RESEARCH IN LAW AND ECONOMICS: A JOURNAL OF POLICY
RESEARCH IN LAW AND ECONOMICS: A JOURNAL OF POLICY Series Editors: Richard O. Zerbe Jr. and John B. Kirkwood Volume 20:
An Introduction to the Law and Economics of Environmental Policy: Issues in Environmental Design – Edited by R. O. Zerbe & T. Swanson
Volume 21:
Antitrust Law and Policy – Edited by J. B. Kirkwood
RESEARCH IN LAW AND ECONOMICS: A JOURNAL OF POLICY VOLUME 22
RESEARCH IN LAW AND ECONOMICS: A JOURNAL OF POLICY EDITED BY
RICHARD O. ZERBE JR. University of Washington, Seattle, WA, USA
JOHN B. KIRKWOOD Seattle University School of Law, Seattle, WA, USA
Amsterdam – Boston – Heidelberg – London – New York – Oxford Paris – San Diego – San Francisco – Singapore – Sydney – Tokyo JAI Press is an imprint of Elsevier
JAI Press is an imprint of Elsevier The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands 525 B Street, Suite 1900, San Diego, CA 92101-4495, USA First edition 2007 Copyright r 2007 Elsevier Ltd. All rights reserved No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333; email:
[email protected]. Alternatively you can submit your request online by visiting the Elsevier web site at http://elsevier.com/locate/permissions, and selecting Obtaining permission to use Elsevier material Notice No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN-13: 978-0-7623-1348-8 ISBN-10: 0-7623-1348-X ISSN: 0193-5895 (Series) For information on all JAI Press publications visit our website at books.elsevier.com Printed and bound in the United Kingdom 07 08 09 10 11 10 9 8 7 6 5 4 3 2 1
CONTENTS ACKNOWLEDGMENTS
vii
LIST OF CONTRIBUTORS
ix
LAYS VS. WAGES: CONTRACTING IN THE KLONDIKE GOLD RUSH Douglas W. Allen
1
BUYER POWER: ECONOMIC THEORY AND ANTITRUST POLICY Zhiqi Chen
17
GENERALIZED CRITICAL LOSS FOR MARKET DEFINITION Malcolm B. Coate and Mark D. Williams
41
PRICE-FIXING OVERCHARGES: LEGAL AND ECONOMIC EVIDENCE John M. Connor
59
PREDATORY PRICE CUTTING AND STANDARD OIL: A RE-EXAMINATION OF THE TRIAL RECORD James A. Dalton and Louis Esposito
155
ON THE OPTIMAL NEGLIGENCE STANDARD IN TORT LAW WHEN ONE PARTY IS A LONG-RUN AND THE OTHER A SHORT-RUN PLAYER Henrik Lando
207
v
vi
CONTENTS
SHOULD VICTIMS OF EXPOSURE TO A TOXIC SUBSTANCE HAVE AN INDEPENDENT CLAIM FOR MEDICAL MONITORING? Thomas J. Miceli and Kathleen Segerson
217
MARKET CONCENTRATION, MULTI-MARKET PARTICIPATION AND ANTITRUST Dennis L. Weisman
233
ACKNOWLEDGMENTS
We would like to thank the University of Washington Daniel J. Evans School of Public Affairs and the Haglund Kelley Horngren Jones & Wilder LLP law firm for financial support.
vii
This page intentionally left blank
LIST OF CONTRIBUTORS Douglas W. Allen
Department of Economics, Simon Fraser University, Burnaby, BC, Canada
Zhiqi Chen
Department of Economics, Carleton University, Ottawa, Ontario, Canada
Malcolm B. Coate
Bureau of Economics, Federal Trade Commission, Washington, DC, USA
John M. Connor
Purdue University, West Lafayette, IN, USA
James A. Dalton
Economics Department, Boston College, Chestnut Hill, MA, USA
Louis Esposito
Economics Department, University of Massachusetts Dartmouth, North Dartmouth, MA, USA
Henrik Lando
Copenhagen Business School and Lefic, Denmark
Thomas J. Miceli
Department of Economics, University of Connecticut, Storrs, CT, USA
Kathleen Segerson
Department of Economics, University of Connecticut, Storrs, CT, USA
Dennis L. Weisman
Department of Economics, Kansas State University, Manhattan, KS, USA
Mark D. Williams
Bureau of Economics, Federal Trade Commission, Washington, DC, USA
ix
This page intentionally left blank
LAYS VS. WAGES: CONTRACTING IN THE KLONDIKE GOLD RUSH Douglas W. Allen ABSTRACT Mine owners during the Klondike gold rush of 1898–1899 used two types of contracts to coordinate workers and their capital: wage contracts and lay (or share) contracts. The key interesting feature of this gold rush was the severe climate and the constraints it placed on the miners. I show that an ‘‘off the shelf’’ incentive model can explain the pattern of contracts, once one understands how the extreme weather environment influenced the behavior of miners.
INTRODUCTION This paper examines an unusual historical episode of contract choice: the Klondike gold rush where wage and lay (share) contracts were used in the mines.1 The theory of contract choice has been examined in many contexts, including gold production; however, the geological features of the Klondike make it an interesting case to consider. Although the Klondike rush was short, the environment was so intense that it placed severe binding constraints on miner’s abilities to mine and steal gold during the mining process. These strong environmental constraints had
Research in Law and Economics, Volume 22, 1–15 Copyright r 2007 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0193-5895/doi:10.1016/S0193-5895(06)22001-3
1
2
DOUGLAS W. ALLEN
strong implications for the transaction costs involved and must have dominated most other considerations of the early miners when it came to the form of employment used. Weather, along with other naturally occurring events, has historically played a significant role in contracting. Weather often limits the duration of production, leaving inputs idle for periods of time. It can drastically increase or decrease output levels by providing the right or wrong conditions for production. And weather can confuse owners of production over how hard workers are trying and whether or not they are stealing. However, the Klondike gold rush stands out as a natural experiment where the weather was extreme. The paper’s general hypothesis is that the extreme mining conditions constrained the ability of miners to steal gold, and constrained the way miners dug for it. Given the weather and climate conditions supervision of labor was often easy and gold theft difficult, thus making wage contracts viable. On the other hand, getting workers to dig at frigid temperatures was often assisted by granting share contracts. However, given the incredible intensity of gold on some creeks, and the technology of mining during the rush, share contracts were not always viable because they ‘‘wasted’’ too much gold and often over-compensated miners. I elaborate on this later, but the share contract reduced the incentive to fully exploit a mine, and given the mining technology of the time it prevented the owner from ever extracting some of the gold. The paper provides one more data point for the case that understanding the incentives and ability of individuals to engage in transaction cost behavior is the best route to understanding contract choice.
KLONDIKE BACKGROUND The Klondike gold rush of 1898–1899 was the last, and perhaps greatest, gold rush in history. It was rapid, intense, and large in terms of people.2 Indeed, in 1899 Dawson City, the central town of the rush, was the largest Canadian city west of Winnipeg. The Klondike River is a tributary to the Yukon River, and is located close to the Alaska border, but within the Yukon territory of Canada. Gold was not found in the Klondike, but along the smaller creeks that run into the Klondike and Indian rivers (Fig. 1).3 The mining area was relatively small, but held some of the highest concentrations of gold ever found.4 Most of the gold was limited to a half-dozen creeks – Bonanza, Eldorado, Hunker, and Dominion being the most famous.
Lays vs. Wages
Fig. 1.
3
Gold Bearing Creeks (From: Chicago Daily Record, Klondike, Book for Gold Seekers, Chicago, IL: Monarch Book Company, 1897).
In total about 12 million ounces have been mined in the area since discovery. Around 6 million ounces came from Eldorado and Bonanza creeks. Gold was discovered when George Carmack staked his discovery claim on Rabbit (later Bonanza) creek on August 17, 1896, and this started what would eventually be three waves of rushes. The first was within a month of the strike when 200 claims were made by local miners in the area. The second rush occurred in the early part of 1897 when about 3,000 people arrived from various parts of Alaska and British Columbia.5 These first two rushes claimed almost every paying creek. Ironically, when the massive rush of 30,000 people arrived in 1898, there was little opportunity for any fortune to be made. Disappointed, many left immediately. Others hung on, either establishing claims on untested creeks, purchasing existing claims, or (in the majority of cases) working for claim owners. When miners worked for others, they did so under wage or lay contracts. When gold was discovered in Nome, Alaska on July 27, 1899, 8,000 people left Dawson City in one week in August, and the Klondike rush was over.6 The Klondike gold rush took place in a modern age of steamboats, railways, and telegraphs. This led to the almost instantaneous arrival of tens of
4
DOUGLAS W. ALLEN
thousands of individuals to the Alaskan shore towns of Dyea and Skagway. However, the remoteness of the Klondike region meant that the last few hundred miles had to be made by foot over the Chilkoot and White passes.7 Thus, for the first three years the Klondike ‘‘hand-tool’’ mining technology resembled that of the California gold rush a half-century before. On the other hand, within two years rail lines were laid over the White Pass, heavy equipment brought in, and modern mines established. The lay contracts disappeared and the large mining companies hired workers with wages. Hence, the Klondike rush provides a short window in which two conditions held: (i) the instantaneous arrival of workers in 1898 meant there was no information on individual histories of workers, and no time had elapsed to develop reputations; and (ii) given the limited technology and supplies, the weather and geological factors of the area had a strong impact on gold production and contracting decisions. The roles of common culture, established social norms, or other factors that take time to form and impact contracting were dominated by the weather and other natural conditions.8
The Weather Constraint Humans have tended to live in temperate and moderate climates, and as a result extreme weather conditions usually are irrelevant or minor considerations in the organization of production. In the Klondike, however, the weather conditions were severe. Weather could not help but influence the organization of the mines because it dictated the type of production, the season of work, and the ability to steal gold at various points of the process. Extensive commentary about the weather is found in the diaries of the miners who made the trek into the Klondike, and rightly so. Temperatures in the winter could reach as low as 701F, and on average there were only 90 frost free days per year.9 At 641 N latitude and 165 miles south of the Arctic Circle, the base city of Dawson was a hostile environment virtually every month of the year. Dark long frigid winters, permanently frozen ground, and short hot, but wet, summers were shocking to all who ventured there. The extreme cold and length of winter was the defining weather characteristic of the Klondike. Unlike southern mines the Klondike ground was permafrost – frozen all year long. In addition, all creeks and rivers were frozen for most of the year. The frozen ground created a unique form of mining. From 1896–1900 the use of hand tools forced the miners to thaw the ground during the winter with small fires over the spot they wanted to mine. When the fire died out they
Lays vs. Wages
5
would scrape a few inches of melted soil out, hoist it out of the mine in a bucket, and place the muck on a ‘‘dump.’’ In the late spring and summer when the creeks would thaw, the dump would be sluiced and panned for gold. Burning continued down the shaft until bedrock was hit 20–30 feet below the surface. If a vein of gold was discovered, miners would ‘‘drift’’ by following it using the same burning method.10 The weather of the area dictated this method of mining for several reasons. First, the frozen ground was stable and allowed the miners to dig without bracing. Second, since all water was frozen during the winter, the mines were dry. Once the spring run off started, mines would often fill with water and require pumping. Third, and most important, given the relative air pressure in the bottom and surface of the mines during the winter, the deadly gases produced by the fires would exit the mines. During the summer these gases would stay below and prevent entry into the mine.11 Throughout most of the winter, the air temperature was so cold that the melted muck loaded in the base of the mine would be frozen solid by the time it reached the top. If not frozen then, it would quickly freeze on the dump. Just as important, any gold in the dump was invisible when frozen.12
LAYS VS. WAGES: THE BASIC TRADE-OFF It was generally considered impractical to mine Klondike gold single handedly, and mine owners hired large numbers of employees.13 Given the small geographical area (only 800 square miles) one might expect a single type of employment contract to arise; however, there were two basic forms of miner coordination in the Klondike: workers were either hired on wages or ‘‘lays’’ (shares). When the rush of miners poured into the area in 1898 and found the entire area staked, most that remained mined under contract with these early miners.14 The form of the wage contracts was straightforward, with miners usually paid by the day or sometimes by the week. Workers were monitored by the mine owner either directly by mining with them, or indirectly through the size of the dump. The lay contracts require slightly more elaboration. Under the Canadian mining laws a creek claim was 500 feet long, running the direction of the creek. The claim would have a width up to the rim of either side of the creek, and varied between 250 and 2,000 feet on the proved creeks.15 A lay contract was seldom given for the entire claim, but rather given for a section. The standard lay section was 50 feet, and meant that a standard claim could have 10 lay contracts at one time. The standard shares ranged from 1/3, 1/2, to 3/4 depending on how many other inputs were
6
DOUGLAS W. ALLEN
supplied by the miner.16 A given lay section of 50 feet would be given to 2–3 miners, and if these miners hit pay dirt they would hire other men by the wage to help clean the dump in the summer. A standard multi-task moral hazard model is sufficient to explain the contract variation in light of the climate constraint.17 Gold is produced using labor and acres of gold bearing land. Nature deposits gold randomly along creek beds, and thus prevents the mine owner from directly inferring labor effort from output. The mine owner uses either wage or share contracts to coordinate the labor and gold land, and chooses the contract that maximizes the value of the mine net of transaction costs.18 For the mine owner, two transaction cost problems arise: worker effort is not completely observable, and gold can be embezzled from the mine. The mine owner must, therefore, monitor miner effort, and this monitoring depends on the type of contract chosen. Generally, a wage worker has a strong incentive to shirk and steal gold, and therefore, has to be extensively monitored. There were a number of potential dimensions over which miner effort could be monitored. For example, the owner could supervise the amount of gold coming out of the mine, or could observe the size of the dump. However, monitoring by output of gold was imperfect given the random nature of the deposits, and even the size of dump would vary by several conditions of nature (most notably the various ground attributes deep in the mine). In practice, the worker effort was proxied by time. Mine owners would often work beside the miners, but would still measure effort with difficulty. Of course, if the mine owner left for supplies, or left for wood for the fires, no direct monitoring took place. An important consideration of wage contracts might normally be the embezzlement of gold during some part of the mining process. A wage worker has a strong incentive to pocket a nugget found in the mine. Some nuggets in the Klondike were worth between $400 and $1,000, and the temptation to hide such a find might have been great. Indeed, the theft of gold nuggets was an ever-present problem in gold mines of the 19th century, given the simple technology and the value of gold relative to its mass. In particular, nuggets picked up during diggings could easily be placed in pockets.19 Unlike other southern mines, however, the severe Klondike cold prevented miners from stealing gold as it was being dug. Consider the following episode: On November 20th, Thomas Flack, William Sloan and a man by the name of Wilkinson sunk a hole eighteen feet deep in Eldorado Creek, and struck a four-foot pay streak that went $5 to the pan, or $2.50 to the shovelful. This was not for a short time, but for weeks and weeks. They shoveled out ton after ton of dirt literally filled with gold and did not know it. (Sola, 1897, p. 74, emphasis added)
Lays vs. Wages
7
There were several reasons why they did not know what they were digging. First it was dark because mining was done in winter. Second, the thawed out muck was frozen by the time it reached the top of the mine, and in this frozen state the gold was not visible. Third, nuggets were not common, and were almost never found during diggings.20 Most nuggets were recovered during clean up when everyone was present and watching. Finally, most of the gold of the region was coarse gold or dust, which simply could not be readily seen or taken by an opportunistic miner.21 Once the summer arrived, the mines filled with water and gas, preventing miners from re-entering and taking gold while others were not looking. Given the short summer months, miners worked full time on their dumps to separate the gold from the pay dirt. Supervision in this close environment was relatively straightforward and large crews of wage workers were feasible.22 Thus, the severe climate of the region virtually eliminated the theft problem with wage workers. The advantage of a lay contract is that it provides some incentive to work at 701F without supervision, through the use of residuals.23 This contrasts with a miner paid by the day who has less incentive to work unless monitored. Since it was not uncommon for miners to own (through purchase) several claims on different creeks, or for the owner of the mine to leave for supplies or collect wood for burning, the lay contract was an obvious method of monitoring labor effort during the dig. Lay contracts, however, have two major drawbacks. First, as with all types of sharing, the lay contract reduces the incentives of the miner to fully exploit a mine. When digging, miners would continue to dig if they felt the expected gains exceeded the costs. Since they only receive a share of the expected benefits, they stop digging ‘‘too soon.’’24 A complete owner of the mine would have an incentive to mine more exhaustively. On the Klondike gold fields this problem became known as ‘‘gophering.’’25 Here the lay miner would drop a hole, and before fully exploring the bedrock below by drifting the mine, would give up and sink another hole close by. Speaking of lays and gophering, Treadgold states: y it is wasteful, snatching a certain amount of gold in a short time without expenditure of capital, but not making the most of any ground y I have noticed several dumps which had been deposited on undermined ground, y before they could be washed, the ground beneath them had caved in y (Treadgold, 1899, p 57).26
The second problem with a lay contract was to choose a payment that did not pay the miner more than his opportunity cost. One method to accomplish this is to adjust the share downwards. Although the record shows three
8
DOUGLAS W. ALLEN
Table 1.
Benefits Costs
The Costs and Benefits of the Wage and Lay Contracts. Wage Contract
Lay Contract
Simple, easy to adjust Requires monitoring Higher incentive to steal gold
Incentive to work in cold temperatures Gophering/waste gold Potential to over-compensate
different shares were used, these varied based on other inputs supplied, and not by creek. As with shares observed elsewhere, there appears to have been little movement in share values. Generally speaking, low shares only exacerbate the ‘‘gophering’’ problem and reduce the incentive to work unsupervised, which may explain why this method of adjustment was not used. Since shares were relatively inflexible, payment adjustment was mostly done through the adjustment of the size of the lay section itself.27 However, there are physical limits to how small a mining section could become before it was physically impossible to mine. First, an area had to be set aside for the dump. Second, and more importantly, there needed to be enough drop in elevation to produce enough water force to sluice the gold during clean up. I can find no evidence that lay claims were ever smaller than 50 feet.28 Thus, if the share and claim size could not adjust, on wealthy creeks the share contract would over-compensate the miners. Table 1 shows the costs and benefits of the two contracts. Thus, a standard incentive based model would predict that wage contracts would be used when efforts to monitor worker effort was low and when there are insignificant opportunities to embezzle gold. Share contracts would tend to exist when the opposite conditions hold, and when the expected deposits did not over-compensate the miners.
THREE TESTS OF THE HYPOTHESIS Eldorado vs. Bonanza Contracting Unfortunately, there are no detailed existing records of wage and lay contracts by mine for the Klondike. However, there are two facts that come out of the historical record: virtually every contemporary writer on the Klondike acknowledged that the two types of contracts were present in the area, and lays were not used on the high quality creeks, they were used on claims with lower densities of gold.29 This observation is most apparent on the two most famous creeks: Bonanza and Eldorado. Although the initial gold find was on
9
Lays vs. Wages
Bonanza, and many miners passed by the small tributary early on, the Eldorado would prove to be one of the richest gold finds in history30. Consistent with the general observation: Bonanza Creek had considerable amounts of lay contracting, but Eldorado Creek had only wage contracts31. The simple trade-off between monitoring costs and effort explains this variation in contracts, given the Klondike circumstances. On extremely rich claims, where the optimal claim size for a lay contract may have been, say 20 feet, the physical characteristics of mining technology of the time prevented the mines from reaching this optimal size. On rich claims, the lay contract simply wasted too much ground and compensated miners too well.32 Thus, more wage contracts were used on the richer creeks like Eldorado. As mentioned above, wage contracts were possible, because the gold was hard to steal. It was hard to steal during the mining process because everything was frozen, during the clean up everyone was watching, and the gold seldom came in nuggets.
The Arrival of Modern Equipment The Klondike gold rush methods, as mentioned, were pre-modern production processes that took place in the modern era. However, neither the cold temperatures nor the steep Chilkoot Pass could keep the region insulated forever. In 1897 the trip from Skagway to Dawson took 3–6 months. By 1901 it would take 2–3 days.33 Aside from speed, rail cars and ships allowed heavy modern equipment to replace the miners shovel. Steam engines were brought in to heat the ground for tunneling and drifting. Heavy pipes and pumps were used to move water from the Klondike River to high dry sites. Dredging equipment worked river and creek beds that had already been mined by hand. These large-scale methods of mining were able to extract more gold from the mines, but required large scales of production to pay. With the introduction of heavy equipment also came the institutional apparatus for advanced real estate and credit markets. The result was that by 1906 the lion’s share of claims was controlled by just two large firms.34 The essential characteristic of the modern heavy equipment was that it eliminated the critical role of weather in the production of gold. Frozen ground was no match for steam and dredging machines. These machines limited a miner’s access to gold and drastically reduced the miner’s ability to conceal gold. They also meant that the unpleasant task of digging gold by shovel in a smoky hole at frigid temperatures was eliminated. Together, these forces encouraged large-scale production, with relatively large numbers of
10
DOUGLAS W. ALLEN
wage-based laborers.35 Most notably, the modern equipment was not constrained by dump location, elevation issues, or temperature. Problems of ‘‘gophering’’ and monitoring workers trying to stay warm disappeared along with lay contracts.36 California and Other Contracts It is quite natural to compare the California gold rush with the Klondike. Though separated by 50 years, as argued, between 1898 and 1900 the Klondike technology was virtually identical. Using Umbeck’s (1981) study for comparison we find an immediate contrast. Umbeck finds that in California miners either used share contracts or land allotments (i.e., individual miners worked their own claim), but not wage employment. Umbeck (1981) argues that share contracts were used to mitigate the risk of mining subject to the cost of contract enforcement. For him, the offsetting feature of sharing was the incentive it provided for stealing gold relative to an owner-operated mine.37 As mentioned, the inability to steal gold in the Klondike during the mining process allowed for wage contracts, which are even more low powered. A final interesting question is: why were there no rental contracts in the Klondike? A rental contract would have fully provided high powered incentives and removed any incentive to steal. In one crude sense there were rental contracts: they were mine sales. Given the short time it took to extract the gold, a sales contract was almost the same as a rental contract. On the other hand, once a mine was rented or sold, the owner had to hire miners to work the mine because the Klondike mines were large.38 However, this brings us back to the original problem of lays vs. wages.
CONCLUSION The general hypothesis of the new institutional economics is that the optimal organization of production depends on the transaction costs of the various methods involved in handling these inputs, and that organizational forms are chosen to maximize the gains from production net of the transaction costs. In order for transaction costs to exist it is necessary for information to be costly and for nature to play a role. When natural inputs to production are costly to observe, moral hazard, adverse selection, and other transaction cost problems become possible. Organizational forms, including contracts, are designed to mitigate these problems.
11
Lays vs. Wages
The Klondike gold rush provides yet one more case study of contracting where contract choice can best be explained by the attempt to mitigate transaction costs in light of the extreme weather that existed in the area. The natural conditions of weather and geography acted as constraints to the general transaction cost problem of optimal organization. The historical record, unfortunately, contains few numbers on many of the details. However, the written records do contain enough information to understand the general organization of the mines. The weather of the area forced miners to mine in ways that hindered theft and generally promoted contracting. Although share contracts encourage worker effort, the high concentration of gold in some areas prevented the use of share contracts. The advancement of modern equipment virtually eliminated the role of extreme weather and led to uniform wage contracts in a modern manufacturing context.
NOTES 1. The term ‘‘lay’’ contract was coined by the miners of the area, but in form was identical to share contracts found in the California gold rush. 2. In terms of the miners per square mile, the Klondike rush had almost three times the density of California. 3. The richest creek, Eldorado, was only 3–6 feet wide at its mouth (McConnell, 1905, p. 40).The extreme concentration, but limited area, of gold in the Klondike resulted from a particular geological circumstance. The highest peak in the region (called ‘‘The Dome,’’ see Fig. 1) was a great massive rock that alone contained gold. As this central location wore down over time, gold moved down its sides along the creeks. See Tyrrell (1907) for a technical report and explanation. 4. Depending how one measures, the Klondike gold region is between 750 and 1,500 square miles. Approximately 200–300 miles of creeks were mined (Innis, 1936, p. 198). The geological survey of Canada estimated the area to be 800 square miles (McConnell, 1905, p. 6). Tyrrell (1907, p. 344) also claims the area to be 800 square miles. 5. Berton (1972, p. 92). 6. Berton (1972, p. 393). 7. The Chilkoot trail, for example, rose from sea level to 4,000 feet in just 18 miles, with a rise of 3,000 feet in the 6 miles before the summit. Miners would cross this pass dozens of times to carry their supplies in. There were other routes to the gold fields, but the White and Chilkoot passes accounted for most of the miners. The ‘‘wealthy man’s’’ route was to take a boat all the way along the Yukon River. 8. Which is not to say the miners arrived in an institutional vacuum, or came without knowledge of social norms or mining laws. Much of the miner’s day-to-day behavior suggests that they came with strong institutional capital with respect to social order, similar to what Zerbe and Anderson (2001) describe. I only mean that
12
DOUGLAS W. ALLEN
for the first two years customs and traditions around how contracts could be made in the Klondike had not developed and needed to be created. 9. Berton (1972, p. 397). 10. Berton (1972, pp. 18–19). 11. The intense winter air created a strong updraft in the mines. Innis (1936, p. 203) notes ‘‘Moderation of the temperature made it more difficult to get adequate draught for fires, and in many cases gases and smoke accumulating at the bottom after a night’s burning made it dangerous for men to work.’’ Diaries of miners who pushed the limits of the waning winter months mention burning eyes as the sign the gases are not leaving. Kirk (1899, p. 146) notes that ‘‘several deaths have already occurred in the Klondike mines from suffocation by gas and smoke.’’ 12. Kirk (1899, p. 148). 13. Haskell (1898, p. 290). 14. Dunham (1898, p. 318). 15. Treadgold (1899, p. 31). 16. Adney (1900, p. 248), Kirk (1899, p. 150). 17. I use the basic analytical framework found in Allen and Lueck (2002). Since the basic analytics of the argument are worked out elsewhere, no formal model is developed here. 18. As is commonly argued in the contract choice literature, the form of contract is used by the principal (mine owner) to mitigate the transaction cost problem. Although a formal legal apparatus was present in the area, the superintendent of mines and the magistrate dealt mostly with claim title issues and criminal matters. They would not generally have been involved in contract issues like worker effort unless there was a matter of theft involved. Unlike California, the Dominion of Canada had a strong presence in the Yukon prior to the rush. The result being that the legal system during the time resembled what it is now. For a more detailed discussion of the rule of law during the gold rush, see Allen (2006). 19. This plays a significant role in Umbeck’s (1981) theory of contract choice in the California fields, and likely explains the lack of wage contracts there. 20. Kirk (1899, pp. 147–148). 21. McConnell (1905, p. 50). 22. Even when claims were mined with lay contracts, the partners would hire wage labor to help with the clean up, suggesting supervision of this task was easy. 23. The lay contract also reduces the incentive to steal gold. However, as just mentioned, this factor was unimportant in the Klondike due to the climate. 24. This is the classic ‘‘Marshallian’’ impact of sharing. 25. Treadgold (1899, p. 57). 26. Given the hand technology of the time, it did not pay to redig the sites. Once dredging machines were introduced these mines were salvaged. 27. It is beyond the scope of this paper to dwell on the rigidity of shares. Allen and Lueck (2005) provide an efficiency argument for this well-known observation in the context of agriculture. Young and Burke (2001) provide an explanation based on ‘‘custom.’’ Interestingly, one of the few studies that reports flexible shares is Hallagan’s (1978) gold study. Regardless, fixed shares existed in the Klondike as strongly as they did on the corn fields of Iowa, and I take this as given.
Lays vs. Wages
13
28. On the other hand, there is evidence that miners did not work on smaller claims. The first miners crudely laid out the initial claims. Later official surveys were done. Readjustments of claims through government surveys often left slivers of land under 50 feet when the original claim was too large. Miners would stake these claims, but they would then sell them back to the adjoining larger claim sites rather than mine them. Ogilvie notes that though these claims were between 15 and 30 feet wide, they were too small to be mined (Ogilvie, 1913, pp. 177–185). 29. See, for example, Kirk (1899, p. 150), Dunham (1898, p. 318), London (1900, p. 73), or Morse (2003, p. 132, quoting a miner named Hiscock’s 1899 diary). 30. Stanley (1898, p. 98) claims that a single shovel on an Eldorado claim resulted in $800 worth of gold. Dunham, along with many others, supports this observation: No. 30 is the great claim of Eldorado. It has produced as high as $20,000 to the box length. About $150,000 was taken out of two cuts – twelve box lengths in all, or about $70 to the square foot. The pay streak is forty feet wide. At one time during the spring the owner could go into the workings and take out a pan of dirt from bed rock and get from $800 to $1,000 to the pan. (Dunham, 1898, p. 324)
31. As mentioned, there is nothing in the historical record that goes into the detail of contract types from claim to claim. Dunham, provides about as much detail as any of the others: The claims on Bonanza were worked almost entirely by the claim owners themselves and by miners who took lays on the claims, and it is probable that not over 100 men were working for wages on the creek at any one time. The claims on Eldorado were worked principally by the claim owners personally and by miners employed by the day, the owners being averse, on account of the extraordinary richness of the ground, to letting it out on lays. The best obtainable data place the number of men working for wages on Eldorado at any one time at 300. (Dunham, 1898, p. 327)
Although the gold on the various creeks varied in terms of purity and coarseness, it did not vary in any substantial way that can explain the difference in contract choice. 32. Over payment could have been dealt with through side payments. That is, workers could have paid mine owners for the right to work on a lay basis. As in so many other cases of sharing, these side payments simply were never observed in the Klondike. 33. Amazingly the White Pass and Yukon Railroad was begun in 1898 and completed in 1900. The narrow gauge railroad climbed 3,000 feet in 20 miles and connected Skagway, Alaska on the coast to Whitehorse, Yukon. From Whitehorse, travelers would move down river to Dawson. 34. See Zaslow (1971, pp. 116–123) for a discussion of the changes that took place between 1900 and 1925. 35. Allen and Lueck (2002) discuss how the same process is often at work in agriculture. 36. The Klondike sheds some light on the role of nature in development. Diamond (1997) argued that the great forces of geography and nature determined
14
DOUGLAS W. ALLEN
development. In the Klondike we see this was true for only a short period of time. Modern methods made nature relatively irrelevant in a short period of time. 37. Hallagan (1978) examines leasing contracts in post-rush California from 1870– 1900 after the government ‘‘arrived’’ in the area. He finds four types of contracts that vary significantly with the type of gold being mined. In the Klondike the nature of the gold was constant throughout the region, and yet the contract types still varied. 38. In Allen (2006) I argue that the security of tenure in the Klondike meant that miners did not worry about other miners encroaching on their claims. The result was larger mines held by operators who then leased them back as lays or hired workers by the wage.
ACKNOWLEDGMENTS My thanks to Yoram Barzel, Karen Clay, William Hallagan, David Jacks, Dean Lueck, Clyde Reed, Gavin Wright, Richard Zerbe, and two anonymous referees for their warm comments.
REFERENCES Adney, T. (1900). The Klondike stampede. New York: Harper & Brothers. Allen, D. W. (2006). Information sharing in the Klondike gold rush. SFU working paper. Allen, D. W., & Lueck, D. (2002). The nature of the farm: Contracts, risk and organization in agriculture. Cambridge: The MIT Press. Allen, D. W., & Lueck, D. (2005). Unlikely customs: Simple cropshare fractions versus continuous cash rents in agriculture. SFU working paper A.N.C. Treadgold (1899). A.N.C. Treadgold. Report on the Goldfields of the Klondike. G.N. Morang, Toronto. Berton, P. (1972). Klondike: The last great gold rush, 1896–1899. Toronto: McClelland and Stewart. Diamond, J. (1997). Guns, germs, and steel: The fates of human societies. New York: W. W. Norton & Company. Dunham, S. C. (1898). The Alaskan gold fields and the opportunities they offer for capital and labor. Bulletin of the Department of Labor, Washington, No. 16, May. Hallagan, W. S. (1978). Share contracting for California gold. Explorations in Economic History, 15, 196–210. Haskell, W. (1898). Two years in the Klondike and Alaskan gold-fields. Hartford: Hartford Publishing Co. Innis, H. (1936). Settlement and the mining frontier. Toronto: Macmillan Co. Kirk, R. C. (1899). Twelve months in Klondike. London: William Heinemann. London, J. (1900). The economics of the Klondike. The American Monthly Review of Reviews, XXIV, 70–74. McConnell, R. G. (1905). Report on the Klondike gold fields. Ottawa: Geological Survey of Canada.
Lays vs. Wages
15
Morse, K. (2003). The nature of gold: An environmental history of the Klondike gold Rush. Seattle: UW press. Ogilvie, W. (1913). Early days on the Yukon. Ottawa: Thorburn & Abbott. Sola, A. E. I. (1897). Klondyke: Truth and facts of the New Eldorado. London: The Mining and Geographical Institute. Stanley, W. (1898). A mile of gold: Strange adventures on the Yukon. Chicago: Laird and Lee. Tyrrell, J. B. (1907). Concentration of gold in the Klondike. Economic Geology, 2(4), 343–349. Umbeck, J. (1981). A theory of property rights: With application to the California gold rush. Ames: Iowa State University Press. Young, H. P., & Burke, M. A. (2001). Competition and custom in economic contracts: A case study of Illinois agriculture. American Economic Review, 91(3), 559–573. Zaslow, M. (1971). The opening of the Canadian North: 1870–1914. Toronto: McClelland and Steward Ltd. Zerbe, R. O., Jr., & Anderson, C. L. (2001). Culture and fairness in the development of institutions in the California gold rush. Journal of Economic History, 61(1), 114–143.
This page intentionally left blank
BUYER POWER: ECONOMIC THEORY AND ANTITRUST POLICY Zhiqi Chen ABSTRACT The objective of this paper is to survey the recent developments in economic theories of buyer power and using the theories as a guide to discuss how antitrust cases involving buyer power can be analysed. An important conclusion that emerges from this survey is that the competition effects of buyer power are quite different depending on whether it is monopsony power against powerless suppliers or countervailing buyer power against large suppliers with market power. A proposed framework of antitrust analysis is presented, and issues related to market definitions and determination of buyer power are discussed.
1. INTRODUCTION The increased concentration of retail industry in Europe and the tremendous success of big-box retailers such as Wal-Mart, Home Depot, and Staples in North America and around the world have enhanced the interest in antitrust policy issues regarding buyer power in recent years. Antitrust authorities in Europe and North America appear to be increasingly concerned about the policy implications of rising buyer power. For example, both the OECD and
Research in Law and Economics, Volume 22, 17–40 Copyright r 2007 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0193-5895/doi:10.1016/S0193-5895(06)22002-5
17
18
ZHIQI CHEN
the European Commission conducted studies to examine the impact of buyer power on competition (OECD, 1998; European Commission, 1999). In the United States, the Federal Trade Commission held a public workshop in 2000 to discuss enforcement policy regarding slotting allowances, a major buyer power issue in grocery retailing.1 A more recent indication of enhanced interest in buyer power in the antitrust community is the publication of a symposium on this subject in the Antitrust Law Journal in 2005. The objective of this paper is to survey the recent developments in economic theories of buyer power and using the theories as a guide to discuss how antitrust cases involving buyer power can be analysed. Section 2 is a review of the theoretical literature on buyer power. An important conclusion that emerges from this survey is that the competition effects of buyer power are quite different depending on whether it is monopsony power against powerless suppliers or countervailing buyer power against large suppliers with market power. These differences can be seen from their differential effects on economic efficiency as well as from the different consequences of their interactions with downstream competition and pricing schemes. In addition, theories of buyer group are also reviewed. Section 3 is a discussion of the antitrust analysis of buyer power. Three analytical issues are discussed: market definition, determination of buyer power, and assessment of anticompetitive effects. On market definition, it is pointed out that buyer power cases often involve two levels of markets. While the definition of downstream markets can be carried out in the conventional way, the definition of upstream markets should focus on seller side substitutability rather than buyer side substitutability. Consequently, the relevant upstream markets are not necessarily aligned with the relevant downstream markets. In other words, the set of competitors on the buyer side of the upstream markets may not be the same as the set of competitors on the seller side of the downstream markets. This last point is especially important in the determination of buyer power. In practice, a useful indicator of buyer power is a buyer’s market share, i.e. the buyer’s share of purchases in the suppliers’ total sales in the relevant upstream market. The above discussion suggests that when calculating the buyer’s share of purchase, one should include in the denominator sales to all buyers in the relevant upstream market, not just those buyers who compete with this particular buyer in the relevant downstream market. For the assessment of possible anticompetitive effects in buyer power cases, I propose an analytical framework that is grounded in economic theories, in particular the recent theoretical developments. The framework uses a classification scheme based on the state of competition in both
Buyer Power
19
upstream and downstream markets. Consistent with the findings from the literature survey, this framework emphasizes the distinction between countervailing power and monopsony power.
2. ECONOMIC THEORY OF BUYER POWER 2.1. Definitions There is no generally accepted definition of buyer power. Existing definitions range from the almost tautological:2 ‘‘[B]uyer power’’ refers to the circumstance in which the demand side of a market is sufficiently concentrated that buyers can exercise market power over sellers (Noll, 2005, p. 589).
to the convoluted: [A] retailer is defined to have buyer power if, in relation to at least one supplier, it can credibly threaten to impose a long term opportunity cost (i.e. harm or withheld benefit) which, were the threat carried out, would be significantly disproportionate to any resulting long term opportunity cost to itself (OECD, 1998, p. 6).
A more useful definition, however, can be found in an earlier report by OECD (1981, p. 10): Buying power may be defined as the situation which exists when a firm or a group of firms, either because it has a dominant position as a purchaser of a product or a service or because it has strategic or leverage advantages as a result of its size or other characteristics, is able to obtain from a supplier more favourable terms than those available to other buyers.
A similar definition of buyer power is proposed by Dobson, Waterson, and Chu (1998, p. 5) who state that buyer power is exercised when ‘‘a firm or group of firms obtain from suppliers more favourable terms than those available to other buyers or would otherwise be expected under normal competitive conditions.’’ My preferred approach is to take the conventional definition of market power,3 invert it to refer to the power on the buyer side, and broaden it to include both monopsony power and countervailing power. To be more specific, I would define buyer power as the ability of a buyer to reduce the price profitably below a supplier’s normal selling price, or more generally the ability to obtain trade terms more favourable than a supplier’s normal trade terms. As will become clear from the discussion below, it is useful to distinguish two types of buyer power based on whether the supplier has market
20
ZHIQI CHEN
power. If there is perfect competition among suppliers, the normal selling price of a supplier is the competitive price. In this case, the buyer power is monopsony power. On the other hand, in a situation where the upstream market is dominated by a small number of suppliers with market power, the normal selling price is above the competitive price. In this latter case, buyer power can be called ‘‘countervailing buyer power’’ or simply ‘‘countervailing power.’’4 Allegations of buyer power typically involve a vertical situation where a firm or a group of firms buys goods (or services) from a supplier, uses them as inputs to produce a value-added product or service, and then sells it to consumers or firms in a downstream market. An obvious example of this kind of vertical relationships is a retailer who buys products from manufacturers and resells them to consumers. Another example, often mentioned in textbooks, is a large company located in a small, isolated town hiring local workers to produce a product sold in national or international markets. A more subtle example is automobile insurance companies whose business can be interpreted as ‘‘buying’’ automotive repair services, ‘‘repackaging,’’ and reselling them in the form of insurance policies. Therefore, exercise of buyer power usually affects two levels of markets: an upstream market and a downstream market. For ease of discussion, I will refer to those firms who buy in the upstream market and sell in the downstream market as ‘‘retailers.’’ I will call the sellers in the upstream market ‘‘suppliers,’’ and the buyers in the downstream market ‘‘consumers.’’ Buyer power then occurs in the upstream market, but it may affect the equilibrium in both upstream and downstream markets.
2.2. Economic Effects of Buyer Power In this section, I will review the economic theories of buyer power, with an emphasis on more recent theoretical developments on this subject.5 In what follows, I will discuss monopsony power and countervailing power separately. As will become clear later (see, in particular, Section 2.2.4), the effects of these two types of buyer power are quite different. In addition, I will also review the theories of buyer groups. 2.2.1. Effects of Monopsony Power As a starting point, consider the textbook theory of monopsony. Here, a single buyer faces a large number of competitive suppliers. In Fig. 1, the downward-sloping demand curve measures the marginal value of the good
21
Buyer Power Marginal expenditure
Price (w)
Supply curve
B
A
w* D wm
C Demand curve
zm
Fig. 1.
z*
Quantity (z)
The Textbook Theory of Monopsony.
to the buyer, and the upward sloping supply curve measures the social cost of supplying an additional unit. Thus, socially optimal price and quantity is determined by the intersection between the demand curve and the supply curve (point A). On the other hand, the marginal expenditure (the additional cost of purchasing an extra unit) perceived by the monopsonist is higher than the supply price of the good because the monopsonist takes into consideration that an increase in quantity purchased pushes up the price. Consequently, the quantity chosen by the monopsonist (zm) is smaller than the socially optimal quantity (z*). By restricting the quantity purchased, the monopsonist succeeds in pushing the price down to wm. The loss in economic efficiency, measured by total surplus, is represented by the deadweight loss triangle ABC. The rectangle w*DCwm, on the other hand, represents a wealth transfer from the seller to the buyer. Note that the above analysis is done without reference to the competitive situation that the monopsonist might face as a seller in the downstream market. In other words, the efficiency loss identified above arises even if the monopsonist faces perfect competition in the downstream market. An imperfectly competitive downstream market changes the marginal value that the monopsonist attaches to the good, but it does not change the fact that the marginal expenditure curve perceived by the monopsonist lies above the supply curve. Consequently, the deadweight loss of monopsony exists independent of the state of competition in the downstream market.
22
ZHIQI CHEN
It is important to recognize two assumptions in the textbook theory of monopsony. First, the supply curve is upward sloping. A monopsonist would not be able to affect the market price and thus cause efficiency loss if the supply curve were horizontal. Second, the transactions between the monopsonist and his supplier are based on a single unit price (linear pricing). If the parties are able to use more elaborate non-linear pricing schemes (e.g. two-part tariffs), there will be incentives for them to design a more efficient contract to capture the lost profit opportunity embedded in the deadweight loss. Therefore, if non-linear pricing is used, the efficiency loss of monopsony may be reduced or even eliminated. Provided that these two assumptions are satisfied, the insight from the textbook theory of monopsony is also applicable to situations of a small number of large buyers (oligopsony). In this regard, Clarke, Davies, Dobson, and Waterson (2002, p. 12) have identified three conditions that, in their view, ‘‘appear necessary for the exercise of buyer power: (i) the buyers contribute to a substantial portion of purchases in the market; (ii) there are barriers to entry into the buyer’s market; and (iii) the supply curves are upward sloping. Under these circumstances, it is straightforward to apply the principles of oligopoly theory to model situations of oligopsony where strategic interaction occurs between a few buyers in a market.’’ Strictly speaking, only conditions (i) and (ii) are necessary for the existence of buyer power. Condition (iii) is one of the necessary conditions for the exercise of buyer power to cause deadweight loss in the textbook model of monopsony. Indeed, Shaffer’s (1991) analysis of slotting allowance and resale price maintenance provides an example where buyer power in the hands of a small number of buyers may cause efficiency loss even if supply curves are horizontal. In Shaffer’s model, duopoly retailers face competitive suppliers who have constant marginal costs and thus horizontal supply curves. Consequently, the deadweight loss analysis of the textbook model does not apply. Nevertheless, Shaffer demonstrates that the exercise of buyer power by the retailers can still cause efficiency loss. The loss arises because the retailers use their buyer power to mitigate their competition in the downstream market. To be more specific, in Shaffer’s model suppliers compete to obtain shelf space at the retail level, and the instruments used to obtain shelf space are slotting allowance and resale price maintenance. Competition among suppliers gives retailers the buyer power to dictate the terms of contracts between the retailers and the suppliers. The use of slotting allowance allows the retailers to extract profits from the suppliers without depressing the unit wholesale prices, and higher wholesale prices mitigate downstream competition among the retailers. Alternatively, resale price maintenance provides a
Buyer Power
23
means for retailers to commit contractually to higher prices in the downstream market, thus raising retailer profits by eliminating their incentives for aggressive downstream pricing. Therefore, both types of contractual provisions can raise retailer prices and profits; thus, both can have anticompetitive effects.
2.2.2. Effects of Countervailing Buyer Power The term ‘‘countervailing power’’ was coined by Galbraith (1952) to describe the market power developed on one side of a market as a result of the market power on the other side. In his controversial book American Capitalism: The Concept of Countervailing Power, Galbraith argues that economic power on one side of a market begets countervailing power on the other side. According to Galbraith, an important manifestation of countervailing power was that of large and powerful retail organizations such as the major chain stores at that time (e.g. A&P and Sears, Roebuck). By exercising countervailing power, he argues, these retailers were able to lower the prices they paid their suppliers and passed on these savings to their customers. Galbraith’s thesis was controversial at the time when the book was first published (see Stigler (1954) and Hunter (1958) for critiques of Galbraith’s book). But subsequently it was largely ignored for a long time, as evidenced by the fact that until late 1990s little theoretical analysis had been done on this subject.6 In recent years, the growing dominance of powerful big-box retailers and the resulting increased concentration in retail markets have renewed the interest in the countervailing power hypothesis. Since 1996, a number of papers have been published that analyse the countervailing power hypothesis in formal models. They include von Ungern-Sternberg (1996), Dobson and Waterson (1997), Chen (2003), and Erutku (2005).7 The main insight from the analyses by von Ungern-Steinberg and Dobson and Waterson is that increased concentration at the retail level does not necessarily lead to lower prices for consumers; under certain conditions it may in fact lead to higher prices. von Ungern-Sternberg (1996) studies a model where a monopolist producer bargains with oligopolist retailers. Retailers offer identical services and they compete in quantity (i.e. Cournot competition). He demonstrates that a decrease in the number of retailers unambiguously leads to an increase in equilibrium consumer prices. As a point of comparison, he also studies the case where the retail market is perfectly competitive and finds that a decrease in the number of retailers leads to a reduction in equilibrium consumer prices. He concludes that
24
ZHIQI CHEN
countervailing power can have positive effects for the consumer only if competition at the retail level is very fierce. Dobson and Waterson consider a similar model where a monopolist supplier bargains with oligopolist retailers. Different from von UngernSternberg, services offered by different retailers are imperfect substitutes. Consequently, in their model, a more competitive retail market manifests itself through closer substitutability of retailer services. They show that consumer prices fall and economic welfare rises with the reduction in the number of retailers only if retailer services are regarded as close substitutes. The analyses by von Ungern-Sternberg and Dobson and Waterson highlight two opposing forces of retailer consolidation on consumer prices and economic welfare. When the number of retailers is reduced, the remaining retailers gain both countervailing power against their suppliers and market power against consumers. The countervailing power over the suppliers tends to reduce wholesale prices, which can lead to lower consumer prices when there is intense competition in the retail market. On the other hand, increased market power in the retail market allows the retailers to boost their price–cost margins, which tends to push up retail prices. Which effect dominates depends greatly on the intensity of retail competition – the former effect is stronger when competition is intense, otherwise the latter effect dominates and consolidation leads to higher prices for consumers. The above discussion suggests that if retailer consolidation leads to higher consumer prices, it is caused by the increased market power in the retail market. In these models, retailer countervailing power by itself benefits consumers and improves economic efficiency by exerting downward pressure on wholesale and retail prices. In contrast, the analysis by Chen (2003) suggests that retailer countervailing power does not necessarily improve economic efficiency. Specifically, Chen (2003) studies the effects of countervailing power in the hands of a dominant retailer. In his model, an upstream supplier sells to a group of downstream retailers, which consists of a dominant retailer and a competitive fringe. The dominant retailer possesses countervailing power, but a fringe retailer does not. An increase in the degree of countervailing power manifests itself in the form of a larger share of the joint profits going to the dominant retailer. Using this framework, Chen demonstrates that a rise in the countervailing power possessed by the dominant retailer reduces the equilibrium price paid by consumers. This, however, comes at the expense of possible efficiency loss in the provision of retail services. As a result, economic efficiency does not always improve with the rise of countervailing
Buyer Power
25
power. Furthermore, the presence of fringe competition is crucial for countervailing power to benefit consumers. Erutku (2005) modifies Chen’s (2003) model by considering a situation where a national retailer competes with a local retailer in each geographic market. National retailer possesses countervailing power against a monopoly supplier, which allows it to get a discount off the list price set by the supplier. A local retailer, on the other hand, does not have any countervailing power and thus pays the list price. Services offered by the national retailer and local retailer are imperfect substitutes. Erutku shows that when forced to offer a larger discount to the national retailer, the supplier may raise the list price, which would push up the retail prices of local retailers. This occurs when the national retailer’s countervailing power is relatively small. If its countervailing power is relatively large, on the other hand, both wholesale list price and retailer prices fall. An increase in the degree of competition between retailers makes the latter situation more likely. The price of the national retailer, on the other hand, falls monotonically with its countervailing power. Therefore, Erutku’s analysis identifies a situation where countervailing power of a large retailer may benefit some consumers but hurt other consumers. All of the analyses reviewed so far focus on the effects of countervailing power on prices. However, price effect may not be the only consequence of retailer countervailing power. In particular, concerns have been expressed about the possible longer term effect on product variety and innovation. This has been the subject of analysis in a number of recent working papers, including Chen (2004), Inderst and Shaffer (2004), and Inderst and Wey (2005). Chen (2004) studies a monopoly supplier’s choice of product diversity and how that choice is affected by retailer countervailing power. He shows that the number of products produced by the monopolist is lower than that of the constrained social optimum. Retailer countervailing power lowers consumer prices but reduces product diversity. Consequently, it alleviates the distortions in prices but exacerbates the distortions in product diversity. In Chen’s model, the former is outweighed by the latter. Therefore, Chen’s analysis demonstrates that retailer countervailing power can cause reduction in the number of products available to consumers, and the efficiency loss caused by this reduction may be large enough to lower aggregate economic welfare. A welfare trade-off between lower consumer price and reduced product diversity can also be found in Inderst and Shaffer’s (2004) analysis of retailer mergers. In their model, there are two suppliers selling differentiated
26
ZHIQI CHEN
products to two retail outlets. They show that following a merger that gives a single retailer control over both outlets, the retailer may want to enhance its buyer power vis-a`-vis the suppliers by delisting a product and committing to a ‘‘single-sourcing’’ purchasing strategy.8 Anticipating further concentration in the retail industry, suppliers will strategically choose to produce less differentiated products, which further reduces product diversity. Inderst and Wey (2005) identify a mechanism through which increased retailer countervailing power can improve economic efficiency by strengthening a supplier’s incentives to reduce production costs.9 In their model, a single supplier sells to competing intermediaries (e.g. retailers) who operate in many separate markets. As an alternative to purchasing from this supplier, a retailer can develop its own supply channel at a cost. A large retailer is able to spread this cost over a larger number of units, which strengthens the retailer’s bargaining position against the supplier. On the other hand, a reduction in the supplier’s marginal cost of production, brought about by the investment in innovation, improves the supplier’s bargaining position against a retailer, because a lower marginal cost will lead to lower per-unit purchasing prices for all supplied firms and, consequently, a retailer that chooses to switch to another source of supply will be more at a disadvantage vis-a`-vis its competitors. Therefore, the supplier can counter the strengthening bargaining position of a larger retailer by making more investment in innovation to engineer a larger reduction in its cost. An increase in retailer countervailing power, then, strengthens the supplier’s incentives to reduce costs. 2.2.3. Buyer Groups Buyer power issues can also arise from buyer groups. If a buyer group is formed solely for the purpose of gaining market power over the group’s suppliers, the effects of such buyer power can be analysed in more or less the same way as in the case of a single buyer. The small literature on buyer groups (Mathewson & Winter, 1997; Dana, 2003) is mainly concerned with the incentives to form buyer groups. According to Mathewson and Winter (1997), buyer groups can form in response to a market inefficiency caused by monopolistic competition, namely the failure of monopolistically competitive markets to achieve the optimal trade-off between lower costs and greater variety or availability of products. They start with the proposition that a monopolistically competitive equilibrium can lead to an excessive number of firms or product variety, selling at excessive prices. They demonstrate that this property of monopolistically competitive markets is enough to generate the incentive for buyers
Buyer Power
27
collectively (or for a large subset of buyers) to offer a subset of sellers the right to their exclusive business in exchange for lower prices. The formation of such a buyer group benefits buyers inside the coalition. However, those consumers and firms outside the group may be made worse off. Outside consumers may face higher prices and fewer suppliers, and outside firms may face a smaller customer base. Dana (2003) suggests that buyer groups can be used as a commitment device in the negotiation with suppliers. He argues that a buyer group can create buyer power by committing to buy exclusively from a single supplier who offers the lowest price. When buyers with heterogenous preferences form such a buyer group, they induce the suppliers to compete more aggressively. Thus, a buyer group makes its members better off by intensifying price competition among suppliers. While some members of the group end up consuming the product they value less, the expected benefit of increased price competition exceeds the expected cost of consuming the wrong product. The incentives to form buyer groups and the effects of these groups analysed in these two papers can also be viewed as the incentives for and the effects of a merger among buyers. Indeed, there is some parallel between Dana’s analysis of buyer groups and that of retailer mergers by Inderst and Shaffer (2004). This lends support to the earlier claim that in many instances analysis of a buyer group can be done in more or less the same way as in the case of a single buyer. There are situations, however, where buyer groups may raise unique competition issues. By definition, a buyer group involves coordination in the purchase decisions of the buyers. Typically, coordination takes place in the choice of suppliers and negotiation of prices. However, if coordination also extends to the quantity purchased by each member and this quantity has a direct impact on the member’s output level in a downstream market, the buyer group could be used as a way to enforce collusion in the downstream market. Even in the absence of quantity coordination, one has to consider the potential facilitating effects because a buyer group reduces the variation in input costs among its members. Tacit collusion is easier to achieve when competitors have similar input costs. Therefore, in situations where members of a buyer group collectively account for a significant portion of the downstream market, the buyer group could be used as an instrument to facilitate downstream collusion. On the other hand, there are also situations where the formation of a buyer group may have nothing to do with the creation or exercise of market power. Consider a situation where there are economies of scale in the
28
ZHIQI CHEN
supplier’s distribution technology so that the average cost of processing a large purchase order is lower than that of a small order. In such a situation, a supplier may choose to offer volume discounts to encourage large purchase orders. Smaller buyers then may find it beneficial to pool their purchase orders so that they can receive the volume discounts.10 A buyer group in such a situation is a way to exploit the efficiency gains associated with economies of scale in distribution technology. 2.2.4. Comments It is clear from the above literature review that the effects of buyer power are quite different depending on whether it is monopsony power against competitive suppliers or it is countervailing power against suppliers with market power. These differences can be seen in the following three areas. (1) The effects on economic efficiency. Generally speaking, exercise of monopsony power causes efficiency loss. With monopsony power, the best possible scenario is that the supplier and the retailer are able to use efficient contracts to avoid any deadweight loss, but even here monopsony power will not provide any benefits to final consumers. Countervailing power, on the other hand, is more likely to benefit consumers. Whether countervailing power improves economic efficiency will depend on the specific situations, and the analyses reviewed here are useful for identifying conditions under which it does. (2) The role of downstream competition. Deadweight loss of monopsony exists even if there is intense competition in the downstream market. On the other hand, the welfare effects of countervailing power depend critically on the state of competition in the downstream market. A common theme in this literature (von Ungern-Sternberg, 1996; Dobson & Waterson, 1997; Chen, 2003; Erutku, 2005) is that consumers are more likely to benefit from countervailing power and consequently welfare is more likely to improve when there is intense competition in the downstream market. (3) The role of linear and non-linear pricing. In the textbook theory of monopsony, the use of linear pricing (a single unit price) plays a critical role in the welfare consequence of monopsony. In that model, the use of nonlinear pricing would reduce or even eliminate the deadweight loss of monopsony. By contrast, countervailing power is more likely to benefit consumers when linear prices are used (see, for example, Inderst & Shaffer, 2004; Erutku, 2005). With linear prices, the exercise of countervailing power will necessarily lead to lower wholesale prices, and lower wholesale prices usually translate into lower consumer prices. The use of non-linear prices, on the other hand, makes it possible for a supplier and a retailer to
29
Buyer Power
reallocate their joint profits without affecting the wholesale prices. Consequently, non-linear pricing may allow the firms to insulate the retail prices from the effects of a shift in market power. The above discussion suggests that it is important to distinguish between monopsony power and countervailing buyer power in the antitrust analysis of buyer power.
3. ANTITRUST POLICY AND BUYER POWER Analysis of an antitrust case usually consists of at least three parts: market definition, determination of market power, and assessment of anticompetitive effects. In what follows, I discuss each of these three parts for cases involving buyer power. The focus of this discussion is on areas where buyer power requires a somewhat different treatment from conventional market power.
3.1. Market Definition Buyer power cases often involve two levels of markets, which may require that market definitions be done for both the upstream markets and the downstream markets. The definition of downstream markets can be carried out in the conventional way (i.e. using the hypothetical monopolist test) since a retailer is a seller in the downstream markets. The definition of upstream markets, in which the alleged buyer power resides, requires more discussion because it is done from the buyer side as opposed to the seller side. In principle, the approach to market definition from the buyer side should be symmetric to the approach to market definition from the seller side. Therefore, I would mirror the hypothetical monopolist test used in merger reviews11 and describe the approach to market definition from the buyer side as follows: A relevant market is defined as the smallest group of products and the smallest geographic area in which a sole profit-maximizing buyer (a ‘‘hypothetical monopsonist’’) would impose and sustain a significant and non-transitory price decrease below its normal level.12 Accordingly, market definition from the buyer side should focus on seller side substitutability, that is the ability by a seller to find alternative buyers. In this regard, an important factor to consider is the supplier’s switching costs. When confronted by a retailer demanding lower than normal prices, a supplier may want to sell its product to an alternative retailer. This option,
30
ZHIQI CHEN
however, may not be a profitable one if it faces significant switching costs, in which case, the alternative retailer has to be excluded from the relevant upstream market. Note that the relevant upstream markets defined using the hypothetical monopsonist test are not necessarily aligned with the relevant downstream markets defined using the hypothetical monopolist test. In other words, the relevant upstream markets and the relevant downstream markets could be quite different in terms of the products being included, or the geographic areas being covered, or both. For example, imagine a buyer power case where a supermarket chain is accused of abusing buyer power against, say, a toothpaste manufacturer. At the downstream level, the relevant product market may be defined as the one-stop shopping of grocery products, and accordingly, the geographic market is likely to be local. The competitors in a downstream market are supermarket chains operating in a particular geographic area. At the upstream level, on the other hand, the relevant product market, defined from the buyer’s side, is likely to be the wholesale purchase of toothpaste. The relevant geographic market will be national if the wholesale purchases are done at the national level. Competitors included on the buyer side of the upstream market may be all the supermarket chains operating in different parts of the country as well as other types of retailers (e.g. pharmacies) that also purchase and resell toothpaste. In such a case, there is significant asymmetry between the relevant upstream markets and relevant downstream markets.
3.2. Existence of Buyer Power Before assessing its impact, one should first determine whether a retailer in fact possesses any buyer power. One should not presume that a retailer has buyer power simply because it is large in size relative to its supplier. A large retailer may not be able to obtain below-normal prices from a supplier if it has to compete aggressively against other retailers for the supplier’s products. The key to the existence of buyer power, therefore, is not the relative size, but whether there is vigorous competition, either actual or potential, for the supplier’s products. To gauge whether there is actual competition for the suppliers’ products, one can start with a calculation of the retailer’s share of purchases in the relevant upstream market. Note from the toothpaste example that the other buyers in this upstream market do not necessarily compete with this retailer in the relevant downstream market. Therefore, one should take care to
Buyer Power
31
include the sales to all buyers in the relevant upstream market, not just those buyers who compete with the retailer in the relevant downstream market. It has been pointed out (Kirkwood, 2005, p. 637) that while a dominant share of purchases in the relevant market may be necessary for the exercise of monopsony power, such dominance is not required for a buyer to exert countervailing power. Kirkwood (2005, pp. 642–644) lists a number of instances where countervailing power had been exerted even though the buyers had neither a very large share of purchases nor a dominant share. Therefore, a very large share of purchases is not a necessary condition for the existence of buyer power. Neither is it a sufficient condition. One must also consider the barriers to entry into the buyer side of the upstream market. If high barriers to entry exist, an incumbent retailer may not have to worry too much about the possibility of its suppliers being bid away by a new retailer offering higher prices. Note, again from the toothpaste example, that the new retailer does not have to be a competitor in the incumbent’s downstream market.
3.3. Competition Effects Dobson et al. (1998) have proposed an approach to analysing the competition effects of buyer power, framed around five questions (see Table 1).13 Questions 1–3 are designed to determine the existence of market power in the hands of suppliers and buyers. Depending on the answers to these questions, the effects of buyer power can be different. For example, ‘‘if the buyer power is against relatively powerless suppliers then there are concerns about abuse of monopsony power, which might include a detrimental effect on producer (suppliers’) surplus and the long-term viability of suppliers. On the other hand, if buyer power is linked with significant seller power at the upstream level then it is more likely that the existence or enhancement of buyer power is beneficial, that is buyer power may have a socially beneficial countervailing effect by negating the detrimental effects of upstream seller power’’ (Dobson et al., 1998, p. 31). Through Question 4 Dobson et al. propose to examine buyers’ market behaviour, with a focus on buyerinduced vertical restraints that are potentially anticompetitive. Finally, Question 5 considers the underlying economic conditions in production/ distribution in order to determine whether there is an efficiency reason for the presence of buyer power. As this approach was developed in 1998 (or earlier), it did not have the benefit of the insights generated by theoretical developments since then.
32
ZHIQI CHEN
Table 1.
Analytical Framework in Dobson et al. (1998).
Question 1. Is there significant buyer power? If not, the considerations of buyer power are not relevant. (By ‘significant power’ is meant the ability to have material effect on prices set or negotiated, on quantities exchanged, or on the viability of traders at one or more stages of the production/distribution cycle.) 2. Is the buyer power against relatively powerless suppliers? If so, it is more likely that buyer power has policy implications. (In contrast, if buyer power is linked with significant seller power at the upstream stage then it is more likely that the existence or enhancement of buyer power is beneficial.) 3. Does the buyer itself have significant selling power? If so, then buyer power may serve as a means of strategically enhancing seller power in the downstream market raising potentially adverse effects 4. Does the buyer attempt to constrain its suppliers’ other actions? If so, such an arrangement should be treated with suspicion 5. Are there significant productive efficiency gains associated with buyer power? If so, then there may be an efficiency justification for the presence of buyer power
Relevant Evidence Significant proportion of the product as a whole purchased by this firm Significant arrangement of terms of purchase by this firm (e.g. upfront fees for distributing a product, such as slotting allowances) Absence of evidence that suppliers dictate terms of sale Low seller concentration in the upstream market
Normal means of assessing seller power (in the downstream market)
Evidence of exclusive supply requirements, specific custom designs or arrangements, idiosyncratic specification, etc. Pecuniary or other economies of scale indicating ‘natural’ monopsony tendency (i.e. average costs lowered by buying being undertaken by a single party)
Here, I present an analytical framework of buyer power that is more firmly grounded in the economic theories of buyer power, in particular recent theoretical developments in this area. Instead of a series of questions, this approach uses a classification scheme based directly on the state of competition in both the upstream market and the downstream market. This framework is summarized in Table 2. As can be seen from the table, the state of competition in the upstream market is classified into three situations: (i) the retailer in question is a competitive buyer, (ii) the retailer has monopsony power, and (iii) the retailer has countervailing buyer power.14 Meanwhile, the condition in the downstream market is divided into two cases: (a) the retailers are competitive sellers and (b) the retailers have
33
Buyer Power
Table 2.
A Framework to Analyse the Competition Effects of Buyer Power. Upstream Market Competitive Buyers Competitive sellers
No buyer power issue when markets are in equilibrium
Monopsony Power
Countervailing Buyer Power
If linear prices are used, efciency loss is likely
Buyer power will likely benet consumers in the short run Buyer power may have potential long-run effects on product variety and innovation in the upstream market, and on the state of competition in the downstream market Efciency loss is possible in both short run and long run There are more potential competition problems than in the case of competitive downstream market
If non-linear prices are used, efciency loss is less likely but the issue of wealth transfer remains
Downstream market
Market power
No buyer power issue when markets are in equilibrium
Efciency loss is possible even if non-linear prices are used
market power over consumers. The discussion below is organized based on the state of competition in the upstream market. 3.3.1. A Competitive Buyer If there is a large number of small buyers in the upstream market, buyer power does not exist when the market is in equilibrium. If there is no buyer power in this case, why include it in this analytical framework? The reason is
34
ZHIQI CHEN
that there may be situations where occasional changes on the demand or supply side shift the balance of power in favour of some retailers in the short run, which may lead to complaints about anticompetitive buyer power. For example, changes in consumer tastes may lead to a permanent decrease in the demand for a product. The excess capacity caused by the drop in demand may trigger a price war among suppliers as they struggle to stay alive. This may confer buyer power on some retailers in the short run. The buyer power, however, evaporates in the long run as some suppliers are forced to exit and the number of suppliers in the market reaches its new equilibrium level. Similarly, advances in technology that enlarge the efficient scale of production may reduce the number of suppliers needed in a market and, as existing suppliers fight for survival, confer buyer power to some retailers in the transition period. However, as long as a reasonably large number of buyers remain in the upstream market, this kind of short-run buyer power is unlikely to raise competition concerns. Note that the above discussion is applicable whether a retailer has market power in a downstream market or not. Market power in a downstream market could raise competition policy issues of its own. But that is not the subject of analysis in this paper.
3.3.2. Monopsony Power Monopsony power arises when a retailer possesses market power against a competitive supplier. As has been discussed in Section 2, monopsony power against competitive suppliers can cause deadweight loss both in the case where the downstream market is competitive and in the case where the retailer has market power in the downstream market. This is particularly the case where a linear pricing scheme is used in the contracts between the retailer and its supplier, in which case, the deadweight loss caused by monopsony power is independent of the state of competition in the downstream market. The effects of monopsony power are more ambiguous when a non-linear pricing scheme is used. In the case of competitive downstream market, the use of non-linear pricing may reduce or eliminate the deadweight loss arising from monopsony power, although monopsony power still causes wealth transfer from the supplier to the retailer. However, the effects of monopsony power may not be as benign in the case where the retailer has market power in the downstream market. In such a situation, as Shaffer (2001) shows, oligopolistic retailers with monopsony power can use non-linear pricing as an instrument to lessen competition in the downstream market.
Buyer Power
35
3.3.3. Countervailing Buyer Power The analysis tends to be more complex when buyer power is exercised against sellers with market power. Depending on the state of competition in the downstream market, a range of scenarios is possible. For example, exercise of countervailing power could benefit consumers in the short run but not long run, benefit consumers in both short run and long run but at the cost of efficiency on the production side, or harm consumers and cause efficiency losses in both short run and long run, just to name a few possibilities.15 Therefore, the subsequent discussion in this subsection is organized based on the state of competition in the downstream market. The Retailer Faces Intense Competition in the Downstream Market. In this case, exercise of countervailing power by a large retailer is likely to benefit consumers, at least in the short run. This is the scenario that has often been put forward in the discussion of buyer power in popular press. Buyer power in the upstream market allows a large retailer to obtain lower prices from its suppliers, but intense competition in the downstream market forces the retailer to pass on at least a portion of the cost savings to consumers. Beneficial effects to consumer do not necessarily mean that exercise of countervailing buyer power is always free of efficiency loss. As Chen (2003) shows, increased retailer countervailing power by a dominant retailer, while benefiting consumers by reducing the prices they pay, can cause efficient loss on the production side. The reason is that exercise of buyer power by one retailer will typically cause redistribution of retailing business in the downstream market. Given that this redistribution of business is based on the retailers’ buyer power (or the lack of) in the upstream market rather than on their productive efficiency in the downstream market, it tends to result in distortions in downstream production. Furthermore, one must consider the effects of countervailing power in the long run. A question of particular interest is, can the intensity of competition in the downstream market be maintained in the long run? Or, to put it in a slightly different way, will the competition in the downstream market be lessened as a result of exercise of countervailing buyer power? In this regard, it may be useful to consider the relative strengths of the competing retailers in the upstream market. If most retailers have approximately the same amount of buyer power against their suppliers, it is unlikely that one of the retailers will be able to use the cost advantage obtained from the exercise of buyer power to squeeze out most of its rivals in the downstream market. In such a situation, it is likely that most of the existing competitors in the
36
ZHIQI CHEN
downstream market will survive, and hence effective competition will likely remain in the long run. If, on the other hand, buyer power is concentrated in the hands of one or two dominant retailers, competition problems may arise in the long run when a significant number of smaller retailers are forced out of the downstream market and, as a result, the dominant retailers acquire significant seller power in the downstream market. Finally, one must consider the long-run effects of countervailing power on product variety and, more generally, on investment in innovation by upstream suppliers. Recall from the discussion in Section 2 that the theoretical predictions on this subject are rather mixed. Depending on the circumstances, retailer countervailing power can cause distortions in product variety that outweighs the benefits of lower consumer prices (Chen, 2004), or strengthen suppliers’ incentive to invest in innovation (Inderst & Wey, 2005).
The Retailer Possesses Market Power in the Downstream Market. In this case, the retailer possesses market power in both upstream and downstream markets, and countervailing buyer power is most likely to cause competition problems. Broadly speaking, competition problems may arise from two directions. First, the exercise of buyer power itself may cause harm to consumers and deadweight loss in the economy. Insufficient competition in the downstream market means that a retailer with buyer power will not be compelled to pass on the cost savings to consumers. On the contrary, it may find it advantageous to raise the purchase prices it pays the suppliers and extract profits from the suppliers in the form of lump-sum payments such as slotting allowances. In an oligopolistic retail market with insufficient price competition, a commitment to higher purchase prices by a retailer will push up its own retail prices, and encourage other retailers to raise prices (Shaffer, 1991).16 Second, a retailer may abuse the dominant position conferred by its buyer power in an attempt to eliminate or stifle competition in the downstream market. Such abuse of dominant position may take the form of pressuring suppliers into not supplying certain competitors of the retailer; imposing market restrictions, such as exclusive dealing, on suppliers; or raising rival’s costs by artificially bidding up the suppliers’ prices. The possible anticompetitive effects of these practices can be analysed in the same way as conventional abuse of dominance cases.
Buyer Power
37
4. CONCLUSIONS An important observation from the literature survey is that the competition effects of buyer power are quite different depending on whether it is monopsony power against competitive suppliers or it is countervailing power against suppliers with market power. Generally speaking, exercise of monopsony power causes efficiency loss. With monopsony power the best possible scenario is that the supplier and the retailer are able to use efficient contracts to avoid any deadweight loss; but even here monopsony power will not provide any benefits to final consumers. Countervailing power, on the other hand, has a better chance to benefit consumers. Whether countervailing power improves economic efficiency will depend on the specific situations, in particular the state of competition in the downstream market. Consistent with this observation, the analytical framework of buyer power proposed in this paper uses a classification scheme that is based on the state of competition in both upstream and downstream markets and emphasizes the distinction between monopsony power and countervailing buyer power.
NOTES 1. Shaffer (1991, p. 12) defines slotting allowance as ‘‘fees paid by manufacturers to obtain retail patronage. They may be cash gifts or payments in kind, such as cases of free goods. Either way, their salient characteristic is that the fee paid does not vary with subsequent retailer sales.’’ 2. This definition conveys virtually no information as it does not specify what market power means in this context. It is equivalent to defining monopoly power (seller power) as the ‘‘exercise of market power by a seller.’’ 3. Market power or monopoly power is usually defined as the ability of a firm to set prices profitably above competitive levels. See, for example, Carlton and Perloff (2005, p. 783) and Viscusi, Harrington, and Vernon (2005, p. 294). 4. This approach is similar to the one in Kirkwood (2005), which makes a distinction between monopsony power and bargaining power based on the way buyer power is exercised. Monopsony power is exerted through the reduction in the quantity purchased, while bargaining power ‘‘is the power to obtain a concession from another party by threatening to impose a cost, or withdraw a benefit, if the party does not grant the concession’’ (Kirkwood, 2005, pp. 638–639). Kirkwood also observes that normally bargaining power can be exercised only in markets where a seller has market power. Therefore, there is little substantive difference between the bargaining power in Kirkwood (2005) and the countervailing power defined here. 5. Surveys of older literature can be found in Blair and Harrison (1993) and Clarke et al. (2002).
38
ZHIQI CHEN
6. To the best of my knowledge, the only theoretical analysis during this period is Veendorp (1987), in which numerical examples are presented to illustrate the point that while countervailing power may lead to lower consumer prices, it does not necessarily improve economic efficiency. 7. In a related paper, Snyder (1996) develops a model where a price-taking buyer accumulates backlogs of unfilled orders for the purpose of weakening tacit collusion among sellers and obtaining lower prices. While the buyer in his model can be interpreted as a downstream firm that buys an intermediate input and converts it into a final product, the model’s assumption of constant buyer valuation of the input amounts to assuming that the downstream firm faces a fixed price in the downstream market. Therefore, by design the firm’s countervailing power in this model has no impact on the consumer price in the downstream market. 8. The source of buyer power identified here is reminiscent of that of a buyer group in Dana (2003). See Section 2.2.3 for discussion on buyer groups. 9. The same kind of arguments can also be found in an earlier paper by Inderst and Wey (2004). 10. For this to occur, it must be the case that the cost of aggregating small orders by the buyer group is lower than the cost of performing the same task by the supplier. 11. See, for example, Section 1.0 of the Horizontal Merger Guidelines in the US, or paragraph 3.4 of the Merger Enforcement Guidelines in Canada. 12. Recall from the discussion in Section 2.1 that the normal selling price is the competitive price if there is perfect competition among suppliers, but it is above the competitive price if the upstream market is dominated by suppliers with market power. 13. The same approach is presented in Clarke et al. (2002). 14. Note that a multi-product retailer may be in more than one of these situations simultaneously. It may deal with suppliers with little market power (e.g. a small vegetable farmer) for some products, and large suppliers with monopoly power (e.g. the Coca Cola Co.) for other products. 15. Related to this, Kirkwood (2005, pp. 647–651) lists five ways in which non-cost justified discrimination as a result of the exercise of countervailing power can harm consumers. The discussion below touches on these five ways in various details. 16. In practice, however, such occurrence may not be very common. Based on his personal experience at FTC, Kirkwood (2005, p. 647) observes that in most of the investigations involving large buyers, the exercise of buyer power led to lower perunit prices from suppliers (in addition to lump-sum payments such as slotting allowances) and some of the savings were passed on to consumers.
ACKNOWLEDGMENTS For comments and discussions I thank Andy Baziliauskas, Tim Brennan, Alan Gunderson, John Kirkwood, Roger Ware, and Thomas Ross.
Buyer Power
39
REFERENCES Blair, R. D., & Harrison, J. L. (1993). Monopsony: Antitrust law and economics. Princeton: Princeton University Press. Carlton, D. W., & Perloff, J. M. (2005). Modern industrial organization (4th ed.). Boston: Pearson Addison-Wesley. Chen, Z. (2003). Dominant retailers and countervailing power hypothesis. RAND Journal of Economics, 34, 612–625. Chen, Z. (2004). Monopoly and product diversity: The role of retailer countervailing power. Carleton Economic Papers 04–19. Carleton University. Clarke, R., Davies, S., Dobson, P. W., & Waterson, M. (2002). Buyer power and competition in European food retailing. Cheltenham: Edward Elgar. Dana, J. (2003). Buyer group as strategic commitments. Mimeo. Northwest University. Dobson, P. W., & Waterson, M. (1997). Countervailing power and consumer prices. Economic Journal, 107, 418–430. Dobson, P. W., Waterson, M., & Chu, A. (1998). The welfare consequences of the exercise of buyer power. Office of Fair Trading Research Paper 16. Erutku, C. (2005). Buying power and strategic interactions. Canadian Journal of Economics, 38, 1160–1172. European Commission. (1999). Buyer power and its impact on competition in the food retail distribution sector of the European Union: Final report. Office for Official Publications of the European Communities, Luxembourg. Galbraith, J. K. (1952). American capitalism: The concept of countervailing power. New York: Houghton Mifflin. Hunter, A. (1958). Notes on countervailing power. Economic Journal, 68, 89–103. Inderst, R., & Shaffer, G. (2004). Retail mergers, buyer power and product variety. CEPR Discussion Paper 4236. Inderst, R., & Wey, C. (2004). Buyer power and supplier incentives. Mimeo. London School of Economics. Inderst, R., & Wey, C. (2005). How strong buyers spur upstream innovation. Mimeo. London School of Economics. Kirkwood, J. B. (2005). Buyer power and exclusionary conduct: Should Brooke Group set the standards for buyer-induced price discrimination and predatory bidding? Antitrust Law Journal, 72, 625–668. Mathewson, G. F., & Winter, R. A. (1997). Buyer groups. International Journal of Industrial Organization, 15, 137–164. Noll, R. G. (2005). ‘Buyer power’ and economic policy. Antitrust Law Journal, 72, 589–624. OECD. (1981). Buying power: The exercise of market power by dominant buyers. Report of the Committee of Experts on Restrictive Practices. OECD. (1998). Buyer power of large scale multiproduct retailers. Background paper by the Secretariat, Roundtable on buyer power. Shaffer, G. (1991). Slotting allowances and resale price maintenance: A comparison of facilitating practices. RAND Journal of Economics, 22, 120–135. Snyder, C. M. (1996). A dynamic theory of countervailing power. RAND Journal of Economics, 27, 747–769. Stigler, G. J. (1954). The economist plays with blocks. American Economic Review Papers and Proceedings, 44, 7–14.
40
ZHIQI CHEN
Veendorp, E. C. H. (1987). Oligoemporistic competition and the countervailing power hypothesis. Canadian Journal of Economics, 20, 519–526. Viscusi, W. K., Harrington, J. E., Jr., & Vernon, J. M. (2005). Economics of regulation and antitrust (4th ed.). Cambridge: MIT Press. von Ungern-Sternberg, T. (1996). Countervailing power revisited. International Journal of Industrial Organization, 14, 507–520.
GENERALIZED CRITICAL LOSS FOR MARKET DEFINITION$ Malcolm B. Coate and Mark D. Williams ABSTRACT This paper generalizes the critical loss concept of Harris and Simons to account for a broader range of possible cost structures. Our analysis presents a specialized market-level equilibrium for a relatively homogeneous good in which the Harris and Simons’ critical loss structure is appropriate for market definition. Then, we broaden the equilibrium and propose a generalized critical loss analysis. Of course, for relatively differentiated goods, market definition analysis would use firm-level modeling and therefore the standard market-level critical loss modeling could be inappropriate.
1. INTRODUCTION For over 15 years, critical loss analysis has played an important role in the market definition analysis. Introduced by Barry Harris in the geographic market analysis of the Occidental/Tenneco merger and further developed in
$
The analyses and conclusions set forth in this paper are those of the authors and do not purport to represent the views of the Federal Trade Commission, any individual Commissioner or any Commission Bureau.
Research in Law and Economics, Volume 22, 41–58 Published by Elsevier Ltd. ISSN: 0193-5895/doi:10.1016/S0193-5895(06)22003-7
41
42
MALCOLM B. COATE AND MARK D. WILLIAMS
a paper by Harris and Simons (1989), the analysis computed the maximum level of lost sales (‘‘the critical loss’’) that could be profitably accepted by a hypothetical cartel when applying the Merger Guidelines-based small, but significant and non-transitory increase in price (SSNIP) test. Evidence, showing the hypothetical cartel (i.e. the hypothetical monopolist) would lose more than the critical level of sales if a SSNIP occurred, would lead to a rejection of the proposed market definition. By contrast, if a SSNIP appeared profitable, the proposed market would be accepted as relevant and a full investigation would be needed to determine the merger’s likely competitive effect. Critical loss analysis has played an important role in a number of merger litigations, as well as countless merger enforcement investigations.1 Critical loss methodology is best seen as an empirical application of the SSNIP market definition test that serves to improve the understanding of certain complex real world situations. Scheffman and Simons (2003) recently noted that critical loss is nothing more than a simple application of arithmetic designed to explore the likelihood of a price increase. However, it is important to understand that the basic methodology presumes that a group of firms can set a single market price, therefore, the goods traded in the market must be relatively homogeneous. If the products, offered for sale in the market, are differentiated, the simplifying assumptions of critical loss analysis will not apply and therefore a naı¨ ve application of the analysis may be misleading.2 In this paper, we define a set of conditions under which the critical loss structure defined by Harris and Simons will accurately reflect the realities facing firms in homogeneous product markets and then provide a method to generalize the analysis to a more complex industry cost structure.3 We start with an exploration of the theoretical support for the standard critical loss methodology within a homogeneous goods framework. Assuming a Bertrand equilibrium structure, we observe that the original Harris–Simons calculation for the Merger Guidelines market definition algorithm may be based on a capacity-constrained equilibrium.4 Our presentation then generalizes the industry cost structure by adding an elasticity parameter for marginal cost to the critical loss algorithm. As shown below, this generalization results in a small change in the value of critical loss for some elasticity parameters and a large change for other parameters. Thus, the need to apply the sophisticated version of the critical loss concept turns on the underlying cost structure of the market in question.
43
Generalized Critical Loss for Market Definition
2. INTRODUCTION TO CRITICAL LOSS ANALYSIS Standard critical loss analysis represents a systematic structure with which to evaluate the impact of the Merger Guidelines’ SSNIP hypothetical in a relatively homogeneous market. The technique first calculates the critical loss (i.e. the level of foregone sales necessary to just offset the profitability of a SSNIP) and then collects evidence to determine if this critical loss in sales is likely to occur. The Harris and Simons critical loss paper illustrates the basic critical loss construct as a pricing problem for the hypothetical industry monopolist. Fig. 1 presents the essence of Harris and Simons’ Fig. 2. It can be readily seen that as prices increase from P0 to P1 and output decreases from Q0 to Q1, industry profits may change. The restriction in output costs the industry the profit associated with the area between the pre-merger price (P0) and the industry marginal cost (MC) curve defined over the distance between the restricted (Q1) and competitive (Q0) levels of output. The higher price generates additional industry profits associated with the box whose area is determined by the size of the price increase multiplied by the restricted level of output (box is P1P0Q1). The essence of critical loss analysis is to find the percentage decrease in volume (X) that would make a monopolist indifferent
Price
MC P1 P0 =MC0 AVC AVC0 AVC1 D
Q1
Fig. 1.
Q0
Standard Market Model.
Quantity
44
MALCOLM B. COATE AND MARK D. WILLIAMS
to raising prices Y percent from current, competitive level of P0 to the less than competitive level of P1. Fig. 1 illustrates both the industry demand and industry supply curves. In homogeneous goods industries, the industry supply (or marginal cost) curve is arrived at by horizontally summing all the individual firm supply curves. Furthermore, an upward sloping industry supply curve implies that each individual firm’s supply curve is upward sloping at the relevant point. The Bertrand assumption on individual firm’s behavior simply requires that each firm operates at a point where price equals marginal cost. While free entry and exit would cause firms to not earn economic profits in the long run, neither the model nor the critical loss framework requires that assumption. Note that the critical loss is purely an empirical concept, determined by the slope of the industry demand curve (hence the elasticity) and the shape of the marginal cost curve (hence the value for the average margin) and these two variables are unrelated in homogeneous goods markets. Thus, it is possible to have both high industry margins and elastic industry demand.5 Firm-level modeling is not relevant, as the competitive assumption implies that the firm-level margin is zero and the firm demand elasticity is infinite at the market equilibrium point. Such point analysis is not considered in the standard critical loss, as the Guidelines hypothetical monopolist test mandates the evaluation of a discrete pricing decision with an interval estimate of the margin. Once the problem is considered in this light, it is clear that each firm generates positive margins on the infra-marginal output lost to a SSNIP. Formally, the critical loss computation starts with the indifferent relationship between the profits in the base case and the profits associated with the higher, less competitive price. The relationship is expressed as Z Q0 Z Q1 P 0 Q0 MC Fixed Costs ¼ P1 Q1 MC Fixed Costs (1) 0
0
where Q1 ¼ Q0(1X) and P1 ¼ P0(1+Y). The equation can be transformed as R Q1 R Q0 Y 0 MC 0 MC X¼ þ (2) P0 Q0 ð1 þ Y Þ 1þY Recognizing that average variable cost (AVC) is the area under the marginal cost curve divided by Q, Eq. (2) can be re-expressed in terms of AVC as X¼
YP0 þ AVC0 AVC1 P0 þ YP0 AVC1
(3)
Generalized Critical Loss for Market Definition
45
Harris and Simons note that while accounting data may contain estimates of the current variable contribution margin (P0AVC0)/P0, it is doubtful that an estimate of average variable cost at the less than competitive price of P1 (AVC1) will be available. Thus, they make the assumption that AVC0 ¼ AVC1. With this assumption: X¼
YP0 Y ¼ P0 þ YP0 AVC0 Y þ AVCM
(4)
where AVCM, the average variable contribution margin, is defined as (P0AVC0)/P0. This formula allows the analyst to calculate the critical loss for any combination of margin and hypothetical SSNIP. Assuming a profit margin of 50 percent and using a 5 percent SSNIP, it is possible to compute the critical loss at 9.1 percent. The critical elasticity simply divides the loss by the 5 percent price increase to generate a value of 1.82. A fundamental problem with this analysis arises from the geometry of the marginal/average variable cost relationship. (This appears to be implicitly recognized by the Harris and Simons in their footnote 17.) If points on the marginal cost curve are above the comparable values on the average variable cost curve, the AVC curve will be increasing with Q. Similarly, a marginal cost value below the average variable cost value will cause the AVC curve to decrease with Q. The only way (with ‘‘well-behaved’’ marginal costs curves) for AVC0 to equal AVC1 is to have a flat and coincident marginal and average variable cost curve. However, if the marginal cost and average cost curves are flat and equal, the assumption that the pre-merger price is competitive (i.e. P ¼ MC) leads to a zero variable contribution margin for the relevant market. Harris and Simon ignored that problem, probably because they were trying to keep their procedure simple. One solution to this technical problem is to construct a relationship between marginal and average cost curves that allows for positive margins for the interval associated with the (SSNIP-caused) output reduction but retains the assumption that price equals marginal cost at the market clearing point. This scenario is illustrated in Fig. 2. For output levels up to Q0, marginal cost is constant, hence marginal and average costs coincide. However, for output levels above Q0, marginal cost becomes infinite. As the marginal cost curve cuts the demand curve, the market price equals marginal cost at the market-clearing level of output Q0. Two possible models could justify this equilibrium structure. First, a capacity constraint could limit the industry to specific level of output (Q0). Bertrand competition would define prices that exceed marginal costs and contribute a margin to cover fixed costs.6 If the market offered premium
46
MALCOLM B. COATE AND MARK D. WILLIAMS
Price
MC=AVC
P1
P0= MC
AVC0=AVC1
D
Q1
Fig. 2.
Q0
Quantity
Capacity Constraint Market Model.
returns, the profit would attract entry, shift the capacity constraint out and reduce price to a zero profit equilibrium level in the long run. Second, a dynamic model could be applied in which long-run cost considerations fix the price at P0. The firms earn a short-run margin on all units of output as short-run marginal costs remain below price. Output cannot be expanded beyond Q0, because each firm lacks the ability to commit to serve new customers in the long run. As the market price would cover all the relevant costs, short-run margins would not attract entry.7 In either case, the critical loss analysis would model the short-run choice facing a hypothetical cartel of market participants. Given the critical loss, it would be possible to collect data on likely substitution linked to a price increase, compute the actual loss, and then either accept or reject the proposed market definition depending on how this actual loss compares to the computed critical loss.
3. GENERALIZED CRITICAL LOSS The fundamental structure of critical loss can be adjusted to allow for a more general cost structure. Manipulation of Eq. (2) shows how to
Generalized Critical Loss for Market Definition
47
perform an exact calculation for the critical loss. The equation can be expressed as R Q0 Y Q1 MC þ (5) X¼ 1 þ Y P0 Q0 ð1 þ Y Þ can define the average marginal cost (AMC) as AMC ¼ R QOne 0 MC=XQ 0 ; which is literally the average of marginal costs between Q1 Q1 and Q0. Eq. (5) can be rewritten as AMC Y (6) ¼ X 1 P0 ð1 þ Y Þ 1þY If one defines the average marginal contribution margin as AMCM ¼ (P0AMC)/P0, one finds the correct measure of critical loss to be X¼
Y Y þ AMCM
(7)
This implies that the critical loss tables in Harris and Simons are correct if one uses the average marginal contribution in place of the average variable contribution margin. Note that as long as the industry quantity is significantly to the right of where the MC and AVC curves cross, even when output is reduced, a rising marginal cost curve is likely to be quite a bit above the average variable cost curve. Assuming the upward sloping marginal cost model is appropriate, the critical loss is likely to be underestimated using average variable contribution margins. Of course, the problem with using marginal rather than average costs is that one is unlikely to find good accounting data on marginal costs. What is needed is an appropriate way to convert between accounting measures of average costs and actual marginal costs. With that in mind, first consider the following homogeneous good, price-setting model shown in Fig. 3. Here, the marginal cost curve is linear with slope m and intercepts the demand curve at (Q0, P0).8 Because the marginal cost curve is linear, the average marginal cost over the segment between Q1 and Q0 is the average of the marginal costs at Q1 and Q0. In other words, AMC ¼ P0mXQ0/2. From this, we can calculate the average marginal contribution margin: AMCM ¼
P0 AMC P0 ðP0 mX Q0 =2Þ mX Q0 X 0MC ¼ ¼ ¼ P0 P0 2 2P0
(8)
48
MALCOLM B. COATE AND MARK D. WILLIAMS
Price
MC
P1
P0=MC0 AMC= P0-mXQ0/2
MC1 D
Q1` (Q0+Q1)/2
Fig. 3.
Q0
Quantity
Illustration of Average Marginal Cost Concept.
where 0MC is the elasticity of the marginal cost curve at Q0.9 Solving for X, one finds X¼
Y Y ¼ AMCM þ Y ðX 0MC =2Þ þ Y
(9)
which can be expressed as 2 1 0 2MC X
þ YX Y ¼ 0
The quadratic formula implies that10 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Y ðY þ 20MC Þ Y X¼ 0MC
(10)
(11)
It is a simple matter to evaluate the formula for a range of values for the price increase (Y) and the marginal cost elasticity, 0MC : The relevant critical losses are given in Table 1. This analysis defines the general critical loss in a homogeneous market with linear marginal costs and the assumption that pre-merger prices equate to marginal cost at the competitive level of output. In some cases
49
Generalized Critical Loss for Market Definition
Table 1.
Critical Loss for Linear Marginal Cost Curve.
Elasticity of Marginal Cost
Price Increase (%) 5.0% 91.6 73.2 61.8 50.0 46.3 35.8 27.0 20.0 14.6 13.2 9.5 N/A N/A N/A
0.01 0.05 0.1 0.2 0.25 0.5 1 2 4 5 10 20 100 1000
10.0% 95.4 82.8 73.2 61.8 58.0 46.3 35.8 27.0 20.0 18.1 N/A N/A N/A N/A
15.0% 96.9 87.3 79.1 68.6 64.9 53.1 41.8 31.9 23.9 N/A N/A N/A N/A N/A
20.0% 97.6 89.9 82.8 73.2 69.7 58.0 46.3 35.8 N/A N/A N/A N/A N/A N/A
25.0% 98.1 91.6 85.4 76.6 73.2 61.8 50.0 39.0 N/A N/A N/A N/A N/A N/A
(marked N/A), the required critical loss would push the linear marginal cost function below zero and hence the hypothetical calculation is not valid. This negative marginal cost problem suggests that a more sophisticated cost model is required to create a generalized critical loss structure. Such a model would allow marginal cost to equal average variable cost for a range of output values but then increase linearly so the standard market equilibrium could be generated. This scenario is illustrated in Fig. 4. This revised specification changes the formula for the average marginal cost. Operationally, it is the area under the marginal cost curve from Q1 to Q0 divided by the distance between Q1 and Q0, which is XQ0. Call the height of the flat part of the average variable cost curve AVC. In this case, the length of the bottom of the triangle under the MC curve but above AVC is (P0AVC)/m where m is the slope of the MC curve.11 The area of that triangle is (P0AVC)2/2m, which implies that AMC ¼ AVC þ
ðP0 AVCÞ2 ðP0 AVCÞ2 ¼ AVC þ 2mX Q0 2XP0 0MC
(12)
Once again, one can define the average marginal contribution margin as AMCM ¼
P0 AMC ðP0 AVCÞ ½ðP0 AVCÞ2 =2XP0 0MC ¼ P0 P0
(13)
50
MALCOLM B. COATE AND MARK D. WILLIAMS
Price
MC
P1
AVC P0= MC
AVC
Q1
Q0
Fig. 4.
Quantity
Generalized Market Model.
Now define the average variable contribution margin as AVCM ¼ (P0AVC)/P0.12 This implies that AVCM2 AVCM AMCM ¼ AVCM ¼ AVCM 1 (14) 2X 0MC 2X 0MC To solve for the critical loss, one must solve the following for X X¼
Y Y ¼ Y þ AMCM Y þ AVCM ðAVCM2 =2X 0MC Þ
(15)
This equation is linear in X. Solving the equation for X yields X¼
20MC Y þ AVCM2 20MC ðAVCM þ Y Þ
(16)
Note that as the elasticity of the marginal cost curve goes to infinity (i.e. the model goes to an L-shaped marginal cost curve), the results converge to the Harris and Simons calculation. Tables 2 and 3 present critical losses for 5 and 10 percent price increases, given the cost elasticity and the margin.
Generalized Critical Loss for Market Definition
51
Table 2. Critical Loss for Generalized Marginal Cost Curve (If Output is in Flat Portion of Marginal Cost Curve) (5% Price Increase). Elasticity of Marginal Cost
0.01 0.05 0.1 0.2 0.25 0.5 1 2 4 5 10 20 100 1000
Variable Contribution Margin (%)
10.0% 366.7 100.0 66.7 50.0 46.7 40.0 36.7 35.0 34.2 34.0 33.7 33.5 33.4 33.3
20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 820.0 1300.0 1788.9 2281.8 2776.9 3273.3 3770.6 4268.4 180.0 271.4 366.7 463.6 561.5 660.0 758.8 857.9 100.0 142.9 188.9 236.4 284.6 333.3 382.4 431.6 60.0 78.6 100.0 122.7 146.2 170.0 194.1 218.4 52.0 65.7 82.2 100.0 118.5 137.3 156.5 175.8 36.0 40.0 46.7 54.5 63.1 72.0 81.2 90.5 28.0 27.1 28.9 31.8 35.4 39.3 43.5 47.9 24.0 20.7 20.0 20.5 21.5 23.0 24.7 26.6 22.0 17.5 15.6 14.8 14.6 14.8 15.3 15.9 21.6 16.9 14.7 13.6 13.2 13.2 13.4 13.8 20.8 15.6 12.9 11.4 10.5 9.9 9.6 9.5 20.4 14.9 12.0 10.2 9.1 8.3 7.8 7.4 20.1 14.4 11.3 9.3 8.0 7.0 6.3 5.7 20.0 14.3 11.1 9.1 7.7 6.7 5.9 5.3
Table 3. Critical Loss for Generalized Marginal Cost Curve (If Output is in Flat Portion of Marginal Cost Curve) (10% Price Increase). Elasticity of Marginal Cost
0.01 0.05 0.1 0.2 0.25 0.5 1 2 4 5 10 20 100 1000
Variable Contribution Margin (%)
10.0% 300.0 100.0 75.0 62.5 60.0 55.0 52.5 51.3 50.6 50.5 50.3 50.1 50.0 50.0
20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 700.0 1150.0 1620.0 2100.0 2585.7 3075.0 3566.7 4060.0 166.7 250.0 340.0 433.3 528.6 625.0 722.2 820.0 100.0 137.5 180.0 225.0 271.4 318.8 366.7 415.0 66.7 81.3 100.0 120.8 142.9 165.6 188.9 212.5 60.0 70.0 84.0 100.0 117.1 135.0 153.3 172.0 46.7 47.5 52.0 58.3 65.7 73.8 82.2 91.0 40.0 36.3 36.0 37.5 40.0 43.1 46.7 50.5 36.7 30.6 28.0 27.1 27.1 27.8 28.9 30.3 35.0 27.8 24.0 21.9 20.7 20.2 20.0 20.1 34.7 27.3 23.2 20.8 19.4 18.6 18.2 18.1 34.0 26.1 21.6 18.8 16.9 15.6 14.7 14.1 33.7 25.6 20.8 17.7 15.6 14.0 12.9 12.0 33.4 25.1 20.2 16.9 14.5 12.8 11.5 10.4 33.3 25.0 20.0 16.7 14.3 12.5 11.1 10.0
The output reductions of over 100 percent are not relevant, because this could imply negative production. Moreover, the model only makes sense when the lower, post-price increase, level of output (Q1) falls on the flat portion of the marginal cost curve. To address this issue, we undertake
52
MALCOLM B. COATE AND MARK D. WILLIAMS
another calculation. For each pair of parameters (contribution margin and marginal cost elasticity), we compute the required reduction in quantity necessary for the marginal cost function to fall to the level of the fixed average variable cost (associated with the set contribution margin). These results are shown in Table 4 (the symbol ‘‘*’’ implies the required output reduction exceeds 100 percent and therefore is impossible). Not surprisingly, as contribution margins increase (average costs fall) and it becomes more difficult for relatively elastic (flat) marginal cost functions to fall quickly enough to reach the relatively low variable cost functions while still predicting positive levels of output Q. Obviously, if the required loss from Table 4 exceeds 100 percent, the kinked marginal cost curve model operationalized in Tables 2 and 3 cannot be used. Moreover, the critical loss values in Tables 2 and 3 are relevant only if they exceed the comparable loss from Table 4. Whenever the computed critical loss (in Tables 2 and 3) is smaller than the required loss (from Table 4), the appropriate critical loss figures are given in Table 1 (as the firm incurs its critical loss while on the downward sloping portion of the marginal cost curve). Table 2 is reproduced as Table 5, with the critical losses from Table 1 substituted in when appropriate. Table 3 is revised in a comparable manner as Table 6 for a 10 percent price effect.
Table 4.
Minimum Required Reduction in Output to Use Kinked Cost Curve.
Elasticity of Marginal Cost
0.01 0.05 0.1 0.2 0.25 0.5 1 2 4 5 10 20 100 1000
Variable Contribution Margin (%)
10.0% * * * 50.0 40.0 20.0 10.0 5.0 2.5 2.0 1.0 0.5 0.1 0.0
20.0% * * * * 80.0 40.0 20.0 10.0 5.0 4.0 2.0 1.0 0.2 0.0
30.0% * * * * * 60.0 30.0 15.0 7.5 6.0 3.0 1.5 0.3 0.0
40.0% * * * * * 80.0 40.0 20.0 10.0 8.0 4.0 2.0 0.4 0.0
50.0% * * * * * * 50.0 25.0 12.5 10.0 5.0 2.5 0.5 0.1
60.0% * * * * * * 60.0 30.0 15.0 12.0 6.0 3.0 0.6 0.1
70.0% * * * * * * 70.0 35.0 17.5 14.0 7.0 3.5 0.7 0.1
80.0% * * * * * * 80.0 40.0 20.0 16.0 8.0 4.0 0.8 0.1
90.0% * * * * * * 90.0 45.0 22.5 18.0 9.0 4.5 0.9 0.1
53
Generalized Critical Loss for Market Definition
Table 5.
Corrected Critical Loss for Generalized Marginal Cost Curve (5% Price Increase).
Elasticity of Marginal Cost
0.01 0.05 0.1 0.2 0.25 0.5 1 2 4 5 10 20 100 1000
Table 6.
Variable Contribution Margin (%)
10.0% 91.6 73.2 61.8 50.0 46.7 40.0 36.7 35.0 34.2 34.0 33.7 33.5 33.4 33.3
20.0% 91.6 73.2 61.8 50.0 46.3 35.8 28.0 24.0 22.0 21.6 20.8 20.4 20.1 20.0
40.0% 91.6 73.2 61.8 50.0 46.3 35.8 27.0 20.0 15.6 14.7 12.9 12.0 11.3 11.1
50.0% 91.6 73.2 61.8 50.0 46.3 35.8 27.0 20.0 14.8 13.6 11.4 10.2 9.3 9.1
60.0% 91.6 73.2 61.8 50.0 46.3 35.8 27.0 20.0 14.6 13.2 10.5 9.1 8.0 7.7
70.0% 91.6 73.2 61.8 50.0 46.3 35.8 27.0 20.0 14.6 13.2 9.9 8.3 7.0 6.7
80.0% 91.6 73.2 61.8 50.0 46.3 35.8 27.0 20.0 14.6 13.2 9.6 7.8 6.3 5.9
90.0% 91.6 73.2 61.8 50.0 46.3 35.8 27.0 20.0 14.6 13.2 9.5 7.4 5.7 5.3
Corrected Critical Loss for Generalized Marginal Cost Curve (10% Price Increase).
Elasticity of Marginal Cost
0.01 0.05 0.1 0.2 0.25 0.5 1 2 4 5 10 20 100 1000
30.0% 91.6 73.2 61.8 50.0 46.3 35.8 27.0 20.7 17.5 16.9 15.6 14.9 14.4 14.3
Variable Contribution Margin (%)
10.0% 95.4 82.8 73.2 62.5 60.0 55.0 52.5 51.3 50.6 50.5 50.3 50.1 50.0 50.0
20.0% 95.4 82.8 73.2 61.8 58.0 46.7 40.0 36.7 35.0 34.7 34.0 33.7 33.4 33.3
30.0% 95.4 82.8 73.2 61.8 58.0 46.3 36.3 30.6 27.8 27.3 26.1 25.6 25.1 25.0
40.0% 95.4 82.8 73.2 61.8 58.0 46.3 35.8 28.0 24.0 23.2 21.6 20.8 20.2 20.0
50.0% 95.4 82.8 73.2 61.8 58.0 46.3 35.8 27.1 21.9 20.8 18.8 17.7 16.9 16.7
60.0% 95.4 82.8 73.2 61.8 58.0 46.3 35.8 27.0 20.7 19.4 16.9 15.6 14.5 14.3
70.0% 95.4 82.8 73.2 61.8 58.0 46.3 35.8 27.0 20.2 18.6 15.6 14.0 12.8 12.5
80.0% 95.4 82.8 73.2 61.8 58.0 46.3 35.8 27.0 20.0 18.2 14.7 12.9 11.5 11.1
90.0% 95.4 82.8 73.2 61.8 58.0 46.3 35.8 27.0 20.0 18.1 14.1 12.0 10.4 10.0
Critical losses can be evaluated for any combination of cost elasticity and variable contribution margin by solving the relevant equations. Similarly, critical losses for parameter values not in the table can be computed from the tables.
54
MALCOLM B. COATE AND MARK D. WILLIAMS
4. APPLYING CRITICAL LOSS ANALYSIS The revisions to the critical loss construct increase the analyst’s flexibility in evaluating a merger. To use critical loss to appropriately define the relevant market, the analyst must first understand the equilibrium process for the market in question. One crucial issue involves deciding whether the market in question is best analyzed using a homogeneous or differentiated goods model.13 To the extent the product in question is relatively homogeneous, either the standard critical loss analysis or our generalized critical loss analysis is appropriate. In contrast, if the product under review is relatively heterogeneous, some type of firm-level equilibrium model would be more appropriate. Within a homogeneous market, it is necessary to identify the equilibrium cost structure to apply the critical loss analysis. Harris and Simons propose a simplification in which price is effectively set independently of short-run cost. For example, the market might face relatively tight capacity constraints or the evidence could suggest that market participants consider long-run issues when setting prices. If either of these assumptions is appropriate, the standard critical loss analysis should be applied to determine the short-run profitability of a SSNIP. Our paper generalizes the critical loss construct for a more conventional short-run marginal cost equilibrium. To apply our model, it is first necessary to obtain some estimate of the relatively constant variable contribution margin on the flat portion of the marginal cost curve. This may very well be the same proxy used in the Harris and Simons’ analysis.14 Next, it is necessary to estimate the elasticity of the marginal cost curve at the pre-merger competitive price. In effect, this requires the economist to obtain some understanding of the pricing equilibrium before assuming the standard short-run marginal cost equilibrium applies to the market in question. One would expect the analyst to identify exactly what factor of production becomes scarce as the market moves to equilibrium and then estimate its rate of increase. In general, the relevant cost elasticity will be quite inelastic, as marginal costs must fall quickly to the constant variable cost for the constant costs to be quantifiable at the industry level. To the extent the elasticity cannot be identified, one could pick a range of values and select a couple of critical loss values from either Table 5 (5 percent) or Table 6 (10 percent). Critical loss statistics could be significantly higher than those associated with the standard Harris and Simons’ work, but not prohibitively high. An interested reader could compare the last row in either Table 5 or 6 with the third or fourth row from the bottom to obtain a feel for the effect. Once the generalized critical loss is computed, the analyst follows the same steps as
Generalized Critical Loss for Market Definition
55
advocated by Harris and Simons to evaluate the extent of the expected substitution. If the product market is considered highly differentiated, a firm-level model should be used to replace the industry-level model implicitly used in standard critical loss analysis.15 Here, the analyst needs to consider the linkage between the margin and the firm’s own-price elasticity. Critics note the observation of high margins necessarily implies relatively inelastic firmlevel demand curves and thus impose severe theoretical limits on empirical estimates of critical loss. While a detailed criticism of this analysis is beyond the scope of this paper, applications of differentiated product models run the risk of serious error if their implicit assumptions are violated.16
5. CONCLUSION This paper highlights the importance of understanding the equilibrium structure of the market before undertaking a critical loss analysis. To the extent a homogeneous market actually clears at the classic price equal shortrun marginal cost condition, the standard critical loss methodology tends to generate markets that are too broad. Our analysis shows that the standard critical loss values approximate the correct values more closely for lowmargin industries. Of course, if short-run marginal costs play little role in driving competition in a homogeneous goods market, then the standard Harris and Simons model is likely to offer reasonable results, as the lost short-run profits can be proxied with average margin data. Finally, critical loss analysis, as presented by Harris and Simons and generalized in our paper is not appropriate if the market offers heterogeneous products. To the extent the products in the market are relatively differentiated, more complex firm-level equilibrium models must be used. However, in using these models, care must be taken to ensure the actual equilibrium structure matches the theoretical equilibrium structure assumed in the model. In market definition, facts must control theory.
NOTES 1. See, for example, California v. Sutter Health System 84 F. Supp. 2d 1057 (N. D. Cal.) aff’d 217 F. 3d 846 (9’th Cir. 2000) opinion amended 130 F. Supp. 2d 1109 (N. D. Cal. 2001); FTC v. Tenet 17 F. Supp. 2d 937 aff’d 186 F. 3rd 1045 (8’th Cir. 1999), U.S. vs. Mercy Health Services 902 F. Supp. 968 (1995). Critical loss is also used in nonhealth care mergers. See, U.S. v. Sungard Data Systems. Inc 172 F. Supp. 2d 172 (2001).
56
MALCOLM B. COATE AND MARK D. WILLIAMS
2. In a differentiated goods market, the demand curve is a function in N+1 space, where N is the number of relevant spatial attributes that serve to define the product. A uniform SSNIP across all products will likely result in different sales losses for each firm. Although the price increase could prove either profitable or unprofitable to the hypothetical differentiated goods monopolist, the standard critical loss analysis may be misleading. 3. A recent review of Federal Trade Commission enforcement activity found that homogeneous goods theories were used in 42 of the 124 sampled merger investigations (Coate, 2006). 4. The Bertrand assumption ensures the initial equilibrium underpinning the critical loss calculation is competitive, where price equals MC at the appropriate firm and industry output levels. Some analysts favor a Cournot assumption for homogeneous goods. The Cournot model defines a Nash equilibrium for a homogeneous goods market when all the participants are sure that their rivals are committed to holding their level of production fixed in response to any output decision by another firm. In effect, the firm nets their rivals’ output functions out of the market demand and optimizes on the residual demand curve to compute its desired level of output. Given equal costs, all firms simultaneously undertake the same analysis, so each firm chooses the same Cournot output strategy. While such a model appears unilateral, this structure is purely an artifact of modeling the firm’s output decision. The commitment necessary to hold output constant in response to any competitive action is obtained via assumption. Cournot firms, by definition, do not allow customers to buy extra product at fixed prices. In effect, the Cournot assumption implies that some institutional consideration must exist to preclude the firms in the market from meeting consumer demands by selling desired output. An agreement among the firms to play Cournot would obviously be sufficient to justify the Cournot assumption, but absent such a finding (or other evidence supportive of Cournot), the more competitive Bertrand assumption seems the appropriate default. 5. This observation could not be made for a differentiated goods market, as the Lerner index equation links the price-cost margin at the firm level with the firm-level elasticity. 6. The Bertrand model supports the competitive outcome, as the rivals cannot restrict output to raise price under a fixed strategy structure. However, mixed strategy equilibriums will generate less competitive pricing (Levitan & Shubik, 1972). This less than competitive outcome could not occur if rivals cannot commit to use an output restriction strategy. For example, under some cost conditions, the monopoly price associated with the residual demand curve facing a firm is equal to (or less than) the constrained price P0. As long as the firms offer to sell out their capacity at any price at or above P0, the competitive output will evolve. While one could certainly build a Cournot model around a capacity constraint, this generalization would not follow the competitive orientation of the Harris and Simons model. (Moreover, for some parameter values, the Cournot price would be below the Bertrand price and thus, the Cournot model would have to be adjusted to use the market-clearing Bertrand prices.) 7. The Harris and Simons graph shows average total cost equal to marginal cost at the pre-merger competitive level of the output. If the marginal cost curve is the long-run marginal cost curve, a short-run marginal cost curve could also exist and approximate average variable cost.
Generalized Critical Loss for Market Definition
57
8. The assumption that marginal cost curve rises implies that firms cannot replicate the costs of earlier units and therefore there must be some shortage of at least some input so that increasing production drives up the cost of the input. In other words, there is some capacity constraint on production. Allowing the marginal cost curve to be linear, but not vertical, implies that the capacity constraint is more flexible than the L-shaped MC curve. 9. Note the use of the point elasticity of marginal cost at the competitive price makes the problem unitless. This does not imply that the marginal cost curve has constant elasticity. The only way a linear marginal cost curve will have constant elasticity is if the curve passes through the origin. 10. Note that X is defined to be positive and therefore one should only consider the positive root. 11. Of course, the parameters of the model are constrained, so the specification of the increasing marginal cost function will actually equal the flat portion of the marginal cost function for a level of output between 0 and Q0. 12. Note that at Q0, the true average variable cost is not AVC. It is our belief that accounting data is more likely to reflect the flat part of the curve. However, if one believes that this is not the case, one can convert between the two if one knows the slope of the marginal cost curve. 13. Some seemingly differentiated products may best be analyzed with a homogeneous goods model. If each participant in the market can quickly make a product with any set of characteristic as cheaply as the incumbent supplier at that point characteristic space, then the product may be considered homogeneous on the supply side. 14. Firm-level accounting data evaluates costs over a range of output levels, so a homogeneous goods market will often generate substantial margins for the relevant interval. Point price-cost margins may be zero, but will be hard to measure with accounting data. Our structure, along with that of Harris and Simons minimizes the need to measure the actual point margin. 15. See, for example, Danger and Frech (2001), Katz and Shapiro (2003), and O’Brien and Wickelgren (2003). 16. This analysis assumes (1) the firm-specific elasticity is relevant (i.e. the products in the market are relatively differentiated and there are no supply-side effects from other firms in the time horizon contemplated by the elasticity measurement), (2) market optimization with respect to price (i.e. Bertrand strategies give rise to a premerger competitive equilibrium), (3) differentiability of the relevant functions (i.e. hence no discontinuities), (4) the time horizon of the optimization matches that implicit in the margin (i.e. this generally requires the assumption of a short-run optimization model), and (5) no current exercise of market power through coordinated behavior. For a detailed critique of these models, see Coate and Williams (2005).
ACKNOWLEDGMENTS We would like to thank Jeffrey Fischer and other anonymous members of the Bureau of Economics staff for helpful comments on this manuscript.
58
MALCOLM B. COATE AND MARK D. WILLIAMS
REFERENCES Coate, M. B. (2006). Economic models and the merger analysis: A case study. Review of Law and Economics, 2, 53–84. Coate, M. B., & Williams, M. D. (2005). A critical commentary on the critical comments on critical loss. Federal Trade Commission. Danger, K. L., & Frech, H. E. (2001). Critical thinking about critical loss in antitrust. Antitrust Bulletin, 46, 339–355. Harris, B., & Simons, J. (1989). Focusing market definition: How much substitution is enough. Research in Law and Economics, 12, 207–226. Katz, M., & Shapiro, C. (2003). Critical loss let’s tell the whole story. Antitrust, 17, 49–56. Levitan, R., & Shubik, M. (1972). Price duopoly and capacity constraints. International Economic Review, 13, 111–122. O’Brien, D., & Wickelgren, A. (2003). A critical analysis of critical loss analysis. Antitrust Law Journal, 71, 161–184. Scheffman, D., & Simons, J. (2003). The state of critical loss analysis: Let’s make sure we understand the whole story. The Antitrust Source: http://www.abanet.org/antitrust/ source/nov03/scheffman.pdf
PRICE-FIXING OVERCHARGES: LEGAL AND ECONOMIC EVIDENCE John M. Connor ABSTRACT This paper surveys published economic studies and judicial decisions that contain 1,040 quantitative estimates of overcharges of hard-core cartels. The primary finding is that the median long-run overcharge for all types of cartels over all time periods is 25.0%:18.8% for domestic cartels and 31.0% for international cartels. Cartel overcharges are positively skewed, pushing the mean overcharge for all successful cartels to 43.4%. Convicted cartels are on average as equally effective at raising prices as unpunished cartels, but bid-rigging conduct does display somewhat lower mark-ups than price-fixing cartels. These findings suggest that optimal deterrence requires that monetary penalties ought to be increased.
INTRODUCTION Since at least 1888, hundreds of economists, historians, commissioners, and jurists have labored mightily to assess the ‘‘effectiveness’’ of cartels. Various criteria have been applied to evaluate cartel performance, including
Research in Law and Economics, Volume 22, 59–153 Copyright r 2007 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0193-5895/doi:10.1016/S0193-5895(06)22004-9
59
60
JOHN M. CONNOR
longevity, stability, efficiency, and social welfare effects, but by far the greatest attention has been lavished on market price effects.1 The increase in transaction prices by a sellers’ cartel is commonly called an overcharge by economists or damages by legal writers. The overcharge rate is calculated by comparing actual cartel-enhanced prices to some competitive benchmark price2 (Connor, 2004a). A price-fixing overcharge is a transfer of income or wealth from buyers to the members of the cartel that occurs as a result of a collusive agreement.3 Ceteris paribus, when a cartel achieves high levels of effectiveness (i.e., longevity, stability, and high overcharges), it tends to generate large customer losses in the form of loss of consumer surplus.4 Although there are other negative effects of price fixing, legal scholarship tends to equate antitrust injury with the overcharge. Effective cartels are also viewed as destructive to the competitive process in the sense that they weaken the natural effects of demand and supply in price formation and cause deadweight social losses.5 The deadweight losses result from the costs incurred by customers when they are forced to substitute inferior substitutes, if any, the costs incurred by the members of the cartel in managing the collusive enterprise, and rent-seeking behavior by the cartel such as efforts directed at forestalling entry. In this paper, I focus on cartel overcharges as the principal type of harm or damages created by price fixing. The last dedicated survey of the cartel literature did not cover overcharges (Bullock, 1901, 1905). Today textbooks of economics conventionally devote considerable space to the market price effects of cartels (see, for example, Carlton & Perloff, 2004, pp. 128–131, 140–145, 148–150). While empirical studies of cartels routinely survey selected works as a prelude to the study being presented, to my knowledge no one has published a work aimed principally at comprehensively surveying and analyzing cartel overcharges. This paper is aimed principally at filling this gap in the legal-economic literature. The size of cartel overcharges is an issue at the empirical heart of a number of legal and economic controversies. First, knowing the size and distribution of cartel overcharges is necessary to justify the underpinnings of U.S. and foreign sanctions for illegal cartel conduct. Many commentators on government fining practices have noted the absence of appropriate empirical data for the rational design of such policies. Second, there is evidence in the economic literature widely varying opinions among experts on the critical legal-economic issue of the size of sanctions needed for optimal deterrence of cartel formation.
61
Price-Fixing Overcharges
Overcharges and Cartel Fines The Sentencing Reform Act of 1984 created the U.S. Sentencing Commission (USSC), a judicial-branch unit charged by Congress with devising guidelines for sentencing for the federal judiciary (USSG Advisory Group, 2003). The first set of guidelines was promulgated in 1987, and after three years of study and public comment was made law in 1989. The guidelines included sanctions for organizations guilty of horizontal price fixing and bid rigging (Cohen & Scheffman, 1989, p. 332). Although the Sherman Act of 1890 is a criminal statute that encompasses other types of restrictive business practices, by long tradition only horizontal price fixing and market-sharing agreements have triggered criminal indictments by the Department of Justice (DOJ).6 The issue of how high cartels typically raise prices was crucial when the USSC established the fine levels for cartels. The USSC’s cartel fine levels followed from its famous conclusion: ‘‘It is estimated that the average gain from price fixing is 10 percent of the selling price.’’7 The Commission added: ‘‘The purpose for specifying a percent of the volume of commerce is to avoid the time and expense that would be required for the court to determine actual gain or loss.’’8 As the Sixth Circuit noted, the Sentencing Commission ‘‘opted for greater administrative convenience’’ instead of ‘‘undertaking a specific inquiry into the actual loss in each case.’’9 The USSC appears to have adopted the 10% presumption because its use was advocated by the (then) head of the Antitrust Division, Douglas Ginsburg.10 The origin of Ginsburg’s 10% figure is not publicly known.11 However, a prominent analysis of the issue by Cohen and Scheffman (1989) published shortly after the Antitrust Sentencing Guidelines were promulgated, states that the economic evaluation of only three price-fixing conspiracies was particularly important in shaping Ginsburg’s views. If this analysis is correct, a critical assumption in setting cartel fines in the United States is supported by a surprisingly small amount of evidence. The USSC’s 10% presumption was attacked as unreliable and overstated almost as soon as it was issued. For example, Cohen and Scheffman (1989) concluded that ‘‘y there is little credible statistical evidence that would justify the Commission’s assumptions which underlie the Antitrust Guidelines’’ (p. 333). ‘‘At least in price fixing cases involving a substantial volume of commerce, ten percent is almost certainly too high’’ (p. 343). Moreover, the specific data that the Commission uses was characterized as exaggerated: ‘‘later research has cast considerable doubt on y these estimates, concluding that the markups, if they existed, were quite small’’ (p. 345).
62
JOHN M. CONNOR
From 1990 to 1999, a series of record corporate fines were imposed for criminal price fixing by U.S. courts, most of which can be attributed to prosecutions of international cartels; a similar upswing may be noted for fines imposed by Canada and the European Commission (EC) from 1995 to 2001 (Connor, 2004c).12 In 1990–1994, U.S. fines on international cartels were practically zero; in 1996–1997, fines averaged $69 million per year; and in 1998–1999, fines were 12 times higher (ibid., Fig. 3). Civil treble-damage cases in the United States have seen a parallel response in the size of settlements (ibid., pp. 263–264). Besides corporate penalties, a full assessment of optimal deterrence requires a consideration of corporate reputational effects of conviction and individual sanctions (Connor, 2005). Attorneys engaged in cases involving international antitrust conspiracies have argued that the Guidelines have resulted in excessive penalties. For example, just as the DOJ’s campaign against international cartels was gathering steam, Adler and Laing (1997) asserted that ‘‘the fines being imposed against corporate members of international cartels are staggering (p. 1),’’ placing the blame on the ‘‘uniquely punitive’’ requirements of the U.S. Sentencing Guidelines.13 After viewing an intensification of this trend for another two years, Adler and Laing (1999) were even more alarmed. What is y troubling is that the company fines y have risen astronomically – to levels far higher than the fines for other serious economic crimes and in amounts that can be unrelated to the economic harm14 caused by the violations (p. 1).
More recently, Denger (2003) too decries the prevalence of excessive pricefixing fines and private settlements. He places the blame for excessive fines on the Corporate Guidelines base fine calculation (p. 3). This approach, he notes, unlike all other white-collar federal crimes, means that the actual degree of direct harm caused does not have to be proven by prosecutors.15 Denger blames this state of affairs on a gap in the economic-legal literature: ‘‘y we have little information on what level of criminal or civil exposure is needed to deter most cartels (p. 4).’’ Concern about the lack of empirical evidence on the size of overcharges caused by price fixing is not confined solely to those sympathetic to the increased exposure of corporate defendants. DOJ official Graubert (2003) notes that the controversy over whether antitrust payments are excessive (which on p. 7 he equates with payouts greater than reasonable damage estimates) is largely attributable to the ‘‘y difficulty of gathering useful data.’’ In a law-review article noting the sharp increase in U.S. criminal fines on international cartels in the late 1990s, Klawiter (2001) believes that these fines and other related antitrust penalties ‘‘y have substantially increased
Price-Fixing Overcharges
63
the level of deterrence in antitrust criminal cases’’ (ibid., p. 756).16 Yet, he laments the paucity of information needed to make a more sweeping conclusion. ‘‘There are no known applicable empirical studies on the adequacy of the present mix of criminal and civil antitrust sanctions from the standpoint of deterrence’’ (ibid., fn. 79). U.S. antitrust enforcement has been a model for many other countries that have more recently adopted such laws (Wells, 2002). Germany’s revised competition law implemented in 1958 became one of the principal influences on the adoption of such statutes by the original six members of the European Economic Community (Goyder, 1998, pp. 18–33). After four years of confidential political discussions within the European Economic Communities’ (EEC) Commission, Regulation 17 was passed; it lays out the powers of the Competition Directorate General (DG-COMP) to fine companies for competition-law infringements (ibid., p. 45). That rule sets a maximum corporate fine of 10% of the company’s total sales in the year prior to the Commission’s decision and specifies that the specific fine will depend on the duration and seriousness of the offense.17 Methods of calculating EU cartel fines are explained in a 1998 Notice (Connor, 2004c, pp. 14–15). Harding and Joshua (2003) state that EU fines are supposed to incorporate both compensatory and punitive components, the latter meant to serve deterrence (p. 240). The EC considers the ‘‘gravity’’ of the offense. EU cartel fines are loosely related to overcharges because cartels with large damages that are geographically widespread add to the gravity. Also, relatively large companies are fined more than smaller participants: in several global cartels, companies in the upper half of the cartel’s size distribution had their fines doubled. After applying are number of other factors, the Commission ensures that fine amount does not exceed 10% of global sales in the year prior to the date of the decision. Rarely does the EC need to worry about reaching the 10% cap (Connor, 2003). Canada is another jurisdiction with relatively tough sentencing for cartels. The Canadian Competition Bureau (CCB) uses a fairly simple standard for setting fines. Although not spelled out in any administrative guidelines, decisions of Canadian courts have, in the absence of aggravating and mitigating circumstances, imposed fines close to 20% of Canadian affected sales (Low, 2004; Connor, 2003).18 A former Canadian prosecutor comments that ‘‘there has not been any economic or judicial analysis of the assumptions behind this proxy for harm that this represents y’’ (Low, 2004, p. 19). Cooperating firms get leniency discounts, and recently recidivists have paid fines as high as 45% of affected sales, yet the large majority of convicted cartelists pay fines equal to 20% of Canadian affected sales.
64
JOHN M. CONNOR
The Canadian 20% rule seems to mimic the base fine of the USSGs. If Canada intends to punish cartels, then the presumed overcharge may also be 10%; if only compensation is the aim, then a 20% overcharge is assumed.
Overcharges and Cartel Deterrence Concerns about the inadequacy or excessiveness of antitrust sanctions are part of the larger issue of the effectiveness of antitrust interventions. To make any headway in assessing empirically the adequacy of anticartel enforcement, it is necessary to have reliable information about the degree of harm generated by private cartels. Cartel injuries to purchasers are positively related to three economic factors: the size of the cartel’s market, the duration of the conspiracy, and the percentage overcharge. Antitrust sanctions should be calibrated to a cartel’s total overcharge; investigation procedures can reduce the probability of cartel formation or the duration of cartels. The U.S. Sentencing Guidelines’ are consistent with the standard optimal deterrence standard promulgated by William Landes (1983). Landes showed that to achieve optimal deterrence, the damages from an antitrust violation should be equal to the violation’s ‘‘net harm to others,’’ divided by the probability of detection and proof19 (Landes, 1983, pp. 666–668). Cohen and Scheffman (1989) argue that fines based on the U.S. Sentencing Guidelines, when coupled with civil and marketplace sanctions will cause ‘‘a serious overdeterrence problem’’ (p. 334). That is, they and other critics of the Guidelines believe that there is a disparity between the size of the corporate fines mandated for antitrust violations and the amount of the economic injuries caused by overt price fixing.20 Specifically, Cohen and Scheffman argue that actual overcharges are well below the 10% level assumed in the Guidelines (pp. 343–347).21 During recent years, their criticism has been repeated with perhaps even more intensity. For example, in a provocative essay that quickly drew rebuttals,22 Crandall and Winston (2003) argue that extant empirical evidence demonstrates that U.S. antitrust policy has been ineffective in deterring anticompetitive conduct. To support their view that the prosecution of overt price fixing is misdirected, they cite five empirical studies of overt collusion that find no upward effects on prices of conspiracies convicted in U.S. courts.23 In his comment on Crandall and Winston, Kwoka (2003) faults them for their ‘‘startlingly selective’’ body of evidence. He suggests that they should have included ‘‘y studies from any source with appropriate evaluation of their credibility’’ (p. 4).
65
Price-Fixing Overcharges
The majority of the overcharges generated by cartels in the past 15 years have been international in membership and global in their geographic impact (Connor, 2001b, 2003). To assess deterrence in the context of international schemes, non-U.S. monetary sanctions must be considered. Optimal deterrence requires cartel sanctions to be somewhat punitive. It is clear that for a single-product firm that participates in a cartel with a 10% overcharge for one year, there can be no punitive component solely with EU fines.24 EU and Canadian fines together are usually less than those imposed by U.S. courts for the same violations, and penalties in other parts of the world are practically zero. In general, global monetary sanctions have amounted to less than 10% of estimated global overcharges (Connor, 2003). Thus, punitive sanctions are the exception not the rule for illegal international price fixing. In sum, there does indeed seem to be a broad consensus among legal and economic writers that the question of the optimality of price-fixing penalties turns mightily on the actual degree of harm caused by cartel conduct, and that not enough is known about this issue. Moreover, even if the creators of the USSC Guidelines were correct that in the 1980s cartels generally raised prices by 10%, the harsher cartel sanctions imposed more recently could mean that this presumption is no longer justified. A secondary objective of this paper is to provide a firm factual foundation for dialogs on optimal deterrence and anticartel policies.
OBJECTIVE The principal purpose of this paper is to collect and analyze all serious quantitative estimates of the monopoly overcharges generated by private, hard-core cartels from all areas and eras. Estimates are taken from published social-science studies and from the decisions of competent judicial bodies. There are two secondary objectives. First, rather than apply a subjective quality filter during the collection phase, the assembled estimates are examined for patterns that might indicate systematic differences in reliability across types of sources. Second, the results of the survey are used to draw lessons about the ability of current antitrust policies to deter cartels.
LITERATURE SURVEY This paper was prepared by checking more than 500 social-science publications.25 The major portion of the overcharge estimates included in this
66
JOHN M. CONNOR
paper is taken from books, book chapters, conference proceedings, or papers published in economic, historical, and legal journals whose readers and contributors are mainly academics. The great majority of these publications are peer reviewed. A minority of the estimates are taken indirectly from newspapers, magazines, and similar journalistic outlets; from reports issued by governments; from academic working papers; and from decisions rendered by courts or antitrust commissions. This section focuses on the evolution of social-science concepts about cartels and their price effects.
Early Cartel Studies The social-science literature on cartels prior to 1945 is characterized by a groping toward a conceptual understanding of the nature of private cartels and the first tentative steps toward quantitative evaluation of the market effects of overt collusion (see the Appendix Table A1). Economic studies of cartels were dominated through the 1920s by books written in German. Especially influential was the German economist Liefmann (1897, 1932). His concept of a cartel as a voluntary, contractual association of independent firms intent on profit maximization26 and monopolistic control of a market became the accepted definition by the late 1930s. Acceptance of the idea was assisted by the simultaneous development of oligopoly theories, which freed economists from their dependence on only two market models: pure monopoly and pure competition. An unfortunate legacy of the German school of cartel studies was its view that gauging price effects was nearly impossible, a presumption that discouraged European economists from attempting to estimate overcharges until the late 20th century.27 However, U.S. social scientists inherited a more pragmatic tradition driven by an awareness of the country’s new antitrust law. As a result, most quantitative estimates of overcharges made prior to 1945 were produced by American social scientists. Some highlights include Jenks’ (1888) path-breaking analysis of the Midwest salt cartel; Jones’ (1914) book on the anthracite coal industry; Edgerton’s (1897) paper on the U.S. Wire Nail Association is a superb analysis of price effects of a shortlived but highly effective international cartel; Andrews (1889) sketched what is quite possibly the world’s first global cartel, the Secre´tan copper syndicate of 1887–1889; Stevens’ (1912a, 1912b) study of the convicted gunpowder trust is notable for focusing on what was believed to be the longest-running discovered cartel in the Nation’s history.
Price-Fixing Overcharges
67
Post-World War II Cartel Studies During and immediately after World War II, a surge in publications examined the roles of cartels in international trade and in war production. Ervin Hexner (1946), a Czech refugee turned scholar at a U.S. university, produced the most comprehensive economic study of international cartels yet published.28 Hexner had an insider’s knowledge of cartels (Barjot, 1994, p. 65). Louis Marlio (1947), a French economist who wrote a detailed account of the international aluminum cartel, had a similar background (ibid., p. 66). Both of these authors found much to admire in the effects of international cartels, whereas post-war works by American authors tended to be distinctly more skeptical, if not hostile concerning the economic and political effects of the interwar cartels (e.g., Berge, 1944; Edwards, 1946). Although they may overstate the issue, Harding and Joshua (2003) draw a sharp distinction between the views toward cartels of North American lawyers and lawmakers and those in Europe: y the North American approach has been, since the end of the nineteenth century, one of categorical censure [and] recourse to criminalization of antitrust violations as a central plank of legal control y On the other hand, the general European approach y has been altogether more tentative, more agnostic y and only in recent years moving towards an uncompromising condemnation of cartel activity y (p. 40).
One finds these disparate but changing views reflected in the social-science literature on cartels. Perhaps, the first publications to attempt to quantify systematically the price effects of cartels were a pair of books produced by a team of economists that had access to information handed over to investigators of Congressional committees and to criminal court proceedings (Stocking & Watkins, 1946, 1948).29 These books set a new standard for rigor and detail in the economics literature on cartels, because they were the first to apply rigorous modern concepts of the emerging field of industrial economics; because of access to the information spawned by numerous Congressional investigations; and because they were among the first to focus on the market effects of international cartels.30 Numerous and continuing citations to their books by leading scholars attest to their status as classics in the field. The negative impacts of cartels during 1920–1945 began to bring about a reappraisal of the welfare impacts of cartels among Europeans just after World War II. In Germany, there was a healthy parliamentary debate over its cartel laws in 1951–1957 (Wells, 2002, pp. 165–174). Through the early 1950s, a majority of the UK’s manufacturing output was affected by cartels
68
JOHN M. CONNOR
(Symeonidis, 2002; Swann, O’Brien, Maunder, & Howe, 1974). The reconsideration of the benefits of cartels began around 1950 with a series of empirical studies by the Monopolies Commission, which investigated the structure and performance of British industries and made recommendations to the government about restrictive practices, dominant firms, and mergers. By the late 1950s, anticartel legislation had been adopted that placed the burden of proof on cartels to prove the economic benefits of their price fixing and related conduct. Germany was the prime mover behind the adoption of tough anticartel provisions in the Treaty of Rome, which solidified the antitrust tradition in the EU and its Member States. In the second half of the 20th century, relatively few books were written about the empirical economics of cartels, but there have been three brief periods of interest.31 First, there was a short-lived U.S. interest in domestic cartels when the ‘‘Great Electrical Equipment Conspiracy’’ burst onto the Nation’s consciousness in 1960–1961.32 The great electrical equipment conspiracy resulted in the release of more publications in a few years than any other single historical event since the beginning of cartel literature. The scope of the conspiracy, the fame of the leading companies involved, and the U.S. Government’s aggressive prosecution of the violators – all these factors lead to a degree of public fascination and publicity about an antitrust action not seen since the Supreme Court decisions against the Standard Oil and American Tobacco trusts in 1911.33 Several trials provided unusually detailed pictures of the cartel’s organization. The books written about the heavy-electrical-equipment conspiracy include at least six monographs documenting the complex organizational details of these long-lasting and widespread bid-rigging conspiracies (Herling, 1962; Smith, 1963; U.S. Congress, 1965; Sultan, 1974/1975; Epstein & Newfarmer, 1980; Bane, 1973). Sultan’s books are by far the most quantitative. In addition, three journal articles were devoted to the cartels (Kuhlman, 1972; Finkelstein & Levenback, 1983; Lean, Ogur, & Rodgers, 1985). These studies have become staples in textbooks in industrial organization (e.g., Carlton & Perloff, 2004). Second, there was a brief revival of focus on international cartels after 1973 when the Organization of Petroleum Exporting Countries (OPEC) first used its power to raise crude petroleum prices.34 Many books and articles were written about the cartel, and two economic studies tried to predict OPEC’s staying power by studying previous international cartels. First, a chapter in a book by Eckbo (1976) is notable for its effort in classifying cartels according to a large number of potentially significant economic dimensions. One dimension is a binary variable that separates cartels with significant price effects from those that were ineffective in this respect.
Price-Fixing Overcharges
69
Second, the most comprehensive quantitative study of cartel price effects appears in a chapter by Griffin (1989).35 Griffin, who has several cartel studies to his credit, specifies a formal cartel model, which allows for a fringe of competitive, non-cooperating producers outside the cartel. From this theoretical model, Griffin derives a simple empirical model that explains variation in the Lerner Index36 of market power. Third, four books may be traced to high profile U.S. and EU prosecutions that began in late 1996. Three were prompted by a well-publicized 1998 criminal trial of three executives involved in the lysine cartel, the record of which provided a degree of testimonial evidence, which is unique for international cartels discovered after World War II (Lieber, 2000; Eichenwald, 2000; Connor, 2001b). Harding and Joshua (2003) provide a legal overview of mainly EU cartel enforcement. Only Connor (2001b) contains an empirical overcharge data. After 1973, most empirical analyses of cartel effects began to appear in professional academic journals. The shift away from monographs as the preferred outlets is remarkable. Of about 80 papers with useful overcharge information, 66 were published after 1973. While a few are historical narratives, these articles tend to focus on statistical tests of theoretical hypotheses or demonstrations of the superiority of a novel estimation technique. They form a small sub-set of the vast literature in economics that measures the price effects of market power because external information is needed to identify markets in which sellers overtly colluded from the much larger number of markets characterized by presumptively tacit collusion. Perhaps, the first published work that uses econometrics to estimate a cartel overcharge is Sultan’s (1974/1975) analysis of the U.S. electrical equipment conspiracy of the 1950s. Fisher (1980) and Finkelstein and Levenback (1983) show that econometric evidence was being presented by experts in U.S. civil, price-fixing trials as early as 1970. Econometric evidence on monopoly overcharges was published to critique government-enforced compulsory cartels; Kwoka (1977) is the first of many analyses of the price effects of agricultural marketing orders. However, quantitative analyses of the size of buyers’ cartels undercharges are rare; Daggett and Freedman (1985) seem to be the first to publish such a study. Sophisticated econometric modeling has spread into historical studies of cartels: a notable pair of studies by Hausman (1980, 1984) examines two UK coal markets from 1699 to 1845 and Levenstein (1997) the century-old bromine cartel. A new development in the cartel literature was the statistical analysis of auctions and bid rigging, much of it inspired by the urge to test game-theoretic notions (Porter, 2001 surveys this literature). Howard and
70
JOHN M. CONNOR
Kaserman (1989) study collusion in public tenders for sewer construction; Froeb, Koyak, and Werden (1993) federal-government procurement of frozen fish; Brannman and Klein (1992) state road-building contracts; and Lee (1999), Porter and Zona (1999), and Pesendorfer (2000) school-milk procurement. These studies were made possible by U.S. laws that mandate public access to bids for public project tenders, a policy that is uncommon outside of North America. Novel methods continue to be applied to estimating cartel mark-ups. There is substantial work focused on understanding cartel stability from which price effects can be derived. Grossman (1996) looked at the 1851– 1913 railroad express delivery market, and several have studied the 19th century Joint Economic Committee railroad cartel (Porter, 1983; Briggs, 1992; Ellison, 1994). Bajari and Ye (2003) applied the Baysian method to a U.S. seal-coating conspiracy. Clarke and Evenett (2003) apply a trade model to importing countries to estimate price increases during the 1990s bulk vitamins cartel. In addition to journal articles, this study draws upon nine 2000–2004 working papers of economists, many of which will become journal papers. One major source is a working paper that compiles data on the price effects of about 40 of 167 private international cartels that were discovered by antitrust authorities only since 1990 (Connor, 2003). Other important sources of scores of overcharge estimates are the decisions of courts and competition-law commissions, most published since 1990.
Quantitative Estimates of Cartel Overcharges Given the importance of the topic for legal-economic discourse, there have been surprisingly few compilations of the empirical findings of cartel overcharges.37 I have been unable to find any research that has as its principal aim collecting or analyzing information on the price effects of overt collusion.38 However, I have found six works that mention a significant number of studies of mark-ups due to overt collusion. The overcharges are assembled as a prelude to scholarly research, not as an end in themselves; none claims to be a comprehensive survey. The most comprehensive quantitative study of cartel price effects appears in a chapter by Griffin (1989).39 He derives a simple behavioral empirical model that predicts the Lerner Index of market power using three factors. The model was fitted to data on 54 cartel episodes, most of which
Price-Fixing Overcharges
71
operated during the interwar period and all of which operated in a legal environment.40 All but four of the cartel episodes were effective at raising price.41 If the but-for price is the purely competitive price, then the Lerner Index is the same as the overcharge, except that it is measured by dividing by the monopoly price instead of the competitive or benchmark price.42 That is, the Lerner Index is a margin on the collusive selling price, while the overcharge is a mark-up on the competitive price. Thus, for the same cartel, the Lerner Index is a smaller number than the overcharge.43 If, on the other hand, the but-for price is supracompetitive, then the Lerner Index might overstate the overcharge. Griffin (1989, Table 1) concludes that the mean cartel margin was 0.31, which is equivalent to a 45% price increase. Cohen and Scheffman (1989) recognize that the average size of pricefixing overcharges generated by overt collusion is a critical issue in evaluating cartel fines. Their paper cites five to seven estimates for price-fixing cases.44 A working paper by Werden (2003) cites 14 studies of cartel overcharges. All of his sampled studies were published since 1991 because he wished to study conspiracies that operated after 1974, the first year in which cartels could be prosecuted as felonies; three studies examined international cartels prosecuted by the DOJ in 1996–1997. Posner’s (1975, 2001) treatise on antitrust law is an avowedly economic treatment of the subject. To illustrate the social costs of cartelization, Posner assembles data on 12 ‘‘cartel price increases’’ in ‘‘y industries having well-organized (mainly international) private cartels’’ (Posner, 2001, p. 303), which he admits are ‘‘crude and probably exaggerated’’ (ibid., p. 304). Given that Posner is an avatar of the Chicago School of Economics, it is noteworthy that his estimates are among the highest of the six studies.45 Levenstein and Suslow (2006) focus on the determinants of success for both the interwar and more modern cartels. Although the authors are modest about their accomplishment,46 this paper contains the fullest accounting of overcharges of any source. This paper provides a total of 21 estimates of price effects for international cartel episodes. The OECD (2002–2003) report on private ‘‘hard-core’’ cartels contains a summary of a 2001–2002 survey of its government members on the economic harm caused by cartels recently prosecuted by the EC and national antitrust authorities.47 While not all of the survey responses can be converted to overcharge percentages, the usable responses represent an unusually authoritative compilation of data on mark-ups by contemporary cartels that have been prosecuted by courts or commissions.48 The six surveys just discussed are summarized in Table 1.
72
JOHN M. CONNOR
Table 1.
Summary of Six Economic Surveys of Cartel Overcharges.
Reference
Number of Cartels
Cohen and Scheffman (1989) Werden (2003) Posner (2001) Levenstein and Suslow (2006) Griffin (1989), private cartels OECD (2002–2003), excluding peaks Total, simple average Total, weighted average
5–7 13 12 22 38 12 102–104 102–104
Average Overcharge (%) Mean
Median
7.7–10.8 21 49 43 46 15.75
7.8–14.0 18 38 44.5 44 12.75
30.7 36.7
28.1 34.6
DATA SOURCES AND COLLECTION METHODS Selection Criteria I have made every attempt to identify and collect all useful information on private, hard-core cartel overcharges available from public sources. A private, hard-core cartel is one that by contemporary U.S. standards could be criminally indicted under the Sherman Act.49 Hard-core or ‘‘naked’’ cartels are those that made explicit agreements to control prices or limit quantities to be produced or sold. Price agreements may cover list prices or transaction prices; the transactions prices may be floor prices, target prices, or, if a common sales agency is employed, actual transactions prices. Prices may refer to sales of goods or services, procurement of inputs, or bids in auctions or tenders. Quantity restrictions most commonly involve fixed market shares for each participant, but may also include territorial exclusivity, customer allocations, or production-capacity agreements. Cartels that focused exclusively on advertising, patent pooling, setting technical standards, R&D, and the like are excluded. Classifying the sampled cartels at times requires judgment. Some cartels operated prior to 1890 when passage of the Sherman Act made participation by U.S. companies illegal, but many cartels headquartered in Europe predate the beginning of effective European anticartel laws. If these cartels were not formed by means of a legally enforced government monopoly, they are generally considered private schemes.50 However, if a government simply required registration or chartering of a cartel but left its management in corporate hands, they are included in the data set. Beginning in 1918 in the
Price-Fixing Overcharges
73
United States and in most European countries in the interwar period, domestic producers were permitted to register and operate export cartels with no or minimal supervision; I consider these private cartels. Similarly, if a government-owned national monopoly or commodity association voluntarily joins an international cartel, that too may be a private cartel. Thus, the mere fact that governments tolerated or turned a blind eye to cartels does not disqualify them from inclusion in the data set. However, commodity agreements known to have been initiated, actively sponsored, or overtly protected by national sovereignty are not included.51 In these ‘‘public’’ cartels, the active involvement of governments are signaled by the signing of a treaty, government ownership of stocks, or the appointment of civil servants to cartel-management positions. There are many fine studies of such agreements, but the inclusion of government-sponsored or -enforced cartels would tend to bias upward the overcharges in the sample (Suslow, 2001). Where judgment was required, procedures were followed that would result in conservative overcharge statistics. With very few exceptions, this paper reports on every scholarly or serious study that contained quantitative information on the price effects of hardcore private cartels.52 Writings by economists, political scientists, economic historians, and legal scholars are included. Written decisions or detailed reports of decisions of antitrust authorities everywhere in the world were examined. While no time limit was placed on the literature search, the large majority of the sources consulted were written after 1945.53 I have examined more than 500 English-language books, journal articles, working papers, reports, and other short analyses of cartel price effects. Many were written primarily as historical case studies or are focused on demonstrating a new method. Some mention price effects only in passing. The great majority of the short cartel studies were written by economists. Nearly all economic articles are written by North American academics using cartel episodes that affected commerce in the United States or Canada.54 The absence of empirical studies by European or Asian academics is striking.55 In general, I aimed at collecting the largest possible body of quantitative estimates of monopoly overcharges, and avoided applying some sort of quality screening. In the vast majority of cases, the writers themselves provided the overcharge calculations. In a small minority of cases, I made inferences from price data contained in the works; the bases for my inferences are briefly outlined in Connor (2004b, Appendix Table 2).56 Few overcharge claims appearing in newspapers, magazines, and newsletters are included because such assertions are usually from anonymous sources who
74
JOHN M. CONNOR
Table 2.
Number of Cartel Markets, by Type.
Type of Cartelized Market
Number
Percent
International membership Members from one nation Bid-rigging schemes Classic cartels Cartel found guilty or liable Currently under investigation (presumed ‘‘illegal’’) Known to have been operating legally No record of sanctions (presumed ‘‘legal’’)
102 177 83 196 185 7 57 30
36.6 63.4 29.8 70.3 66.3 2.5 20.4 10.8
Total
279
100.0
Source: Appendix Table 1 dated August 2005.
may not be disinterested parties in an ongoing law suit or in some public policy debate, roles that may color their assertions.57 In some cases, overcharge estimates may originate from articles in industry trade journals, but if they were cited by economists, historians, or legal scholars with some background in cartel studies, such estimates are reported in the present survey. We did include estimates appearing in a few book-length cartel studies by journalists, public servants, or other professional writers of nonfiction. Clearly, this catholic approach to data-gathering will create concerns in the minds of many readers about the reliability and precision of the overcharges. There may be substantial variation in the quality of the price data, the methods used, degrees of judicial scrutiny, and the professional orientation of the sources that could affect reliability as perceived by any individual. I noted above the lack of clarity among professional writers about the essential characteristics of the cartels until at least the 1920s. Consequently, some readers may wish to dismiss scholarship before that decade, while others will be untroubled by semantic differences. Economists may well give greater preference to writings by professionals in their own field than to opinions reached by judges, commissions, or juries, whereas legal scholars will often give greater credence to the latter. Legal professionals may have strong preferences for high court decisions over state or district courts, or they may have strong opinions about European versus American antitrust jurisprudence. Similarly, many economists might trust results published in refereed scientific journals more than other publication outlets that receive less peer scrutiny, prefer modern quantitative methods to deep historical case studies, or express skepticism about the analyses of economists writing before the Age of Game Theory.
75
Price-Fixing Overcharges
To contend with the disparate preferences of readers, I have chosen to cast my net widely, but look across the sources for evidence of systematic bias. Indeed, the analysis of these data by source, time period, or method may provide useful insights in itself. I hope to provide the interested reader with enough information to make up his or her own mind about reliability.58
Social-Science Studies The first of two major sources consists of books, monographs, reports, and refereed journal articles written by specialists in many fields: economists, historians, political scientists, lawyers, and in a few instances journalists.59 Newer publications were located by using various bibliographic search engines, by noting the references cited by authors in the works themselves, and by searching on-line library catalogs. These studies vary substantially in terms of depth and the degree of professional commitment to the study of cartels. Some economists and historians have spent substantial portions of their careers specialized in cartel analysis, but most of the publications quoted herein are by social scientists for whom cartels were just a passing interest. Other sources of information include the Web pages of scores of antitrust agencies, lists of court and commission decisions, and multilateral organizations. There are varying methods used to derive the effects of cartels on prices. In economics, older studies tended to use a rather informal method of price analysis that now comes under the rubric of the ‘‘before-and-after method’’ (Connor, 2004a). That is, armed with knowledge of when overt collusion occurred, the author would compare prices during the affected period with prices before the cartel began or after it ended; in some cases, the basis of comparison would be a price war that erupted during the affected period. The base price was typically assumed to be the long-run competitive equilibrium benchmark price (now rather succinctly, if inelegantly, termed the ‘‘but-for price’’). Although some were careful to take such factors into account, in many cases, the possibility that shifts in demand or supply conditions could have caused the benchmark price during the affected period to depart systematically from the before or after price was ignored; moreover, the idea that price wars could generate unsustainably low prices was not often recognized. Some economists of the time realized the importance of averaging before or after prices for periods long enough to eliminate the influence of transitory disturbances in markets, but others were satisfied to identify one month’s or one day’s price as the but-for price.
76
JOHN M. CONNOR
A second way of calculating a benchmark price is the yardstick method. In this type of analysis, an economist would collect prices for analogous markets that were believed to be free from cartelization. For a localized conspiracy, the competitive yardstick could be prices in a nearby city or an adjacent state with similar demand or cost conditions; the trend in cartel prices could then be compared with the trend in the yardstick during the collusive period. Yardstick price movements can also be constructed for a noncartelized product made in the same region that is made with the same inputs, utilizes a similar technology, and is consumed by the same customers.60 If a cartel colludes against only some of its customers, then the discounts offered to other similarly situated customers could yield a yardstick. Third, sometimes the costs of production and the margins earned by firms in the relevant lines of business may provide collateral indicators of variations in the degree of competitiveness of a firm or market. Cost-based estimates are relatively uncommon because detailed internal business records are needed. Both the before-and-after and yardstick methods require expert judgments about the market in question, but both remain the leading methods used in courts of law or commission hearings to determine the fact of injury or the amount of damages. Fourth, since the 1970s, the rigor and precision displayed in deriving estimates of cartel overcharges have made several advances (Baker & Rubinfeld, 1999). Driven by developments in oligopoly theory and the increasing availability of detailed company and market data, increasingly it is econometric models of the alleged collusive market that are specified and fitted to the available data.61 Game theory has influenced contemporary concepts of collusion, the design of competition policies, and empirical modeling of oligopolies (Werden, 2004). One type of econometric modeling is an elaboration of the before-and-after method. A structural model of the market before or after the conspiracy can be estimated and used to predict the competitive price during the conspiracy (Brander & Ross, 2005, pp. 17–20). A second type of econometric model can specify demand, supply, and an oligopoly model (usually Cournot or Bertrand) and fit the model to data from the collusive period (ibid., pp. 21–23). The most common approach is a reduced-form model. These models usually specify the demand and supply conditions in the relevant market as a function of the observed market price before, during, and after a conspiracy; the analyst then investigates through statistical tests whether and to what extent changes in prices or output fail to respond to normal, competitive market forces (ibid., pp. 23–29).62 Because these models can simultaneously
Price-Fixing Overcharges
77
incorporate multitudinous factors that explain prices, economists tend to regard overcharge estimates from such models as more accurate than analyses that depend on more informal ways of accounting for such factors.63 Consistent with most previous empirical studies of cartels, each cartel episode is treated as a unique observation. Most cartels are organized and fall apart only once; not counting brief disciplinary price wars, this describes one episode. However, many cartels are formed, disband, reform, and disband several times; each cycle is an episode. The reasons for analyzing episodes rather than one cartelized market over time are fairly straightforward. Each time a new collusive episode begins, chances are that the methods and membership composition have changed; pauses between episodes are often quite lengthy. Because the agreement and the players are different, in effect a new cartel is launched.
Decisions of Antitrust Authorities Decisions of courts, commissions, and other antitrust authorities are the second major source of overcharges. In theory, one should be able to determine how high cartels raise prices by a straightforward examination of a statistically significant sample of the thousands of U.S. antitrust cases that involved cartels. However, the amount that prices changed, or even whether prices were affected at all, is not relevant to the issue of whether a cartel violated U.S. antitrust law.64 In U.S. criminal antitrust cases, it is unnecessary for prosecutors to present evidence of the extent of any overcharges or undercharges.65 In civil cases, however, the damages awarded to a successful plaintiff are equal to three times the overcharges, so in these cases, plaintiffs must demonstrate how much prices increased or decreased due to the actions of the cartel. Finding overcharge rates in civil actions proved to be extremely difficult because almost every private antitrust suit for damages settles or is dismissed before an overcharge can be calculated by a neutral observer and made part of the public record of the case. As a consequence, final verdicts involving cartels, where a judge or jury calculated an overcharge, are surprisingly rare. Announced settlements are poor guides to actual overcharges66 (Connor & Lande, 2005). Data collection aimed at obtaining the largest possible sample of verdicts in collusion cases, namely, final decisions in United States antitrust cases involving horizontal collusion, broadly defined to include bid rigging and
78
JOHN M. CONNOR
related practices, where a judge, jury, or commission calculated the damages.67 Three sources were explored: computer-assisted searches of data bases,68 reading through a large number of articles and treatises on cartels and on antitrust damages, and messages to groups of knowledgeable antitrust professionals.69 Every qualifying final collusion verdict is included.70 The small sample size (26) of overcharges from final U.S. decisions is disappointing. Besides U.S. court decisions, the web sites of many foreign antitrust authorities were examined.71 In the jurisdictions employing Common Law, most cartels are sanctioned after government negotiations that result in guilty pleas or by monetary settlements with private parties out of court. When this is the method of resolution, the press releases practically never mention the degree of harm caused by the cartel. Very few cartels defend themselves in court, and very few of the trials result in published decisions that reveal the overcharges. In other legal systems, antitrust commissions hold confidential hearings to determine guilt and impose sanctions. These decisions are announced in press releases that seldom mention the extent of cartel damages.72 However, in some jurisdictions, a detailed report is released a year or two after the decision, and some of these reports have prices that can yield useful overcharge information.73 Additionally, commission decisions can be appealed to a court that renders a decision with a recitation of the facts of the case.74 The data are organized according to three levels of analysis: markets, episodes, and overcharge estimates. By ‘‘market’’ is meant the industry or product that was subject to price fixing. Markets are precisely self-identified by the participants in the conspiracy, though occasionally there are alternative names for the same market.75 The name of the market is eponymous for the cartel. Episodes, discussed more fully below, are distinct periods of collusion separated by price wars, temporary lapses in agreements, or changes in cartel membership or methods. Episodes may be adjacent in time or may be separated by significant gaps of time.76 The markets marked by adjacent multiple episodes will typically be regarded by antitrust law as one infraction, but as economic phenomena as multiple cartels. Most of the analyses in this paper will use overcharges as the units of observation. Each episode will in principle have one true ‘‘average’’ (episode-long) overcharge and one ‘‘peak’’ overcharge.77 However, because there are sometimes multiple publications about the same episode and because a single analyst will sometimes apply alternative methods of estimation, this paper often records several estimates for a single episode.
79
Price-Fixing Overcharges
GENERAL DESCRIPTION OF THE SAMPLE Markets Publications from economists, historians, and related sources yielded useful overcharge or undercharge information on cartels that operated in 279 markets (Table 2). If one group of sellers decided to fix prices of a product in one geographical region and a different group colluded on the same product in a separate geographical region, these will be viewed as two markets. Of the 279 markets, 37% were cartelized by international agreements, where ‘‘international’’ describes the membership composition of the cartel and not necessarily the geographic spread of the cartel’s effects. Some international cartels affected directly the commerce of only one nation, though the vast majority was international in both senses. National cartels account for the remaining 63% of the cartelized markets.78 In this category, I count some purely national cartels that were formed for the sole purpose of controlling a nation’s export sales; in the United States, these are called Webb–Pomerene Associations. In addition, some domestic cartels had side agreements with international cartels that protected their domestic market from exports from the international cartel’s members. Almost 30% of the sample consists of markets affected by bid-rigging cartels.79 Although most cartels have some sales to government entities or industrial customers that purchase by tenders, these cartels are explicitly described to have been principally or exclusively engaged in bid rigging. This proportion is certainly an underestimate because the sources did not always provide enough detail on the cartels to be certain of the degree of bid rigging. Recall that the U.S. Sentencing Guidelines assume that bid rigging leads to higher overcharges than otherwise identical conspiracies. The remaining 70% of the cartelized markets may be called ‘‘classic’’ cartels, those that set market selling prices and/or market quotas for each of its members.80 Two-thirds of the cartels were found to be in violation of antitrust laws by at least one legal body.81 Sometimes these are called ‘‘discovered’’ cartels. The determination of guilt or liability may take the form of guilty pleas (or nolo contendere in U.S. courts up until the 1960s); of a decision at trial by judge or jury; of a commission decision to impose fines, consent decrees, or other sanctions; of the payments of civil penalties; or of negotiated settlements by defendants in a suit. The remaining 31% of the cartelized markets are known or believed to be ‘‘legal,’’ because they operated prior to the enactment of antitrust laws in the jurisdictions in which they functioned, or
80
JOHN M. CONNOR
extra-legal, because they were not known to have been punished by an antitrust authority. Other legal cartels were organized and registered under antitrust exemptions, such as export cartels or ocean shipping conferences.
Episodes Although I have collected data on 279 cartelized markets, there are multiple overcharge estimates for a large minority of the markets. There are more estimates than cartelized markets for three reasons. First, about half of the markets experienced multiple phases or ‘‘episodes’’ for which the price effects differed. The sources have distinguished a total of at least 512 episodes. This term, which might better be called an observational time period, though commonly used in cartels studies, requires some additional explanation. If a cartel had more than one episode, then each episode is marked by changes in membership composition, the terms of the collusive agreement, method of management, geographic focus, or other major change. In other words, when a cartel is re-formed, it enters a new phase. Between episodes, pricing discipline often breaks down; in some of the cartels, the interregnum is a period of contract renegotiation. The aluminum market, for example, went through six distinct phases that sometimes were adjacent in time and sometimes were several years apart. This heavily researched cartel has 28 overcharge observations. One study from which I obtained a dozen observations summarized the results of 109 bid-rigging convictions in the fluid milk markets of the Southeastern United States within a few years (Lanzillotti, 1996). I count each conviction as an episode.82 If one prefers to count the Lanzillotti summary and two other ‘‘group studies’’ as three episodes, then the total becomes 332. However, some studies that I count as one episode incorporate multiple temporal phases (e.g., Ellison’s study of the Joint Executive Committee). Thus, there are reasons to believe that the number of episodes is an undercount. Second, for a few episodes, more than one study has been published. For example, for the various aluminum cartels, I drew on nine studies written by eight authors. Third, for a given episode, alternative methods of estimation are sometimes available, in a few instances by the same author writing in the same publication. In general, the distribution of episodes across types of cartels is quite similar to the distribution of cartelized markets (Connor, 2004b, Table 3). The major difference is that international cartels tended to have a larger
81
Price-Fixing Overcharges
number of multiple episodes than did domestic ones. The 88 international markets in the sample that were cartelized had on average 1.6 episodes, whereas national cartels had only 1.3 episodes on average. As a result, a larger share (44%) of the cartel episodes had international membership. The number of episodes per market does not vary significantly across other type of categories.
Overcharges Two kinds of cartel mark-up data are available. First, researchers usually report the average price increases over a representative portion or a whole episode (Table 3). This is the measure most relevant for forensic purposes and is the one that will be the focus of most analyses in this paper. I have collected 770 average overcharge estimates. In some cases, the averages are carefully weighted by the sales in each year or month of the episode; but in most cases, the authors give equal weights to the price changes in each sub-period during the total affected period. Sometimes, it is not clear from the source whether the averages are weighted or unweighted; if the conspiracy period is marked by steady slow market growth, it matters little which is reported. Some of the overcharge estimates are said to be minimum estimates. To be conservative, all such minimum estimates are counted as averages.83 Finally, some analysts give minimum and maximum estimates. The majority (57%) of the average overcharge estimates are drawn from international cartel episodes. More than two-thirds of the estimates come
Table 3.
Number of Average Overcharge Observations, by Type of Cartel.
Type
Number
Percent
International membership National members only Bid-rigging schemes Classic price-fixing cartels Cartels found guilty or liablea No record of sanctions (‘‘legal’’)
441 329 159 611 529 241
57.3 42.7 20.7 79.4 68.7 31.3
Total
770
100.0
Source: Appendix Table 1 dated August 2005. a Included are seven cartels still being investigated by antitrust authorities in 2005.
82
JOHN M. CONNOR
from cartels that were sanctioned and nearly four-fifths from ‘‘classic’’ pricefixing schemes. Second, 270, one-fourth of all the 1,040 overcharge figures that were assembled, are peak price effects. One-third of the episodes have peak estimates. In some cases, the peak price was reached for only one day during a cartel period of several years; in other cases, the peak may be the highest one of several years. Peak price changes indicate the potential for maximum harm when a cartel is at its most disciplined. Classifying a particular estimate as an average or peak figure in a minority of cases required judgment. If the original source is unclear about which type of estimate is being presented, in order to be conservative I have assumed it is a peak estimate.
SUMMARY OF THE RESULTS Number of Overcharge Observations There is a total of 1,040 quantitative estimates of overcharges and undercharges drawn from 259 publications.84 The 770 average overcharge estimates are shown in Fig. 1 arrayed by the cartel episode’s end year. The six periods distinguished in this and subsequent tables were selected to represent different antitrust regimes in the United States and abroad.85 In addition, the periods correspond roughly to the major changes in the 300 250 200 150 100 50 0 17701890
Fig. 1.
18911919
19201945
19461973
19741990
19912005
Number of ‘‘Average’’ Overcharges Collected. (Total of 770 Observations/ Spreadsheet Dated August 1, 2005.)
Price-Fixing Overcharges
83
relationship of antitrust jurisprudence to economics (Kovacic & Shapiro, 2000). The era up to 1890 is an obvious choice because of the enactment of the Sherman Act in the United States and the 1889 Anti-Combines Act in Canada.86 During the early decades of the 20th century, numerous U.S. court decisions made the scope and power of the U.S. anticartel law apparent to lawyers, enforcement officials, and business persons (Wells, 2002).87 The year 1919 is chosen as a break point because it represents the end of a period of U.S. antitrust activism and because of World War I. During 1914–1919, nearly all international cartels, a few of them with U.S. corporate members, had ceased operating. Many of the pre-war cartels were re-established after 1919, but in the majority of instances without the active participation of U.S. firms. The years 1945–1946 are another logical break point. During 1939–1945, nearly all of the interwar international cartels were disbanded; moreover, wartime price controls and cost-plus government contracts made cartels superfluous. Scores of U.S. criminal prosecutions of international cartels during 1940–1945 clarified the illegality for U.S. firms of many more subtle forms of cartel participation, such as patent pools, cross-licensing of technologies, and the creation of overseas subsidiaries as loci for cartel participation. The post-World War II era is characterized by the emergence of industrial-organization as a separate discipline within economics, of rapid advances in empirical methods of analysis, and of the adoption of effective anticartel laws outside of North America. Kovacic and Shapiro (2000) note that in the United States by the 1940s ‘‘y there was considerable consistency between judicial decisions and economic thinking y’’ (pp. 51–52). Moreover, the vast expansion of higher education in North America and Europe brought about a parallel expansion of the economics profession as a whole and, consequently, an acceleration in the total resources devoted to theoretical modeling (particularly after 1980) and related empirical testing on collusion.88 Beginning in the 1960s, economists in North America began to work more closely with prosecutors and the private bar in antitrust cases, and many of them began to analyze and write about those activities. This is a major factor responsible for the fact that nearly 80% of the estimates of ‘‘national’’ cartels (most of them prosecuted in North America) are drawn from the post-1945 time period. The post-war era is divided into three sub-periods. The transition years 1945–1973 correspond with three relevant changes in anticartel enforcement. First, the antitrust idea became firmly implanted in the laws of countries outside North America for the first time: Germany and Japan in 1947, the United Kingdom in 1956, and the European Economic Communities
84
JOHN M. CONNOR
(EEC) in 1958. Second, the European Commission (EC), the administrative arm of the EEC, after a decade of registering cartels, successfully prosecuted its first cartel in 1969. Third, U.S. price-fixing enforcement penalties became significantly more severe in 1974. Class action suits became far more common by the mid 1970s because of changes in federal court rules, a change that permitted plaintiffs to attract better lawyers and economic expertise (White, 1988, Table 1.1). Another milestone in U.S. anticartel legislation was the 1974 law that made price fixing a felony, thereby lengthening maximum individual prison sentences and strengthening the bargaining power of the DOJ. Although the prosecution of price fixing of relatively inconsequential domestic conspiracies was at a high level in 1974–1990, the DOJ did not give a high priority to investigating international cartels, nor did it have any success in the courtroom in the few international cases it did pursue (Connor, 2001a). Kovacic and Shapiro (2000) identify 1973–1991 as the years during which the Chicago School of Economics had its greatest influence on antitrust law and enforcement. By 1990, all the present criminal sanctions available to the U.S. government were in place. In 1990, penalties for corporations rose from $1 million to $10 million.89 Moreover, in the early 1990s, the DOJ had in place three devices that improved detection and prosecution of cartels: the U.S. Sentencing Guidelines for corporations (1989), the automatic amnesty policy for corporate whistle-blowers meeting certain criteria (1993), and a demonstrated ability since 1994 to impose fines above the $10-million statutory cap by means of an alternative sentencing provision. These devices were in some cases adopted by the EU and other antitrust authorities, which significantly improved the investigation and prosecution of international cartels. Both U.S. and EU prosecutions of international cartels increased markedly. Except for a hiatus in 1946–1973, the number of observations per year has grown over time (Fig. 2). The first cartel for which price effects can be found is the Coal Gild of northeastern England (also known as the ‘‘Newcastle Vend’’), which made its first collusive agreement on London coal prices in 1699. Although highly unstable, the Vend finally collapsed in 1845, making it the most durable cartel in the data set.90 The primary factor that explains the upward trend in the number of overcharges is the growth in the number of international cartels with usable data (Fig. 3).91 Up until 1890 when price fixing was legal everywhere in the world, only one estimate is available about every six months on average. During this early period, the vast majority of price effects are reported for domestic cartels operating in the United States, the United Kingdom, and
85
Price-Fixing Overcharges 20
15
10
5
0 17701890
Fig. 2.
18911919
19201945
19461973
19741990
19912005
Number of ‘‘Average’’ Overcharges per Year. (Total of 770 Observations by End Date/Spreadsheet Dated August 1, 2005.)
120
Percent
90 60 30 0 17701890
18911919
19201946
19461973
19741990
19912005
Fig. 3. Proportion of Cartels that are International. (Total of 770 Overcharge Observations/Spreadsheet Dated August 1, 2005.)
Germany. Although there were large numbers of domestic cartels extant in the late 19th century, the small size of the fledgling economics profession, a literary approach to writing in economics, and inevitable destruction of most business records over time contributed to the fewness of quantitative overcharge observations for 19th-century cartels. From 1891 to 1945, most data are drawn from studies of international cartels. Four to six overcharge estimates are available per year during these
86
JOHN M. CONNOR
periods. The proportion of international schemes is especially high during the interwar period and after 1990 and especially low during 1946–1990. It is likely that there were more domestic cartels operating legally in Europe in the early 20th century than there were international cartels, but the latter were given more publicity because they appeared to be novel forms of business organization.92 The increasing awareness of the illegality of price fixing in the United States may also account for the absence of internal records of domestic cartels in the United States after 1890. Moreover, because the penalties were so low (a maximum of $5,000 per count), relatively few court decisions bothered to give details about sales or prices during the conspiracy. During 1891–1919, there are 3.8 price observations per year; the rate rises to 5.6 per year in the interwar period. More data are available for international cartels during 1891–1945 than for cartels composed of companies from a single nation. About two-thirds of the observations are drawn from international cartels. One reason is that international cartels mostly were based in Europe, where they operated with legal impunity. That is, they had freedom to set prices. In Weimar Germany for a few years after 1923, cartels were regulated. In a few European countries, cartels were required to register with the government. In others, cartel contracts were enforceable in the courts. Many of the interwar international cartels were organized as federations of national cartels and were aimed primarily at creating national monopolies and assigning shares for export sales.93 As nearly all of them were believed by their members to be legal at the time, their activities often were openly reported by the business press.94 Members of these cartels did not attempt to hide their activities; indeed they often publicized their operations, particularly if they achieved putatively efficiency-enhancing industry rationalization, protected national markets, increased national employment during stressful economic times, or promoted price stability. During this period, many countries passed legislation specifically authorizing cartels that controlled national exports, even if that meant agreements on prices in various overseas markets. In a few cases, including the United States, these cartels were used as cover organizations for domestic price fixing. In the early and mid 1940s, many of the interwar cartels were investigated by the U.S. Congress, indicted by the DOJ, and sued by private parties. Combined with the expanding size of the economics profession and the growing interest among economists in imperfect competition, the transparency of non-U.S. cartels led to a large number of empirical cartel studies. For 50 years after the end of World War II, the number of known international
Price-Fixing Overcharges
87
cartels declined markedly. Perhaps because of the aggressive prosecution of cartels by the DOJ in the early 1940s, it appears that international cartels were by and large driven underground for decades after 1945. From 1946 to 1989, an average of five or six overcharge estimates could be found, nearly all of them domestic conspiracies. Few international cartels were discovered or prosecuted until the early 1990s – less than one international cartel episode every two years. Several explanations have been offered for the hiatus in international cartel formation in the decades following 1945. The destructiveness of World War II left the United States with as much as 65% of world industrial capacity in the late 1940s. As a result, manufacturers in Europe and Japan were oriented mainly toward rebuilding their domestic markets; not only were few industrial partners available for international agreements, it seems that U.S. firms were less prone to form cartels than firms from countries with no or weaker antitrust cultures. In the 1950s and accelerating in subsequent decades, U.S. firms embarked on a period of rapid foreign direct investment as the preferred means of entering overseas markets; leading European and Asian firms adopted this strategy increasingly after the late 1960s. Until the early 1980s, most United States markets were subjected to little import competition, but by the 1990s, imports were exerting a powerful influence on price competition across a wide spectrum of commodity markets. Most international cartels have arisen only in industries with internationally traded merchandise and populated by multinational corporations with strong leading positions. For all these reasons and probably several others as yet unknown, international cartel formation was seemingly at an historically low level until the 1980s. By contrast, the large number of overcharges available for the data set after 1990 is attributed to the launching of an historically high number of international cartels since the early 1980s. Most of these cartels could not have been contemplated without the direct participation or passive cooperation of leading U.S. companies that still tend to be among the leaders in most cartelized markets. Other factors that may be responsible for the surge in overcharge estimates may include greater interest in collusive phenomena by economists, shifts in antitrust enforcement priorities, expansion in the sheer number of antitrust authorities worldwide, and improved carteldetection programs. The number of overcharge observations in 1990–2005 exceeds 17 per year, which is more than double the rate of the interwar period. A second important trend is that most cartel data now arise from prosecuted cartels (Fig. 4). Prior to 1946, about one-third of the observations
88
JOHN M. CONNOR 100
Percent
70
40
10 -20 17701890
Fig. 4.
18911919
19201946
19461973
19741990
19912005
Proportion of Cartels Sanctioned. (Total of 770 Observations/Spreadsheet Dated August 1, 2005.)
refer to cartels known to have been sanctioned.95 Prior to the 1940s, U.S. anticartel sanctions were weak by today’s standards, but increasingly after 1911 or so businesspersons became aware of the legal dangers of overt collusion in the domestic market. However, until the early 1970s, national and international cartels comprised of European companies could form cartels subject only to registration requirements in most European countries (and the EEC after 1960).96 The EC began imposing fines on unregistered cartels that affected EEC trade beginning in 1969 (Harding & Joshua, 2003, p. 121). During 1974–1990, U.S. corporate sanctions on cartels became significantly harsher, and the European Union’s prosecutions moved in the same direction (Connor, 2003). Both jurisdictions imposed historically unprecedented penalties on international cartels beginning in the late 1990s. After 1990, virtually all the observed cartels in the sample were prosecuted or fined by one or more antitrust authority. This pattern does not necessarily mean that the probability of discovery by prosecuting bodies has gone up, but it probably does represent a heightened aggressiveness in anticartel enforcement as well as a shift in research methods by social scientists.97 A third factor that is associated with the increase in cartel overcharge observations is the surge in studies of bid-rigging conspiracies since 1945 (Fig. 5). Prior to the 1950s, only two cartels were identified as primarily engaged in bid-rigging conduct.98 Remarkably, in the 1945–1989 periods, almost half of all the overcharge observations in the sample were primarily bid-rigging conspiracies. Awareness of the importance of bid rigging among economists may have been triggered by the well-publicized U.S. electrical
89
Price-Fixing Overcharges 60 50
Percent
40 30 20 10 0 17701890
Fig. 5.
18911919
19201946
19461973
19741990
19912005
Proportion of Cartels that Rigged Bids. (Total of 770 Overcharge Observations/Spreadsheet Dated August 1, 2005.)
equipment conspiracies discovered around 1960. Post-war studies of bidrigging cartels focused on national cartels in the United States, most of them local milk or construction conspiracies. The immediate victims of most of these bid-rigging conspiracies were governments. Relatively few international cartels rely primarily on rigging auctions or tenders for public projects. What may seem like a surge in this practice may in fact be a consequence of changes in data availability. Most of the articles on bid rigging have drawn on public records of state or federal agencies that have been the objects of these conspiracies. It is possible that the increase in bidrigging cases seen in the data is simply due to the advent of open-records laws at the state and municipal levels similar to the federal Freedom of Information Act.
Trends in Average Overcharges Over Time Table 4 displays the medians of all average overcharges reported, distinguished by membership type, legal type, and mode of pricing conduct, and by time period. Median percentages are displayed instead of the means because nearly all the cells contain positively skewed figures. That is, a few very high overcharges in any particular category tend to overwhelm the larger number of low-to-medium percentages when calculating the more common type of average, the mean. Moreover, while there is no upper limit on overcharge estimates, they are not allowed to fall below zero. In such
90
JOHN M. CONNOR
Table 4.
Median Average Overcharges, by Year and Type. Median Percenta
Cartel Episode End Date Membership
Legal Status
Bid Rigging
All Types
National
International
Found Guilty
Legal
Primary Conduct
Other
1780–1891 1891–1919 1920–1945 1946–1973 1974–1990 1991–2004
20.0 19.0 18.3 15.0 18.0 21.0
50.8 49.5 35.8 33.6 41.5 24.6
14.0 24.4 37.7 15.0 20.0 24.6
22.6 34.0 31.3 26.3 22.5 21.3
27.5 – – 13.0 20.0 19.0
21.0 29.5 32.8 23.0 18.5 25.0
22.3 30.4 34.0 15.0 24.0 24.0
All years
18.8
31.0
23.3
28.5
18.8
25.5
25.0
Source: Appendix Table 2 dated August 2005. a Medians of the point estimates or, where appropriate, of the midpoint of a range of estimates. Includes many zero estimates. See Table 5 for the numbers of observations in each cell.
situations, the means are larger than the medians, and the median is a better representation of the central tendency. The median cartel overcharge for all types and time periods is 25.0% and for successful cartels 27.5%.99 The median overcharge for national cartels is 18.8%; whereas for international cartels, it is 31.0% (65% higher). Cartel mark-ups are above average for all types of cartels in the 1891–1945 periods, well below average during 1946–1973, and close to the all-periods average for the other three time periods. Variation over time appears to be related to changes in the mix of cartels types. For example, overcharges are relatively high when the period mix is rich in international cartels but poor in bid-rigging cartels. There is a strong downward trend in overcharges of international cartels, but weaker evidence of trends for the other types.100 In the period after 1990 when anticartel sanctions were the highest, the overcharges of discovered cartels are below the all-period averages for each type. The distinct decline in average overcharges of cartels that ended after 1990 is most evident among international cartels.101 Somewhat surprisingly, it appears that the interwar cartels, nearly all of them Eurocentric international legal agreements, attained only slightly higher than average levels of price effectiveness. Perhaps, the steadiest overcharges may be seen in the column of legal cartels where the average overcharges hover near the 30–35% range in all but the most recent period.102
Price-Fixing Overcharges
91
It is difficult to know what to make of the downward trends for some types of cartels. Besides the possible influence of the spread of effective anticartel enforcement, several alternative hypotheses may be put forward. Perhaps, the application of more sophisticated quantitative methods by researchers in recent decades systematically yield lower estimates of price effects than the earlier studies that relied on simpler before-and-after comparisons. Perhaps, expected profit rates in cartelized industries have declined as an effect of globalization, and those companies that join cartels are satisfied with smaller percentage increases from collusion. Industry mix could provide an explanation. The sample drawn from the earlier periods tends to contain more minerals and metals conspiracies, whereas the later estimates have a higher proportion of chemical, construction, and services firms represented. Because the most recent periods contain a higher proportion of cartels that were caught by antitrust authorities, the more recent estimates may be drawn from a population of cartels that is relatively incompetent in hiding their activities; similarly, the greater antitrust scrutiny in the United States from 1940 and from Europe since the 1960s could prompt cartelists to refrain from full monopoly pricing increases so as to reduce the chances of detection. Some of these hypotheses will be investigated below.
Average Overcharges Across Types There are significant differences in the height of overcharges when the sample is split according to three cartel characteristics: national or international in membership, bid-rigging or classic price-fixing conduct, and sanctioned or unsanctioned cartels history (Fig. 6). In the aggregate and for all time periods, higher mark-ups are associated with international membership, classic price-fixing methods, and no history of official sanctions. The strongest characteristic pattern that emerges is that in every historical period, international cartels have had higher overcharge rates than domestic (mostly U.S. based) cartels (Fig. 7). Up to the 1990s, international cartels were on average 133% more effective in raising prices than ‘‘national’’ cartels (cartels that fixed prices in one country and export cartels comprised of firms from single countries). This is not so surprising in the pre-World War II era because international cartels were formed without concern about prosecution, and most of the pre-war sample of national cartels operated in the United States.103 But the fact that the differences persisted in the post-war period is somewhat unexpected. Besides antitrust-enforcement considerations, the greater pricing power demonstrated by international
92
JOHN M. CONNOR
Median Percent
40
30
20
10
0 National
International
Guilty
Unsanctioned
Bid Rigging
Classic
Types of Cartel Episodes
Fig. 6.
60
Average Overcharges by Cartel Type (includes some interpretations of court and commission decisions).
National
International
Percent
50 40 30 20 10 0 1770-1890 1891-1919 1920-1945 1946-1973 1974-1990 1991-2005
Fig. 7.
Total
Median Average Overcharges over Time. (Total of 770 Overcharge Observations/Spreadsheet Dated August 1, 2005.)
agreements may reflect a greater degree of freedom from threat of entry than for geographically more localized cartels. International cartels in all eras tended to attract members that controlled the lion’s share of production in all the regions of the world with modern production facilities. Also, international cartels by their very nature deal with internationally tradable commodities, mostly homogeneous producer intermediates with relatively low long-distance transportation costs. Finally, international cartels can more easily engage in third-degree price discrimination among national markets than cartels organized within a single geographic market.
Price-Fixing Overcharges
93
In the 1990–2005 period, the superior pricing power of international schemes ebbed. The median overcharge fell to an historical low of 24.4%. In a sharp break from the first five periods, overcharges of international cartels averaged only 16% higher than national ones. The reasons for the convergence of national and international cartel mark-ups are difficult to divine.104 A second pattern noted the inferior price effects of bid-rigging cartels compared with conventional conspiracies that set selling prices or allocate market shares. In the sampled cartels classic price-fixing conduct led to 29% higher median overcharges than observed for bid-rigging methods. Bidrigging cartels often are organized to exploit tenders for government publicwork projects. Some economists have hypothesized that government buyers are less competent in detecting rigged bids than are professional industrial buyers.105 Relatively few international cartels engage primarily in bid rigging, so this conduct category may be confounded with the geographic types just discussed above. Nevertheless, this finding directly contradicts the hypothesis of Cohen and Scheffman (1989) and the U.S. Sentencing Guidelines that impose higher penalties for bid rigging. It also challenges a rationale of the U.S. Government’s antitrust policy shift in the 1980s that shifted resources toward the targeting of bid rigging against governments. Third, an examination of the columns that contrast cartels according to their legal status sheds light on sample selection bias, an important methodological issue in cartel studies. Many such studies depend on samples of convicted cartels, and critics of these studies have asserted that cartels discovered through government investigations or sued by private plaintiffs are as a group inept compared with cartels that either had no fear of sanctions or remained clandestine. ‘‘y [I]t is not known whether cartels that find themselves in court are unsuccessful or merely unlucky’’ (Carlton & Perloff, 2004: 216–217). In particular, an influential study by Asch and Seneca (1976) finds that price fixers that were caught in 1958–1967 were significantly less profitable during collusion than a control group of unprosecuted firms.106 Lower profitability ought to be reflected in relatively low overcharges. The data in Table 4 suggest a resolution of this paradoxical finding. Cartels punished in the time period covered by the Asch and Seneca study were indeed relatively inept: their median overcharges of 13% are the lowest by far of ‘‘guilty’’ cartels in any of the six time periods. Moreover, their sample appears to have been drawn disproportionately from domestic bid-rigging conspiracies, the categories that throughout history have generated the lowest overcharges. While a more precise analysis is needed, it appears that the Asch and Seneca study may itself be flawed by sample selection bias.
94
JOHN M. CONNOR
It is worth noting that there are few unsuccessful cartels in the data set. Only about 6% of the overcharges indicate that an analyst judged an episode to have produced no significant effect on market prices. I do not wish to make too much of this result, because it may represent selection bias in the studies relied upon. Injurious cartels may be inherently more interesting to analysts and the results more publishable than those about incompetent cartels.
Size Distribution of Overcharges Given the interest in the factual foundations of the U.S. Sentencing Guidelines applied to cartel sanctions, it is logical to examine the size distribution of the estimates. Fig. 8 classifies the average estimates into eight size categories. Because the Guidelines are predicated on the assumption that the average cartel has a 10% overcharge, that break point is of special interest. Because of the interest in prosecutable cartels, the discussion of Table 5 will focus on the effective cartels (non-zero overcharges). Perhaps, the most striking result is that 62% of the cartel episodes have overcharges above 20%.107 The mean overcharge of the 38% of episodes in the two lowest size ranges (0.1–19.9) is 10.3%. These are the cartels imagined to be typical by the creators of the U.S. Sentencing Guidelines. The 62% of the cartel episodes with overcharges of 20% or higher have a mean overcharge of 55.3%, more
Percent of Total
30
20
10
0
0%
0.1-9.9%
1019.9%
2039.9%
4059.9%
6010099.9% 199%
200%+
Fig. 8. Cartel Overcharges by Size Category. (Total of 770 median overcharges, average of entire episode or a representative portion of the collusive period (spreadsheet dated August 3, 2005).)
95
Price-Fixing Overcharges
Table 5. Mean Average Overcharges by Size Category. Percentage Rangea
Zero or lessb 0.1–9.9 10.0–19.9 20.0–39.9 40.0–59.9 60.0–79.9 80.0–99.9 100.0–199.9 200–886 Total
Number of Observations
50 111c 159 223 119d 49 14 28 19 770
Distribution of Observations (%) Mean
Total
Non-Zero
0 6.4 14.4 28.8 48.1 68.3 89.4 129.5 422.2
6.5 14.4 20.7 29.0 15.5 6.4 1.8 3.6 2.5
0 15.4 22.1 31.0 16.5 6.8 1.9 3.9 2.6
40.6e
100
100
Source: Appendix Table 2 dated August 2005. a Point estimates or midpoints of ranges. b Four negative numbers are converted to zero. c Four estimates of ‘‘weak cartels’’ are assumed to be 1% overcharges. d Fifteen estimates of 50% are from Eckbo (1976). e For effective cartels (those with positive overcharges), the mean is 43.4%.
than five times the level assumed by the Guidelines’ authors. If the Guidelines were truly designed to deter recidivism, even if the probability of detection is 100% five-eighths of the cartels will be under-deterred.
Peak Overcharges So far only the ‘‘average’’ overcharges have been examined – those that refer to the mean price change over all or most of an episode. Fig. 9 and Tables 8 and 9 explore the peak price effects attained by cartels – the maximum mark-ups observed for one week, one month, one quarter, or one year of an episode, depending on the price series available.108 It is well known that oligopolistic arrangements typically generate price changes that fall short of what a pure monopolist in a blockaded market would set in order to obtain maximum profits. Tacit collusion generally results in mark-ups above but closer to competitive levels than monopoly levels. While overt collusion should be somewhat more effective than tacit collusion at raising prices ceteris paribus, information failures, potential competition, and cheating also typically result in sub-monopoly price effects. Because the peak periods
96
JOHN M. CONNOR
Median Percent
60
Average
Peak
30
0 National
International
Guilty
Unsanctioned
Bid Rigging
Classic
Cartel Episode Types
Fig. 9.
‘‘Peak’’ versus ‘‘Average’’ Overcharges.
are generally too brief for significant changes in the structure of the industry to change, the observed peak overcharges are measures of the short-run market power exercised by cartels when the discipline of the members is at its most cohesive.109 Thus, the peak price effects are instructive about the potential harm that cartels can cause when they are unfettered by coordination problems. From Fig. 9, it is apparent that the peak overcharges are about double the average overcharges for all types of cartels. The range of the ratio is narrow: from 83% for national cartels to 108% for bid-rigging cartels. Table 6 shows the median peak overcharge over time and across types of cartels.110 The trends in peak effectiveness over time are all negative.111 The decline in peak overcharges of international cartels is particularly strong.112 The pattern of peak overcharges across cartel types is similar to that for the average overcharges. In all time periods, international cartels were able to reach higher levels of peak-price effectiveness than the ‘‘national’’ cartels – on average 85% higher. Peak mark-ups are only slightly higher for prosecuted cartels. And since 1945, when bid-rigging studies became plentiful, cartels that fixed prices or production levels were significantly more harmful than bid-rigging agreements. Table 7 provides calculations of how much higher peak overcharges were compared with the longer run averages for given episodes. Generally speaking, the peaks are about double the longer run average mark-ups. A high peak/average ratio is a rough indicator of price stability during a conspiracy; low ratios may be interpreted as cartels that achieved few operational problems. There are few strong trends in these ratios over time.113
97
Price-Fixing Overcharges
Table 6.
Peak Cartel Overcharges, by Year and Type.
Cartel Episode End Date
Median Percent Membership
Legal Status
Bid Rigging
National
International
Found Guilty
Legal
Primary Conduct
Other
1770–1891 1891–1919 1920–1945 1946–1973 1974–1990 1991–2003
42.4 26.2 46.0a 45.0 33.3 31.2
104.5a 81.0 68.0 53.0 72.0 53.4
46.8 41.5 60.0 48.3 35.0 50.0
51.2 71.6 66.0 33.0 39.0a 25.0a
18.5a 11.9a 69.0a 43.0 35.0 37.5
55.5 54.5 63.1 50.0 51.6 50.5
All years
33.3
60.5
48.3
55.0
35.0
54.5
Source: Appendix Table 2 dated August 2005. a Fewer than four observations.
Table 7.
Peak/Average Ratios of Cartel Overcharges, by Year and Type. Ratio of Mediansa
Cartel Episode End Date Membership
Before 1891 1891–1919 1920–1945 1946–1973 1974–1990 1991–2003 All years
Legal Status
Bid Rigging
National
International
Found Guilty
Legal
Primary Conduct
Other
2.12 1.38 2.51 3.00 1.85 1.49 1.77
2.06 1.64 1.90 1.58 1.73 2.17 1.95
3.34 1.70 1.59 3.22 1.75 2.03 2.07
2.27 2.11 2.11 1.25 – – 1.93
– – – 3.31 1.75 1.97 1.86
2.64 1.85 1.92 2.17 2.79 2.02 2.14
Source: Tables 4 and 6. –: Too few observations. a The ratio of the median peak overcharges to the median full-period overcharge.
Overcharges by Location of Cartel Law-makers and antitrust enforcement officials may be interested in the locus of decision-making by the cartels in the sample. Fig. 10 and Table 8 classify the cartels according to the location of the cartel’s headquarters or
98
JOHN M. CONNOR
Percent
40
30
20
10
0 EU-wide
Fig. 10.
Table 8.
Asia
Global
Other
North Nations of America Europe
Median Average Overcharges by Geographic Location.
Average Overcharges by Cartel Headquarters Location.
Principal Location of Cartel Members USA and Canada Single nations in Western Europe Multiple nations in Western Europe (EU) Asia and Oceania Global Africa, South America, and Eastern Europe
Number of Estimates
Average Overcharge Median Percent
Mean Percent
234 136
20.25 16.95
28.53 47.98
126
42.70
53.66
53 248 23
28.80 28.00 18.80
52.69 50.24 23.89
Source: Total of 770 observations from Appendix Table 2 dated August 2005.
the place of residence of the great majority of the cartel’s corporate members. In most cases, corporate membership mix corresponds to a cartel’s geographic field of operations.114 Cartels may be composed of member companies with headquarters in only one country or one continent; in most of these cases, the cartel is a ‘‘virtual’’ joint venture with no permanent address. By contrast, many early 20th century cartels established secretariats with professional staffs in London, Zurich, or similar locations. In more recent decades, trade associations or management consulting firms have assisted with cartel operation. In these cases, the geographic locus is easy to identify. Cartels with
99
Price-Fixing Overcharges
corporate members from multiple regions are more difficult to classify, but if a supra-majority of the companies were headquartered in North America, Western Europe, or Asia, the cartel is categorized in one continent. Global cartels are those with a diverse mixture of participants from two or more continents; nearly all global cartels aimed at controlling prices in at least Western Europe, North America, and East Asia. There are some significant differences in average cartel overcharges across geographic regions. Those that operated across multiple Western European countries115 have the highest overcharges, but curiously those organized across national boundaries in Western Europe were as a group the least successful. North American conspiracies were also quite low.116 Median overcharges for global conspiracies were relatively high.117
Overcharges and Market Size A commentary in the USSGs asserts that there is an inverse relationship between the size of affected sales and the height of the overcharges achieved by cartels. No conceptual or empirical justification is provided for this assertion. Studies of cartels available to the Commission analyzed neither factor (e.g., Hay & Kelly, 1974; Asch & Seneca, 1975; Fraas & Greer, 1977; Posner, 1976). Eckbo’s (1976) and Griffin’s (1989) studies have information on price effects but do not link them to cartel size. Moreover, subsequent empirical evidence does not support a positive market sizeovercharge connection.118
DECISIONS OF ANTITRUST AUTHORITIES A survey of final verdicts of U.S. courts in collusion cases finds that 25 collusive episodes had a median average overcharge of 21.6% and a mean average overcharge of 30.0% (Connor & Lande, 2005).119 The nine cases that reported peak overcharges produce a median peak overcharge of 71.4% and a mean peak overcharge of 130%. All but five found that the cartel had raised prices by more than 10%. Due to the small number of final verdicts, it would not be meaningful to analyze these verdicts in even smaller groups. Fig. 11 and Table 9 combine that U.S. court survey with other overcharge estimates derived from cartel decisions by other antitrust authorities. There are 264 such observations – 36% from analyses of guilty findings of U.S. and Canadian courts, 24% from decisions of the EC that imposed fines,
100
JOHN M. CONNOR
60
Median
Mean
Percent
50 40 30 20 10 0 No.America
Fig. 11.
UK
EU
Europe Nations
Japan
Korea
Average Overcharges from Courts and Commissions. (Number of Decisions is shown in dark bar.)
Table 9. Cartel Overcharges from Decisions of Antitrust Authorities. Antitrust Authority
Number of Observations
Median Percentage
Mean Percentage
North America US and Canada, pre-1990 US and Canada, 1990–2005
94 55 39
20.0 18.5 21.6
36.8 38.7 34.0
United Kingdom UK Monopolies Commission UK, 1990–2005
28 24 4
19.0 13.4 41.9
73.3 74.3 67.6
Other European Nations France Germany Italy Other Europe
26 9 4 4 9
18.8 22.0 11.0 30.5 9.0
22.8 27.4 12.0 28.7 20.5
European Union European Commission, pre-1990 European Commission, 1990–2005
64 11 53
27.5 25.0 29.0
35.0 31.7 57.1
China Japan FTC Taiwan FTC Korea FTC Australia Mexico
2 10 8 13 2 1
21.1 28.4 67.5 20.0 10.5 18.8
21.1 26.3 94.9 29.4 10.5 18.8
248
23.5
43.0
Total
Source: Appendix Table 2 dated August 2005. Several decisions have alternative estimates by single authors, and some have estimates by multiple authors.
Price-Fixing Overcharges
101
20% from commissions of European nations, and most of the rest from Asian antitrust authorities. Texts of most of these decisions can be found on the web sites of the authorities or in various searchable law archives (Lexis Nexis, WestLaw, the Official Journal of the European Communities, EUR-Lex, and the like). In some cases, press releases or press summaries contained sufficient information to calculate an overcharge, but more commonly an analyst used the product definition and conspiracy dates in the opinion and applied this information to prices from a third party to calculate an estimate. As in the case of U.S. final verdicts, only a small minority of available decisions contain the appropriate quantitative data.120 The median overcharge is 23.5%, and the mean is 43%; both of these figures are close to the entire sample of 770 overcharge estimates. Besides the EC, a large number of observations come from decisions about mostly domestic schemes made by the UK Monopolies Commission in the 1950s and 1960s. Most of the remaining decisions are from other commissions that typically fined international cartels discovered since 1990. The estimated overcharges from decisions of the EU, Taiwanese, and Japanese authorities are relatively high. With few exceptions, overcharges from a jurisdiction are highly positively skewed. In three jurisdictions, there are enough observations to examine changes over time. In each case, median overcharges from 1990 to 2005 are higher than from earlier periods. This rise is most likely attributable to the higher priority being given to prosecuting international cartels in the last 15 years.
THE RELIABILITY ISSUE Many readers may have prior beliefs about the most appropriate data and methods to be used to derive estimates of the price effects of cartels. Some might regard a lengthy historical investigation with access to the internal communications of a cartel’s managers as the surest path to the truth. Others might give greater credence to such communications only where the cartelists had reason to believe that their activities were legal or where the managers are writing about an illegal cartel years after the statute of limitations had passed. Some might assume that disinterested social scientists are likely to be closer to the mark than prosecutors, plaintiffs’ counsel, defendants’ counsel, or other interested parties. Indeed, the cross checks of a more global retrospective analysis might contradict delusions of cartel managers about their power over markets. Among economists, ever cognizant of the march of progress in quantitative research methods, there may
102
JOHN M. CONNOR
be a tendency to find peer-reviewed studies applying methods of the most recent vintage to highly disaggregated, detailed data the most reliable. Among legal scholars, many will view criminal trials or guilty pleas as the gold standard of fact finding, regarding civil commission hearings and other processes with skepticism. Three approaches are taken to learn whether the various overcharge estimates are sensitive to the methods, data sources, time period, or disciplines of the authors. These analyses are reported in detail in Connor (2004b, pp. 56–67) and are summarized here.121
Sources of the Estimates Confidence in the estimates may be judged in part by the sources from which the overcharge estimates were derived. The majority of the estimates are drawn from the traditional end-product outlets of academic research: academic books, book chapters, and peer-reviewed journals account for 65% of the total (Connor, 2004b, Table 11). In addition, 15% of the estimates were taken from economist’ working papers, most of which were distributed since 2000, examined modern international cartels, and appear to be intermediate versions of book chapters and journal manuscripts.122 The majority of the government reports (4% of the estimates) were authored by civil servants with specialized training in economics, and some were written by academics commissioned by the agency; typically, these reports would be vetted by a panel of experts. Similarly, the legal decisions of the UK Monopolies Commission were reviewed and approved by panels that contained a couple of leading professors of industrial economics working alongside senior civil servants attached to the Commission. Much the same process was used for the Congressional committee reports on cartels. Court and competition-law commissions accounted for 12% of the estimates. In sum, four-fifths of the estimates are drawn from the formal or informal writings of academic social scientists, and most of the remainder was the product of professionally trained individuals subject to the checks and balances of internal reviews.123
Sensitivity to Publication Dates Here, the hypothesis examined is whether there are systematic differences between the average overcharges across time, using the date of publication
103
Price-Fixing Overcharges
of the study as a proxy for analytical advances.124 The intuition here is that the authors of more recent empirical studies of cartels have learned to avoid the methodological pitfalls of their predecessors.125 Among the economic studies that dominate the sample, there is an undeniable trend away from mere narrative historical case studies sometimes embellished with simple graphical illustrations toward more formal statistical modeling. In industrial economics, there is a trend away from evaluating cartels from the point of view of the theory of pure monopoly toward a more sophisticated and nuanced view informed by game theory and other conceptual advances. The types of publication outlets have also changed over time (Fig. 12). Before 1974, books and chapters in edited collections accounted for 58% of the publications that contained usable overcharge data. Most of these works show evidence of meticulous scholarship, but the share of them subject to blind reviews is small. After 1973, books became a minor component (11%) of this survey’s source materials. Instead, the greatest sources of overcharge estimates shifted to the published decisions of courts and commissions (44%) and to academic journal papers (34%). That is, in recent decades, most estimates are drawn from papers that have been peer-reviewed, from an adversarial forum, or from decisions likely to be reviewed by courts of appeal. Some may regard review processes as likely to induce more reliable calculations. The results of a temporal analysis are displayed in Connor (2004b, Table 12A). The publications are classified according to four periods that correspond roughly to milestones in social-science analysis of cartels (before
Refereed Journal Papers Books Governments and Antitrust Authorities Book Chapters Working Papers
Before 1974 After 1973
Other
0
Fig. 12.
20
40
60
80
100
Publication Sources of Average Overcharge Estimates. (Total publications 259 as of August 2005. Antitrust Authorities counts decisions.)
104
JOHN M. CONNOR
1891, 1891–1945, 1946–1989, and 1990–2004). There is no evidence for concluding that overcharges vary systematically by date of publication. For example, in the case of cartels that ended in the pre-antitrust era, one sees that both contemporary and early writers arrived at moderate estimates of cartel price effects – median estimates of 22–30%. Studies published prior to 1990 tended to calculate relatively low median price effects.126 However, as the methods of scholarship presumably improved, the estimated price effects of cartels active in the most laissez-faire of economic environments actually rose to a median of 30%.127 For cartels that ended between 1890 and 1989, current research produces median estimates that are quite close to contemporary chroniclers of the same cartels.
Intra-Episode Comparisons The third check on reliability of estimates across various analytical methods controls for changes in the composition of the sample by focusing on pairs of estimates applied to identical cartel episodes. Recall that a cartel episode refers to a single market, time period, and form of cartel organization. There are 291 pairs of observations available for this analysis of reliability, which examines six general methods of estimation. The most widely used is the socalled before-and-after method in which the price during the episode is compared to one of three ‘‘but-for’’ or base prices. The second most popular method is statistical modeling, which accounts for 20% of the estimates. The yardstick method accounts for about 10% of the sample. Overcharges derived from costs of production or profits are the least frequently employed method (about 3%). These five methods have been sanctioned by U.S. courts for determining damages in price-fixing trials (Connor, 2004a). Sixth, approximately 10% of this study’s estimates are quotes from or interpretations of decisions made by antitrust authorities.128 By and large, different authors and different methods applied to identical cartel episodes do not result in markedly different estimates. The correspondence among the three before-and-after methods is quite close. Nevertheless, there are two differences worth commenting on. One somewhat surprising result is that the before-and-after method produces cartelovercharge estimates that are higher than econometric modeling applied to the same episodes. Econometric techniques offer the opportunity to the analyst to make precise allowances for several sources of shifts in demand and supply, for seasonality, for trends in technology, and for feedback effects. If in fact econometric techniques are the most accurate, what this
105
Price-Fixing Overcharges
result seems to suggest is that authors of traditional before-and-after analyses are failing to adjust for all the competitive factors that might drive up the competitive benchmark price. However, this result could be explained by other factors, such as the time available to perform a calculation or to differences in access to confidential price data.129 Second, compared with the before-and-after, the cost-based and yardstick techniques yield relatively high overcharge estimates.130 This suggests that the methods that use costs or profits fail to fully account for all competitive industry costs, perhaps those related to product marketing or overhead. Similarly, as yardsticks are frequently chosen to be prices in proximate regions in which the cartel did not attempt to fix prices, this result suggests that analysts may be identifying prices in regions with lower cost structures than the conspiracyaffected markets. Possibly, the full costs of transportation and transfer from geographic yardsticks to the affected geographic market are underestimated. If the yardsticks are product substitutes, analysts may have underestimated quality differences between the cartelized product and the analogous product.131
CONCLUSIONS This paper’s major goal is to collect and analyze the largest possible body of serious, quantitative estimates of price-fixing and bid-rigging overcharge rates. In publications dating from 1888 to mid 2005, some 1,040 such estimates were assembled. This survey analyzes information found in 259 social-science studies and adverse legal decisions regarding mark-ups by private, hard-core cartels. Those publications yielded 770 observations of ‘‘average’’ overcharges.132 The primary finding is that the median133 cartel overcharge for all types of cartels over all time periods is 25.0%:18.8% for cartels with solely domestic membership and 31.0% for international cartels.134 Thus, international cartels have been about 65% more effective in raising prices than domestic cartels. Cartel overcharges are skewed to the high side, pushing the mean overcharge for all successful135 cartels to 43.4%. The 270 ‘‘peak’’ cartel overcharges in the sample are typically double those of the long-run averages.136 This paper’s findings are generally consistent with the few, more limited works that comment on cartel overcharges.137 Six previously published economic studies with samples ranging from 5 to 38 overcharges report a simple average median overcharge of 28.1% and a weighted average mean overcharge of 36.7% of affected sales. Moreover, the social-science and legal
106
JOHN M. CONNOR
sources yield generally similar estimates. The results of the survey of final verdicts in decided U.S. horizontal collusion cases, only three of which were international cartels, show an average median overcharge of 21% and an average mean overcharge of 30%.138 Outside the United States, decisions of competition commissions contained information that resulted in 154 overcharge estimate; median average overcharges across jurisdictions typically139 ranged from 19 to 29% of affected commerce, and the mean range was 21–73%. Overall, the decisions of all courts and commissions implied a median cartel mark-up of 23.5% and a mean of 43.0%. The authors’ professions, types of publications, years of publication, degrees of peer review, and analytical estimation methods from which these estimates are derived vary greatly. There is some indication that estimates prepared from the yardstick method and those based on decisions of antitrust authorities are higher than other approaches.140 Otherwise, however, extensive examinations of variation in overcharge rates across such categories give no reason to regard any sub-set of the sample as inherently biased or unreliable. The results of the survey have significant policy implications. First, this paper’s introduction noted that there is a view among some antitrust writers that there is little evidence that cartels raise prices significantly for a period long enough to justify extant anticartel laws and, especially, extant cartel penalties. Consequently, they argue for the repeal or scaling back of the fines or damages that result from collusion. This survey’s results, which are based upon an extraordinarily large amount of data spanning a broad swath of history of all types of private cartels, sharply contradict these views. In fact, the data suggest the opposite. Mean overcharges are two to four times as high as the level presumed by the USSC. Second, bid rigging was no more injurious to direct buyers than other forms of collusion. These results suggest that antitrust sanctions’ guidelines should not treat bid rigging more harshly than other forms of collusion. Third, international cartels are typically more destructive of competitive market forces than domestic conspiracies. Connor and Lande (2005) propose raising the overcharge presumption for U.S. fines to 15% for domestic cartels and 25% for international cartels.141 This is a conservative and modest proposal in light of this article’s demonstration that cartels typically generate at least two or three times the antitrust damages presumed by the current Sentencing Guidelines. Average fines imposed since 1995 by Canada and the EU on identical cartels have been lower than U.S. government fines (Connor, 2005). When the effects of private suits are factored in, it is clear that the U.S. court
107
Price-Fixing Overcharges
system is already shouldering the bulk of the world’s burden of punishing international cartels. This survey suggests that overcharges generated by cartels discovered in most jurisdictions are higher than North Americacentered cartels. Consequently, anticartel laws and fine-setting practices abroad are in even greater need of strengthening. Global cartels have historically generated greater overcharges than other international conspiracies. Despite the evident increases in cartel detection rates and the size of monetary fines and penalties in the past decade, a good case can be made that current global anticartel regimes are under-deterring (Bush et al., 2004; Connor, 2005). Global cartels are more difficult to detect, have less fear from entry of rivals, achieve higher levels of sales and profitability, and systematically receive weaker corporate antitrust sanctions than comparable domestic cartels. Base fines of 20% of cartelists’ affected commerce, even when adjusted by significant culpability multipliers,142 will do little to deter most of these cartels. One sanguine development is that for most types of cartels, there are reductions seen in cartel mark-ups over time. Because the post-1990 era has been the period with by far the highest level of fines imposed, this decrease is consistent with the theory of optimal deterrence. It also suggests that the recent worldwide trend towards the intensification of cartel penalties has been desirable. If procedures for calculating criminal fines correspond more closely to the actual levels of cartel overcharges, monetary sanctions against price fixing will more closely provide optimal deterrence.
NOTES 1. Longevity, also called duration, measures the lifespan of a cartel or, if it has more than one, the length of time of one episode. Some researchers use the term stability synonymously with duration, but more commonly, it refers to the absence of price wars or other reversions to competitive conduct during a cartel’s time span. Stability is perhaps equivalent to low variation in a cartel’s ‘‘discipline,’’ where discipline may be measured by how close a cartel’s selling prices are to its desired target price or the theoretical monopoly price. In the context of commodity agreements or marketing orders, stability will show up as lower variation in prices compared with the absence of such an agreement. Efficiency can refer to static allocative efficiency (low net social welfare loss) or, rarely, to technical efficiency or dynamic efficiency (rates of technological change). Allocative inefficiency is smaller than but closely correlated with the overcharge. 2. The benchmark is referred to as the ‘‘but-for price’’ – the market equilibrium price that would have been observed were it not for the overtly collusive conduct of the sellers. The benchmark may be the purely competitive price, or it may be a
108
JOHN M. CONNOR
somewhat higher price generated by legal tacit collusion. The overcharge rate is similar to the Lerner Index of market power, except that the Lerner Index’s benchmark is the realized market price. 3. The overcharge from a buyers’ cartel is symmetrically defined as a price decrease. 4. Customers are direct buyers and they are usually industrial buyers, but overcharge pass-on will transfer the losses in whole or in part to final consumers as indirect buyers. If cartels improve technical or dynamic efficiency, this may offset the buyers’ losses. 5. In large markets for manufactured products, the dead-weight loss is typically one-fifth to one-tenth as large as the overcharge, and the two losses are highly correlated (Peterson & Connor, 1995). 6. Criminal filings are made in cases of per se, covert, intentional conspiracies by participants who are aware of the probable anticompetitive consequences (Hovenkamp, 1999, pp. 585–586). While there are a few exceptions, potentially illegal anticompetitive conduct such as information sharing, signaling, refusals to deal, resale minimum-price maintenance, tied sales, exclusive dealing, patent or trademark pooling, mergers, monopolization, and attempts to monopolize are treated as civil matters. More than 90% of all naked cartel cases are brought as criminal actions, but a small number of such cases are, at the discretion of the DOJ, filed as civil matters. 7. The USSC Guidelines start with a base fine double the 10% presumed overcharge and use it in conjunction with the assigned base offense level (of 10) for antitrust offenses. They adjust this offense level by a number of factors, such as whether bid rigging and other aggravating factors were involved, and by mitigating factors as well. This adjustment results a pair of ‘‘culpability multipliers’’ that are between 0.75 and 4.0. The product of the base fine (20% of the affected commerce) and the culpability multipliers results in the fine range that is to be imposed on a cartel member. Thus, the fine range recommended for convicted cartelists is at its lowest 15% and at its highest 80% of affected sales. These fines usually are adjusted downwards for cooperation or as a part of the Division’s leniency program. The USSC’s Commentary also notes that ‘‘In cases in which the actual monopoly overcharge appears to be either substantially more or substantially less than 10%,’’ it might not employ the 20% base fine. But, in practice, the DOJ almost always uses the figure of 20% of affected commerce as their starting point in their criminal fine calculations. 8. See USSC Guidelines for the United States Courts, 18 U.S.C. Section 2R1.1, Bid-Rigging, Price-Fixing or Market-Allocation Agreements Among Competitors, Application Note 3. 9. See United States v. Hayter Oil Co., 51 F.3d 1265, 1277 (1995). The court noted: ‘‘The offense levels are not based directly on the damage caused or profit made by the defendant because damages are difficult and time consuming to establish. The volume of commerce is an acceptable and more readily measurable substitute y.’’ 10. In a statement to the Commission, Assistant Attorney General Ginsburg stated that ‘‘the optimal fine for any given act of price fixing is equal to the damage caused by the violation divided by the probability of conviction y such a fine would result in the socially optimal level of price fixing, which in this case is zero’’ (USSG, 1986, p. 14). He stated his judgment that ‘‘price fixing typically results in price
Price-Fixing Overcharges
109
increases that has harmed the consumers in a range of 10 percent of the price y’’ and that these violations had no more than 10% chance of detection (ibid., p. 15). 11. A common procedure for preparing testimony for the Assistant Secretary would include the preparation of a confidential ‘‘white paper’’ by the Antitrust Division’s economists. 12. Cartel fines in the United States and Canada peaked in 1999 and peaked in the EU in 2001 (Connor, 2004c). 13. Adler and Laing are correct that the fining standards of the DOJ do not compute fines simply as a function of damages, but rather as a function of the company’s affected commerce, which is loosely related to damages (see fn. 7, supra). However, these authors do not document their claim that antitrust fines are harsher than other corporate crimes. 14. In the case of injuries from naked price fixing, most legal scholars agree that the appropriate rule for damages is a multiple of the overcharge plus the deadweight loss, where the multiple is the inverse of the probability of detection (Hovenkamp, 1999, pp. 645–649). However, the difficulties of calculating the probability of detection and of the dead-weight loss have encouraged U.S. courts to equate pricefixing damages with the direct-purchaser overcharge (ibid., pp. 651–656). Therefore, in most legal writings, if not otherwise specified, harm can be presumed to be the overcharge. 15. Denger appeals primarily to an increase in settlement rates in treble-damage direct-purchaser suits to establish the unfairness of the high fines imposed on corporate price fixers, an increase that, he believes, cannot be explained by increases in overcharge rates. He cites about eight domestic U.S. law cases that settled for 2–4% of sales in the 1970s and one international case in 2001 that settled for 18–20% (pp. 3–4). It is argued below that settlements are inappropriate evidence in this context. 16. Klawiter contrasts enforcement powers in the late 1990s with the clearly suboptimal maximum fine of $10 million available to the DOJ in the 1970s and 1980s. 17. Rule 17 was amended in 2004, but these provisions were unaffected. 18. Under Section 45 of Canada’s Competition Act, fines are limited to C$10 million, but foreign price-fixing conspiracies can be prosecuted under Section 46, which has no fine limit (Low, 2004, p. 17). 19. In 1986, the Assistant Attorney General for Antitrust, Douglas Ginsburg, estimated that the enforcers catch less than 10% of all cartels. See USSG (1986, p. 15). If he is correct, optimal fines for cartels should be 10-fold damages! The percentage of cartels that are caught and proven is probably much higher today. See Spratling (2001). There is, however, neither evidence nor speculation that it exceeds 33%, so there is no reason to believe that the treble-damage remedy should be lowered. See also the discussion in Landes (1983, p. 115, fn. 1). 20. Those critical of aggressive antitrust policy have often embraced the notion that cartels are fragile coalitions. However, empirical studies of discovered cartels from the 1920s to the early 2000s find that the average duration is between four and seven years (Zimmerman, 2005). Legal U.S. export cartels – a sample unaffected by possible bias inherent in studies of prosecuted conspiracies – endured an average of 5.3 years (Dick, 1996). 21. For larger price-fixed markets ‘‘y ten percent is almost certainly too high’’ (p. 343). In part, they rely on evidence of price-fixing settlements rather than awards.
110
JOHN M. CONNOR
They assert that there is ‘‘a sparse amount of economics literature’’ on cartel markups; my reading of their article turns up seven to ten observations with a median range of 8–14% (see Table 1). 22. See Baker (2003), Werden (2003), and Kwoka (2003). According to Kwoka (2003, fn. 2), Crandall and Winston’s earlier drafts ‘‘y endorsed consideration of outright appeal of the antitrust laws.’’ 23. Space constraints do not appear to be responsible for such a skimpy treatment of this topic, for they list 59 references. Somewhat inconsistently, Crandall and Winston do admit that the large DOJ fines meted out to cartels in the 1990s possibly deterred the most harmful cartels. They appear to refer to the lysine, citric acid, and vitamins cartels – international cartels with overcharges in the 15–40% range. 24. For diversified companies for which a cartelized product is a small share of its total sales, EU fines can come closer to optimal deterrence levels. However, given a low probability of detection, durable cartels with average overcharges are difficult to deter with EU fines alone (Wils, 2005). 25. The reference section of Connor (2004b) lists about 350 sources with useful information about the private cartels in this paper’s sample. Only about 200 contained usable quantitative overcharge estimates. 26. An issue among economists up to the 1940s was whether cartels raised average prices in a manner consistent with monopolies or whether cartels simply stabilize price movements with no net increase in prices. Liefmann was in the minority that accepted profit maximization as the goal of a cartel. 27. Liefmann accepted that raw-material cartels typically did raise prices, but considered the price effects of industrial cartels an open question (see the appendix). While most of his contemporaries considered such calculations impossible, Liefmann took the position that precision was difficult because of simultaneous changes in demand and supply. 28. Hexner (1946) spent dozens of pages toying with alternative definitions of the cartel, ultimately adopting one quite close to Liefmann’s. 29. Stocking and Watkins had access to the results of a number of major investigations. The Temporary National Economic (or ‘‘Kilgore’’) Committee published its hearings a few years before their books were published (U.S. Congress, 1938– 1940). Other Congressional committees investigated the munitions industry and patent pools. The authors also had information on U.S. prosecutions of more than 40 international cartels. 30. Technically, the 1911 conviction of American Tobacco et al. was the first U.S. prosecution of an international cartel, but the international aspect of the case was a minor aspect of the case. 31. Overcharges were taken from about 50 books and chapters in edited books, of which 30 were published after 1950. Compared with the total number of economics books printed after 1950, the share of them devoted to cartel studies is smaller than before. 32. When the guilty pleas were received in the Philadelphia U.S. District Court in early 1961, nearly every daily newspaper in the United States placed the events on their front page. 33. The conspiracies are notable for their duration (up to 40 years), the as yet unsurpassed size of the sales involved ($7 billion per year in the late 1950s), the large
Price-Fixing Overcharges
111
number of well-known companies involved (General Electric, Westinghouse, etc.), the size of the fines imposed (over $2 million), the size of the damage awards in three trials and private settlements ($400–$500 million) from more than 1,900 suits, and the imposition for the first time of significant prison sentences for several top executives. 34. I do not include OPEC’s price effects in this survey because it was created and enforced by what amounts to a multilateral treaty organization. 35. Eckbo (1976) comes close. Eckbo studies 51 episodes in 18 markets, but does not really calculate overcharges so much as place them somehow in high/low categories; Griffin terms Eckbo’s approach subjective. 36. The Lerner Index is the same as the overcharge, except that it is measured by dividing the market price by the monopoly price instead of the competitive price. That is, the Lerner Index is a margin on the collusive selling price, while the overcharge is a mark-up on the competitive price. Thus, for the same cartel, the Lerner Index is a smaller number than the overcharge, though the difference is small for small overcharges. The Lerner Index is L ¼ (PC)/P, where P is the observed market price and C the but-for or competitive price. Because C is equal to marginal cost in competitive equilibrium, L is also a profit margin on sales. L is zero in perfectly competitive markets and has a maximum value of one. The monopoly overcharge is a mark-up: MO ¼ (PC)/C. MO is also zero in perfectly competitive markets, but can approach positive infinity when C is very small. Because P is always greater than or equal to C, MO is greater than L whenever L is positive. Simple algebraic substitution allows one to express MO as a function of L, viz., MO ¼ L/(1L). 37. Of the leading textbooks in industrial organization, Carlton and Perloff (1990) devote more space to cartels than most – almost 50 pages out of 852 total pages. This work mentions by name 60 cartels, most of them interwar, international cartels. Other textbooks have far fewer numbers of cartels cited. 38. Hay and Kelley (1974) authored a classic review of 65 U.S. price-fixing conspiracies, which Fraas and Greer (1977) extended to 606 cases from 1910 to 1972. Both studies contain a wealth of information about the number of conspirators, duration, industry, and specific collusive methods employed. However, neither survey covered the topic of price effects, presumably because of the paucity of such data. 39. Eckbo (1976) comes close. Eckbo studies 51 episodes in 18 markets, but does not really calculate overcharges so much as place them somehow in high/low categories; Griffin politely characterizes Eckbo’s approach as ‘‘subjective.’’ 40. I exclude 18 of Griffin’s episodes from government-sponsored cartels. 41. The present survey also finds that about 7% of the sampled cartels failed to raise prices. 42. If P is the collusive price and C the competitive price, then the Lerner Index is (PC)/P, whereas an overcharge is (PC)/C. That is why, the Lerner Index is a margin (a mark down from the market price) and the overcharge is a mark-up (on costs or the but-for price). 43. Suppose the competitive benchmark price is $1.00 and the cartel mark-up (or overcharge) is 5%. Then, the Lerner Index L is (1.051.00)/1.05 ¼ 0.0476 ¼ 4.76%. However, if an overcharge is 25%, L ¼ (1.251.00)/1.25 ¼ 0.20 ¼ 20%. One can derive algebraically a one-to-one linear functional relationship between the overcharge
112
JOHN M. CONNOR
and L; it equals L/(1L). In this paper, I convert estimates of the Lerner Index of monopoly to the overcharge. 44. One of them (Block & Nold, 1981) is irrelevant because it quotes the ratio of out-of-court settlements to annual sales for several U.S. bread price-fixing cases. As Cohen and Scheffman recognize in fn. 66, both the numerator and denominator of this ratio are inappropriate indicators of an overcharge; nevertheless, in the text of their article, they persist in citing this ratio. 45. The Chicago School of Industrial Economics is well recognized in textbooks (e.g., Martin, 1994, pp. 8–11) and by members of the school itself (Posner, 1979). The Chicago School generally maintains that sustained collusion by private firms alone is empirically rare. Posner’s (2001) insistence on widespread cartel success is a departure from the School’s normal themes. 46. ‘‘I have very little evidence on the excess profits y [from] cartelization. For fifteen cartels y I have anecdotal evidence of price increases y’’ (p. 20). 47. A few non-members that participated in an OECD-sponsored ‘‘Global Forum on Competition’’ also submitted responses to the survey. ‘‘Hard-core’’ is a European term that refers to conspiracies that fix prices and/or quantities. Other cartels (soft core?) cooperate on information, technology, marketing, and the like. The distinction seems roughly to correspond to criminal versus civil violations under U.S. law. 48. In a few cases, the harm was reported as a monetary value and the size of affected commerce was missing, but I was able to find a reasonable estimate of the affected commerce from an alternative source. For example, the U.S. DOJ provided a monetary estimate of the U.S. harm caused by the international lysine cartel of 1992–1995, and I found the value of affected commerce in a sentencing opinion written by a federal judge in a criminal jury trial that convicted three of the cartel’s managers. I was able to derive 16 overcharge percentages, of which 12 were long-run and 4 were peak. 49. Bringing criminal indictments for only hard-core cartels is a matter of custom, not law. Some of the 5–10% of U.S. DOJ horizontal or vertical conspiracy cases handled through civil indictments could be criminally actionable. In addition, some hard-core cartels are brought as civil matters because prosecutors judge that the criminal burden of proof cannot be met. 50. Wallace and Edminster (1930, Appendix A) provide a convenient chronology of most government-sponsored export-control monopolies: the Japanese camphor monopoly of 1899, the Italian citric acid monopoly of 1910, the Greek currant monopoly of 1895, and the New Zealand kauri-gum monopoly of 1927 are examples of clearly public cartels. 51. In some cases particularly in the early 1930s, the earlier phases of an international cartel were controlled by national producers’ organizations that negotiated voluntary quota reductions; when cheating threatened the effectiveness of the cartel, colonial or metropolitan governments stepped in to pass mandatory supply-control legislation. The early phase of the cartel I deem private, but not the latter. 52. See Connor (2004b, Appendix Table 5) for a brief list of excluded studies and the reason for their exclusion. I am indebted to dozens of colleagues who responded to appeals for information useful for this study. 53. Unless available in translation, I have mostly confined this survey to English language sources. Many antitrust authorities now translate their press releases and
Price-Fixing Overcharges
113
annual reports into English; moreover, members and some nonmembers submit summaries of their annual reports in English to the OECD. The preponderance of sources published after 1945 is explained by the growth of the field of industrialorganization economics. Although theoretical concepts of competition and monopoly go back at least to Adam Smith, the field is generally regarded as having developed a separate identity only in the 1930s. 54. Several historical studies of cartels were authored by Europeans or Japanese scholars. A few economic studies of cartels were written by UK or Australian economists (Clarke & Evenett, 2003; De Roos, 2006). 55. One might speculate as to why this is so. The supply of well-trained industrial economists in Europe is unlikely to be an explanation. The principal European organization for industrial economists (EARIE) was more active in sponsoring meetings the past decade than its U.S. counterpart (IOS), and the EARIE meetings had a good proportion of empirical and legal-economic papers. However, the structure of academic departments at European and Asian universities may explain the paucity of useful studies. Compared to U.S. departments of economics, European departments tend to be smaller (perhaps falling below the threshold necessary for collaborative teamwork on large-scale data sets), more focused on IO theory, and have different expectations for Ph.D. dissertations. Perhaps, a more important factor is the inability of academics to obtain access to the structural and price data needed to calculate overcharges. Civil cases are unusual in Europe, so the little work being done on cartel overcharges is done in-house by antitrust authorities. Unlike North America, there is little mobility between the staffs of European antitrust authorities and universities or think tanks. Finally, a survey of European and North American industrial-organization economists reveals that there are very few attitudinal differences between the two groups on economic theory, but the former were less likely to expect economists to influence competition policies (Aiginger et al., 2001). 56. If a credible study of a cartel concludes that it was ‘‘ineffective,’’ I have coded this comment as a zero price effect and included this observation in the averages. Likewise, conclusions that the impact of collusion was ‘‘overwhelmed’’ by natural market forces are interpreted as a zero overcharge. However, vague conclusions that a cartel episode was ‘‘effective’’ are not tabulated in the quantitative summaries. 57. Some scholars may have relied on what they judged to be credible journalistic reports of overcharges. 58. The influences of types of publications and methods of analysis are formally analyzed in Connor and Bolotova (2006). 59. I have confined journalists’ accounts of cartels primarily to book-length treatments of cartels, in the belief that such monographs are in-depth accounts of a cartel collected from many sources, some of them anonymous, over a period of time sufficient for the author to provide a balanced account of conflicting claims. Books by journalists typically do not focus on the quantitative economic aspects of the case at hand, so in practice, there are relatively few overcharges drawn from these sources in the present study. I rarely include overcharge estimates embedded in newspaper or magazine articles, though some specialists may judge such assertions to be sufficiently reliable to include in their published studies. For example, Elzinga (1984) cites Demaree (1969), and Carlton and Perloff (1990) cite Smith (1963).
114
JOHN M. CONNOR
60. The danger with this method is that the product yardstick may be a substitute for the cartelized product, and, hence, price-responsive to a cartel overcharge. 61. These data are often proprietary facts revealed during the discovery phase of litigation or submitted to an antitrust authority under compulsory legal processes. In addition to transaction prices of the defendants, production and marketing costs of details of business contracts may be handed over on a confidential basis. 62. Either a dummy variable is included for the assumed collusive period, or the model can forecast or backcast benchmark prices from a noncollusive period. 63. By contrast, if a cartel operated during a span in which cost conditions (input prices, expansion of capacity, inventories, and technology) were steady and demand conditions (consumer preferences, disposable income, available substitutes, and the like) did not shift, then fancy econometric models and the more traditional methods will yield the same overcharges. For durable cartels, constancy of all these factors is unlikely. 64. See the discussion in Sullivan and Grimes (2000, pp. 165–233), which shows that in per se cases, the plaintiff does not have to prove whether prices rose (or even whether defendants had market power). The issue of whether prices rose can be an element of a rule of reason case, but rule of reason cases do not give rise to criminal fines, so are not the subject of this paper. 65. Even at the sentencing phase of criminal price-fixing trials, prosecutors rarely offer information on damages. 66. Class-action settlements in North America are widely advertised in legal notices, but information about the rate of overcharge is not usually provided. Notices on the Internet are fleeting. 67. I excluded cases that were overturned on appeal. 68. Computerized searches were not, with only a few exceptions, particularly helpful. Most searches turned up hundreds of useless citations. 69. For example, inquiries were made on the antitrust list serves of the ABA Antitrust Section, the National Association of Attorneys’ General, and of the American Antitrust Institute. 70. Many of the verdicts found were only expressed in dollar amounts, which could not be translated into percentages. 71. The most useful sites were: The European Commission (EC); the Australian Competition and Consumer Commission (ACCC); the Canadian Competition Bureau (CCB); the German Bundeskartellampt (BKA); the Fair Trade Commissions of Japan, Korea, and Taiwan; and the competition authorities of Finland, Sweden, Norway, Denmark, the Netherlands, France, Italy, and Israel. 72. Italy, the Netherlands, and Korea are exceptions to this rule; these overcharges are collected in Connor (2003). Moreover, these antitrust authorities and some others have reported a few of their decisions and overcharge estimates to the OECD (2002–2003). 73. I read almost 100 EC decisions that imposed fines on cartels (listed in Burnside (2003, Annex 1) and others published since 2003). The UK Monopolies Commission also released detailed reports, and I read about 40 of the ones that declared the cartel was ‘‘not in the public interest.’’ 74. Occasionally, the commission reported an absolute overcharge, and the size of affected sales needed to be estimated.
Price-Fixing Overcharges
115
75. For example, the ‘‘nitrogen’’ cartel is in fact dry salts of nitrogen used as fertilizer, not the gaseous form. The largely successful ‘‘vitamins’’ cartel is best regarded as a series of overlapping ventures, each of which focused on one of 16 products. 76. Episodes are in principle different from phases of cartels that give rise cartels instability. Episodes mark changes in cartel organization, whereas stability is measured by changes in the degree of cartel discipline or cohesiveness. 77. In the rare instances where a cartel kept the market price absolutely constant for the whole episode, the two overcharge concepts will be the same number. 78. A few markets were cartelized by both types; typically, a domestic cartel was expanded to respond to foreign competition. The potash cartel is one example; originally German, it became international shortly after World War I because after potash mines in Lorraine became part of France, a joint Franco-German scheme was established. Thus, the potash cartel as a whole is counted as international; however, the earlier domestic episodes (and overcharges imposed at that time) are classified as national. 79. In Europe, bid rigging is generally referred to as collusion involving ‘‘tenders.’’ 80. Only a couple of cartels were oligopsonies. 81. Counted in this category are criminal convictions; adverse decisions of the UK Monopolies Commission, which made recommendations to the government similar to consent decrees; adverse decisions of the European Commission and similar civil authorities; and those cartels that paid court-approved damages. Seven unfinished probes by antitrust authorities are placed in this category. 82. However, I was able to extract only eight of these episodes’ price effects, plus one overall estimate, from this source. One other study of UK national cartels provided a summary mark-up estimate for 40 cartels. Otherwise, all the other episodes are counted in the manner described. 83. I have preserved these ranges in the appendix tables of Connor (2004b), but have used the midpoints of the ranges for the tables in this paper. The median ranges, if any, are quite narrow. 84. The same estimates sometimes appear in multiple publications. Here, I count only the total number of books, articles, and reports that contain one or more original estimates. The very few undercharges are entered as positive numbers. 85. They are also convenient to chart changes in the historical views toward cartels or in methods of analysis. 86. There were written laws against price fixing in ancient times (Assyria, for example), in 15th century England, and in revolutionary France. None is known to have been effective against private hard-core cartels. 87. But few economists. The first time the Supreme Court took notice of economists was in the 1925 Maple Flooring decision (Kovacic & Shapiro, 2000, p. 47). 88. Even in recent decades, however, there is a notable absence of empirical publications by European economists working out of European research institutions. Obviously, there are many European analysts, most lawyers by training, located in EU and national antitrust authorities’ bureaucracies and performing cartel studies, but few of them publish outside of their governments’ official organs. 89. Raised to $100 million in April 2004; maximum prison sentences rose from 3 to 10 years.
116
JOHN M. CONNOR
90. This cartel was controlled by mine owners who sent coal by coastal ships from Newcastle to London. The number of mines was quite large at times. Coal was mined in many parts of Britain, but when railroads from the Midlands reached London in the early 1840s, the Newcastle owners’ transportation-cost advantage disappeared. 91. Although there is a dip in 1946–1990, the correlation between the number of observations per year and a linear time trend is r ¼ +0.72. 92. When the UK, Germany, and the EEC began requiring registration of cartels in the 1950s, hundreds came forth in each jurisdiction. 93. I do not include national cartels that were fostered by governments (some governments even compelled all the companies in an industry to join) in this data set; likewise, I exclude many international commodity-stabilization schemes that were regulated by government ministries under parliamentary laws or came about because of a multilateral treaty. The second tea cartel in the 1930s, which was authorized by several parliaments of the British Empire and regulated by the Colonial Office, is one example. However, I do include a few international cartels with one or more members consisting in part of government-appointed committee members, government-owned corporations, or government-sanctioned national cartels, if they were formed by a voluntary agreement among the members. An example is the sugar cartel in the late 1930s. Many of the European export cartels also created national monopolies for their members. 94. U.S. companies apparently believed that patent pooling with foreign firms was legal; others joined cartels indirectly through controlled overseas subsidiaries. These and other subterfuges were judged illegal by U.S. courts. 95. This ratio may be deceptively high. Many durable cartels straddled eras that bridged shifts in public attitudes or antitrust enforcement. Almost all the sanctionedcartel observations prior to 1890 derive from the Newcastle Vend, which was not ‘‘punished’’ until the 1830s when a British Parliamentary committee issued an unfavorable report but no further consequences. Similarly, the U.S. anthracite coal cartel operated for four decades before it was indicted. 96. Export cartels that in theory did not affect the jurisdiction’s commerce were permitted in the United States from 1918 and in most other nations throughout the 20th century. Today, fewer than one-third of all countries permit export cartels, and many that have an antitrust exemption appear ready to repeal the loophole (Levenstein & Suslow, 2004). 97. In the last decade, announcements of probes, guilty pleas, and fines on cartelists are more and more to be found in convenient internet sites and through internet search engines than formerly. 98. They are UK copper smelting (1787–1829) and cast-iron pipes in the United States (1895–1896). There are no examples from the early 20th century. 99. ‘‘Successful’’ cartels are those with non-zero overcharges. 100. The correlation of median overcharges of international cartels to a linear time trend is r ¼ 0.805 and for domestic cartels r ¼ 0.196; there are too few periods to test the bid-rigging types; the trends for the other types are insignificant (for example, among cartels found guilty, the coefficient is r ¼ 0.238, the only positive trend). Grouped data are from Table 6; time is the midpoint year. The time trends are similar but slightly stronger among the successful cartels; using 336 micro-observations, the correlation of average overcharges with the episode end date is 0.162 (spreadsheet
Price-Fixing Overcharges
117
dated August 2005). Connor and Bolotova (2006), fitting a sub-set of this study’s sample to a regression model, confirm the decline. 101. It is rather odd that the notable surge in discovered international cartels after 1990 came at a time when the profit incentives for cartel formation were at an historic low (Connor, 2003). Of course, if profits declined in the 1980s and 1990s, it is possible that the percentage increase in expected cartel profits may have been at an historic high point. Uctum (1998) presents evidence of just such a decline in the USA, Canada, Germany, and Japan from the 1950s or 1960s. 102. This last observation should be ignored because there is only one legal cartel formed after 1990. 103. Few international cartels in 1900–1945 had U.S. corporate members. Those U.S. companies that did join international conspiracies may have believed that they had structured their participation in international cartels in ways that would not run afoul the Sherman Act. 104. One possibility is the rise in exports of manufactures from China. Prior to 2005, there is no example of a Chinese company joining an international cartel. 105. Cohen and Scheffman (1989, p. 345) also cite low normal profits and declining demand. 106. The authors interpret their results in two ways. Firms are more likely to collude when industry conditions cause profits to decline, or cartels that are relatively ineffective at raising prices are also inept at hiding their illegal conduct and, consequently, the most likely to be detected and indicted by the antitrust authorities. 107. Note that from a legal perspective, each episode is an actionable offense. For the highest overcharges, the implied own-price elasticities of demand are very large. One of the highest overcharges (800%) is for tungsten carbide, for which General electric had a monopoly in the United States in 1927–1937. This newly developed material was sold at $453 per lb to most customers and at $360 per lb to a few favored buyers; up to 1927, the Krupp sold it at $50 per lb in the United States and during 1927–1937 at $45–$50 per lb in Europe (Stocking and Watkins, 1948, p. 132). These numbers imply that the U.S. elasticity of demand was 81.5 to 64.8. U.S. consumption was less than one-twelfth that of Europe. 108. There is no need to examine effective cartels separately, because nearly all of the peak price effects are non-zero. 109. Peak price changes may well be affected by short-run shifts in demand. Exogenous, unanticipated shifts in demand may exaggerate the peak price changes. However, in some cases, these shifts are endogenous. Especially when a well-financed cartel felt free to announce a new agreement that buyers perceived as likely to be effective, ‘‘panic buying’’ often ensued, which amplified the purely collusive effect on prices. 110. These data are over-weighted by observations taken from the interwar period. Approximately one-fourth of the 210 observations available for Table 8 refer to interwar cartels, which have been well studied by economic historians who often had available public commodity-exchange prices. Almost 30% of the observations on peak prices are for the period since 1991. 111. The correlation for bid-rigging types was not calculated because only three periods have sufficient observations. 112. The correlation over time (the midpoint year of each period) for international cartels is r ¼ 0.900 and for national cartels r ¼ 0.220; for guilty cartels
118
JOHN M. CONNOR
r ¼ 0.074 and legal cartels r ¼ 0.213 (first four periods) or 0.586 (six periods); and for classic cartels r ¼ 0.284. 113. These ratios could be relevant for assessing whether cartels intended to maximize price increases or gave greater weight to controlling variation in their collusive prices. Apologists for cartels, particularly those writing about international cartels during the Great Depression, tended to assert that cartels did not aim to raise prices so much as stabilize prices (Marlio, 1947; Pyndyck, 1979). There is little evidence in Table 9 that the interwar, international cartels achieved greater price stability than those before or after. 114. The major exception is export cartels, which are categorized in their country or region of origin but set prices in the ‘‘rest of the world.’’ 115. In the past few decades, these correspond to intra-EU international cartels. 116. Connor and Bolotova (2006) confirm that North American cartels and those from Western Europe as a whole have significantly lower overcharges. 117. When this analysis is repeated using post-1989 data, the ranking remains the same but differences are smaller. 118. The only appropriate data of which I am aware are those contained in Connor (2003, Tables A.1–A.12). This working paper has developed affected sales and overcharge data for a minority of modern international cartels; approximately 92 pairs of such data are available; sales are in current U.S. dollars and generally fall into the decade of the 1990s. Correlation statistics were calculated for a number of sub-samples. The first sample of 50 cartels examined the largest geographic market for each cartel; the coefficient was not significantly different from zero (r ¼ 0.105). To see whether extreme observations might unduly affect the result, I repeated the experiment but dropped first all cartels with $5 billion in sales or more and second all cartels with overcharges of 65% or higher; in both cases, r became closer to zero (0.065 and +0.019, respectively), which indicates that extreme observations do not account for the low correlations found. Finally, I examined geographic sub-groups of the cartels: global, U.S., EU and other single national markets. The correlations for these four samples varied from 0.17 to +0.24, none statistically significant. 119. These figures were calculated by Robert Lande and research assistants under his direction in 2004. Less than 1% of all U.S. published court opinions contain both the dollar damages and the affected sales of a cartel. For a discussion of the merits of examining only final verdicts, see Connor and Lande (2005). 120. Guilty pleas and sentencing memoranda of the DOJ and Canadian Competition Bureau almost never mention damages. The EC has fined almost 100 cartels since 1969, but the full decisions are not always published, and only a small number included price data. About 37 EC decisions yielded usable overcharges information. However, the web site of the Italian, French, Korean, and Taiwanese antitrust commissions contains the detail necessary in a large minority of cases. 121. The three analyses reported in this section are somewhat preliminary because factors like publication dates are probably correlated with things like a change in the mix of cartel types. Multiple regression analysis is an appropriate tool to handle such analytical issues, and that is the approach taken in Connor and Bolotova (2006), a research project begun in January 2005 and is scheduled to be completed in April 2006. In footnotes to follow, I will alert the reader to findings from Connor and
Price-Fixing Overcharges
119
Bolotova (2006) that may be different from Connor (2004b). Note that Connor (2004b) employs a slightly smaller data set than that shown in tables above. 122. Several of them have notes to that effect. 123. Controlling for other factors, Connor and Bolotova (2006) find that government reports tended to have systematically lower overcharges, while court and commission decisions were higher. 124. Note that the cartel end date places a lower limit on the publication date, but they are not the same. The results in this sub-section do not contradict the earlier finding that end dates and overcharges are negatively correlated, a finding verified by Connor and Bolotova (2006). 125. Alternatively, one might infer that analysts may have increasingly employed techniques that have won court approval as forensically reliable (see Connor, 2004a). 126. The samples of cartels in each time period overlap, but are not identical. I will correct for changes in the sample immediately below. 127. Note that the mean does not fluctuate over time for the earliest group of cartels, but I regard the mean as less indicative of central tendency than the median. 128. Seventh, ‘‘method unspecified’’ estimates are on average quite close to the before-and-after price method. 129. See Connor (2004a) for such an example. Connor and Bolotova (2006) find no differences between before-and-after estimates and econometric estimates. 130. These two methods seem to be conservative relative to statistical modeling, but the number of pair-wise observations is quite limited. Historical case studies, many by historians with access to original documents, tend to produce lower estimates (Connor & Bolotova, 2006). 131. Yardstick prices are more likely to be available at geographic points close to large centers of supply (concentrations of production or major ports of importation). Public price reporting of products with multiple grades normally is restricted to the most common, least differentiated grade. 132. Average overcharges are those calculated from an entire cartel episode, not just a peak or isolated result. 133. All medians presented in this section incorporate all relevant zero estimates and omit peak results. 134. This study has a majority of episodes and estimates taken from international cartels. 135. Those with positive overcharges. 136. If one assumes that the peak mark-ups are the result of a cartel having achieved something close to monopoly price levels, then the lower average overcharges imply that historical cartels are constrained by substitutes, fear of entry, internal discord, or other factors that frustrate optimization. This is a common finding in studies that measure the degree of monopoly power. 137. All of the relevant estimates in the six works are incorporated in the sample assembled for this paper. 138. In addition, the nine cases that reported peak overcharges produce a median peak overcharge of 71.4% and a mean peak overcharge of 130%. 139. Two jurisdictions I place outside the normal range: two decisions of the Australian authority were exceptionally low (median of 11%), and eight decisions of the Taiwan FTC were abnormally high (median of 68%).
120
JOHN M. CONNOR
140. Two other types (historical case studies and government reports) tended to be low. 141. As an anonymous reviewer suggests, such changes need to be considered alongside appropriate levels for private settlements. These recommendations are particularly complicated by corporate leniency programs and by the joint fining policies of overseas antitrust authorities for international conspiracies. 142. For a variety of factors, however, very few firms actually pay a fine amounting to 20% or more of the amount of commerce affected. Most violators have their fines reduced by 60–80% of the maximums for a variety of reasons. 143. Other terms include monopolies, trade associations, conventions, comptoirs, ententes, and intergovernmental commodity agreements. 144. Until World War I or later, the word ‘‘cartel’’ or Kartell was not in general use among Anglophone economists; Sayous (1902), a French economic historian, discusses 16th and 17th century cartels. Sayous (1902, p. 381) appears to be the first academic writer in a U.S. journal to use the word cartel in its economic sense. He clearly distinguishes private cartels from government-run schemes, trusts, holding companies, and the like. The more famous Dutch East India Company, he argues, was a government-supported monopoly. Sayous believes that a cattle-procurement monopoly by butchers of Anvers, France in the 16th century also qualifies as an early European cartel. Notz (1920, 1929), whose work is discussed below, helped popularize the term in the United States. 145. In a footnote on p. 184, Bullock quotes with approval Jenks observation that trusts and cartels also aim ‘‘to check competition,’’ that is, prevent market entry. 146. However, pools often were organized to obtain only short-run profits, whereas cartel connotes a more enduring scheme. ‘‘Cartel,’’ from the German cognate Kartell, came into general use in British writing in 1902 (Connor, 2001b, p. 20). Although more common in the 19th century, modern cartels do not usually endow a joint venture with capital contributions, though they may set up a sales office or secretariat. The first work in the United States that I have seen referring to German cartels is to ‘‘combinations’’ that ‘‘regulate’’ industries (Bullock, 1901, p. 207). Ripley (1916, p. xiv) cites German kartells. On the continent of Europe, ‘‘syndicate’’ or comptoir was often used to describe a cartel, with a joint sales agency often implied. 147. Other early examples (1908–1915) of convicted cartels based upon patent pooling are paper (1908), electrical equipment (1911), umbrella frames (1907), bicycle coasters (1912–1913), shoe machinery (1914), cash registers (1915), harvesters (1914), and watch cases (1915) (Ripley, 1916, pp. 604–605). 148. The books include a couple of government reports of investigations and proceedings of major conferences. Moreover, there was no sharp distinction between academic journals and serious pieces in intellectual magazines like The Atlantic at the time. Bullock includes one book written in French, but none of the large German literature. 149. Among the earlier post-Bullock monographs in English with significant economic content are books by Liefmann (1897, 1932), Jenks (1900, 1901, 1903), Jenks and Clark (1917/1929), Hirst (1905), Jones (1914, 1921), Levy (1927, 1968), Michels (1928), Seager and Gulick (1929), Domeratzky (1928), Notz (1929), von Beckerath (1930), Piotrowski (1933), and Plummer (1934/1951). Levy (1968), a careful historian,
Price-Fixing Overcharges
121
cites about 30 books on cartels and closely related subjects published before 1927, the great majority in German. 150. The first appears to be Kleinwa¨chter (1883), but this author was not as influential as Liefmann. Hirst (1905) seems to be the first book in English to have Kartell or Cartel in its title. 151. That is, if two or more national cartels are joined by a governmentto-government treaty, the result is not a cartel proper. It is the voluntary nature of the agreement that is the defining characteristic of true cartels, according to Liefmann. This distinction is a useful one for the present survey, because I wish to focus ‘‘private’’ cartels that are indictable under U.S. antitrust law. Private cartels may contain state-owned companies or legal export cartels as members, but if the arrangement is sanctioned by national laws, protected by national sovereignty, or the result of international treaties, I deem them ‘‘public.’’ Compulsory cartels, a type popular in Europe and Japan in the 1930s, are a special type of public cartel. 152. Liefmann (1932) has no doubts that cartels frequently raise prices (or prevent them from falling during recessions), but he is a bit of a perfectionist, insisting that ‘‘y it is impossible to say what the prices would have been if there had been no cartel’’ (p. 104). 153. However, Beckerath undercuts his agnostic position by noting that most cartels have members with varying costs and set their common price so as to allow its highest-cost member to make a profit (p. 265); it follows that at such a price, all the others are making economic profits. 154. This method also may result in an inaccurate benchmark price if the elasticity of demand in the export market differs from that in the domestic market and this difference is not taken into account. In Hirst’s study, however, I judge this factor to be a minor source of inaccuracy, because the export markets (mostly the Benelux countries) were geographically proximate to Germany and were at similar levels of industrial development. 155. Jenks seems to be the originator of the cost-based method of calculating overcharges. The 1921 edition of Jenk’s book received a glowing review by a wellknown cartel economist (Dana, 1922). 156. The Association was still operating successfully in 1888. Jenks judges that the Association had only four limited and brief upward effects on prices; as an exclusive marketing organization, it may have lowered the costs of transportation and selling in the upper Midwest. 157. Other economists with interests in cartels worked at California, Columbia, Cornell, and Stanford universities. 158. Perhaps, the most important U.S. study of cartels to appear in the 1930s was a long monograph on seven international cartels or dominant firms in markets for nonferrous metals: nickel, platinum, aluminum, tin, copper, lead, and zinc (Elliott et al., 1937). This book was the result of a multiyear project by several economists working at Harvard University and Radcliff College. Each cartel study was authored by a different member of the project team. 159. A similar book was edited by Curtis (1931). 160. Sayous (1902, p. 381), a French economic historian, appears to be the first academic writer in a U.S. journal to use the word cartel in its economic sense. He clearly distinguishes private cartels from government-run schemes, trusts, holding
122
JOHN M. CONNOR
companies, and the like. The more famous Dutch East India Company, he argues, was a government-supported monopoly. Sayous believes that a cattle-procurement monopoly by butchers of Anvers, France in the 16th century also qualifies as an early European cartel. 161. However, the government refused repeated appeals by the Company of the North to impose import barriers on whale oil or bone. The Company of the North became weakened by the entry of three other Dutch companies that required a reallocation of market shares and by the growth of the Danish whale-fishing fleet in the 1630s. 162. Liefmann (1932) notes that these numbers do not count hundreds of local German price-fixing agreements among hair dressers, hotels, and other service providers. 163. Notz dwells on private cartels because compulsory cartels were mostly a phenomenon of the 1930s. However, he does briefly mention a phase of the German potash cartel that was nationalized during the Weimar Republic. 164. The United States was not a member of the League of Nations and sent only observers to the 1927 conference. 165. These years bracket what is generally called the Progressive Era in American history. Some historians limit the period to the beginning of the first T. Roosevelt administration in 1901 to the late Wilson administration ca. 1919. 166. Until World War I or later, the word ‘‘cartel’’ or Kartell was not in general use among Anglophone economists; Sayous (1902), a French economic historian, discusses 16th and 17th century cartels. Notz (1920, 1929) helped popularize the term in the United States. 167. The paper contains an intriguing hypothesis about the optimality of price fixing. The cartel’s organizers were well aware that most U.S. pools at the time were ephemeral because most manufacturing processes permitted quick entry, about six months in this industry. To discourage entry, the perpetrators consciously decided to raise prices higher than the monopoly level within a few months. They reasoned that potential entrants would view such unsustainable prices as evidence that the members were irrational and that the pool would quickly crash before the outsiders could start production. This information-obfuscation tactic worked because large-scale entry was thwarted for a year, which allowed the cartel to operate successfully for 19 months, about 12 months longer than if a more moderate pricing policy had been adopted.
ACKNOWLEDGMENTS The author is indebted to Professor Robert H. Lande, who contributed to the Motivation and Conclusions sections; he also was responsible for preparing the material on overcharges from antitrust verdicts in U.S. courts. Anonymous reviewers and this Review’s editor made a large number of constructive suggestions, for which I am grateful. Jeff Zimmerman was of great assistance in proofreading the drafts and rechecking the tables
123
Price-Fixing Overcharges
summarizing the social-science overcharges. Leslie A. Stetler prepared the final manuscript for publication.
REFERENCES Adler, H., & Laing, D. J. (1997). Explosion of international criminal antitrust enforcement. The Corporate Counselor, 11(March), 1. Adler, H., & Laing, D. J. (1999). As corporate fines skyrocket. Business Crimes Bulletin, 6, 1–6. Aiginger, K., et al. (2001). Do American and European industrial organization economists differ? Review of Industrial Organization, 19, 383–405. Allen, G. C. (1923). An eighteenth century combination in the copper-mining industry. Economic Journal, 33, 74–85. Andrews, E. B. (1889). The late copper syndicate. Quarterly Journal of Economics, 3, 508–516. Asch, P., & Seneca, J. J. (1975). Characteristics of collusive firms. Journal of Industrial Economics, 23, 223–237. Asch, P., & Seneca, J. J. (1976). Is collusion profitable? Review of Economics and Statistics, 68, 1–12. Bajari, P., & Ye, L. (2003). Deciding between competition and collusion. Review of Economics and Statistics, 85, 971–989. Baker, J. B. (2003). The case for antitrust enforcement. Journal of Economic Perspectives, 17(Fall), 27–50. Baker, J. B., & Rubinfeld, D. L. (1999). Empirical methods in antitrust litigation. American Law and Economics Review, 1, 386–435. Bane, C. A. (1973). The electrical equipment conspiracies: The treble damage actions. New York: Federal Legal Publications. Barjot, D. (Ed.) (1994). International cartels revisited (1880–1980). Caen, France: EditionsDiffusion du Lys. Benni, M. C. L., et al. (1930). Review of the economic aspects of several international agreements. Geneva: League of Nations. Berge, W. (1944). Cartels: Challenge to a free world (1st ed., 2nd printing). Washington, DC: Public Affairs Press. Block, M. K., & Nold, F. C. (1981). The deterrent effect of antitrust enforcement. Journal of Political Economy, 89, 429–445. Brander, J. A., & Ross, T. W. (2005). Estimating damages from price fixing. Manuscript (May 6, 2005). Brannman, L. E., & Klein, J. D. (1992). The effectiveness and stability of highway bid-rigging. In: Empirical studies in industrial organization: Essays in honor of Leonard W. Weiss: Studies in industrial organization (Vol. 16). Norwell, MA: Kluwer Academic. Briggs, H. (1992). The effects of concentration and cartels on nineteenth-century railroad freight rates. In: Empirical studies in industrial organization: Essays in honor of Leonard W. Weiss: Studies in industrial organization (Vol. 16). Norwell, MA: Kluwer Academic. Bullock, C. J. (1901). Trust literature: A survey and criticism. Quarterly Journal of Economics, 15, 167–217. Bullock, C. J. (1905). Pools, trusts and corporations. Boston: Ginn.
124
JOHN M. CONNOR
Burnside, A. (2003). European cartel enforcement and investigations. Speech at the annual spring meeting of the antitrust section of the American Bar Association. Bush, D., Connor, J. M., Flynn, J. J., Ghosh, S., Grimes, W., Harrington, J. E. Jr., Hawker, N., Lande, R., Shepherd, W. G., & Semeraro, S. (2004). Brief Amici Curiae No. 03-724 in the Supreme Court of the United States, F. Hoffmann-LaRoche et al. v. Empagran et al., March 15. Carlton, D. W., & Perloff, J. M. (2004). Modern industrial organization (4th ed.). Boston: Pearson Addison Wesley. Clarke, J. L., & Evenett, S. J. (2003). The deterrent effects of National Anticartel Laws: Evidence from the international vitamins cartel. Antitrust Bulletin, 48, 289–326. Cohen, M. A., & Scheffman, D. T. (1989). The Antitrust Sentencing Guideline: Is the punishment worth the costs? Journal of Criminal Law, 27, 331–366. Connor, J. M. (2001a). Global price fixing: ‘‘Our customers are the enemy’’. Boston: Kluwer Academic. Connor, J. M. (2001b). ‘‘Our customers are our enemies’’: The lysine cartel of 1992–1995. Review of Industrial Organization, 18, 5–21. Connor, J. M. (2003). Private international cartels: Effectiveness, welfare, and anticartel enforcement (Staff Paper 03-12). West Lafayette, IN: Department of Agricultural Economics, Purdue University, November. Connor, J. M. (2004a). Global cartels redux: The amino acid lysine antitrust litigation. In: J.E. Kwoka, L. White (Eds), The antitrust revolution (4th ed.). Oxford: Oxford University Press. Connor, J. M. (2004b). Price-fixing overcharges: Legal and economic evidence (Staff Paper 04-16). West Lafayette, IN: Department of Agricultural Economics, Purdue University, November. http://agecon.lib.umn.edu/cgi-bin/view.pl Connor, J. M. (2004c). Global antitrust prosecutions of modern international cartels. Journal of International Competition and Trade, 4, 239–267. Connor, J. M. (2005). Optimal deterrence and private international cartels. Paper presented at the Second Biennial Conference of the Food System Research Group, University of Wisconsin-Madison, June 16–17. Connor, J. M., & Bolotova, Y. (2006). A meta-analysis of cartel overcharges. International Journal of Industrial Organization, 24, 1109–1137. Connor, J. M., & Lande, R. H. (2005). How high do cartels raise prices? Implications for Reform of Antitrust Sentencing Guidelines. Tulane Law Review, 80(December), 513–570. Crandall, R. W., & Winston, C. (2003). Does antitrust policy improve consumer welfare? Assessing the evidence. Journal of Economic Perspectives, 17(Fall), 1–26. Curtis, R. E. (1931). Trusts and economic control: A book of materials. New York: McGraw-Hill. Daggett, R. S., & Freedman, D. A. (1985). Econometrics and the law: A case study in the proof of antitrust damages. In: L. M. Le Cam & R. A. Olshen (Eds), Proceedings of the Berkeley conference in honor of Jerzy Neyman and Jack Kiefer, 1. New York: Wadsworth. Dana, D. E. (1992). Jones’ The Trust Problem in the United States. Quarterly Journal of Economics, 28(November), 175–182. De Roos, N. (2006). Examining models of collusion: The market for lysine. International Journal of Industrial Organization, 24, 1083–1107. De Rousiers, P. (1927). Cartels trusts and their development. Geneva: League of Nations. Demaree, A. T. (1969). How judgment came for the plumbing conspirators. Fortune, (December), 96–99, 175–182.
Price-Fixing Overcharges
125
Denger, M. L. (2003). Too much or too little, a summary of discussion. Washington, DC: American Bar Association’s Antitrust Remedies Forum, April. Dick, A. R. (1996). When are cartels stable contracts? Journal of Law and Economics, 39, 241–283. Domeratzky, L. (1928). The International Cartel Movement: Trade information bulletin no. 556. Washington, DC: U.S. Department of Commerce. Eckbo, P. L. (1976). The future of world oil. Cambridge, MA: Ballinger. Economic and Financial Section. (1927). Electrical industry. Geneva: League of Nations. Edgerton, C. E. (1897). The Wire-Nail Association of 1895–96. Political Science Quarterly, 12, 246–272. Edwards, C. (1946). Cartelization in Western Europe. Washington, DC: Bureau of Intelligence and Research, U.S. Department of State. Eichenwald, K. (2000). The informant: A true story. New York: Broadway Books. Elliott, W. Y., et al. (1937). International control in the non-ferrous metals. New York: Macmillan. Ellison, G. (1994). Theories of cartel stability and the Joint Executive Committee. RAND Journal of Economics, 25, 37–57. Elzinga, K. G. (1984). New developments on the cartel front. Antitrust Bulletin, 29(Spring), 3–26. Epstein, B., & Newfarmer, R. S. (1980). International Electric Association: A continuing cartel. Report to the Subcommittee on Oversight and Investigations, House Committee on Interstate and Foreign Commerce, U.S. Congress. Washington, DC: USGPO. Finkelstein, M., & Levenback, H. (1983). Regression estimates of damages in price-fixing cases. Law and Contemporary Problems, 46, 145–169. Fisher, F. (1980). Multiple regression in legal proceedings. Columbia Law Review, 80, 702. Fraas, A. G., & Greer, D. F. (1977). Market structure and price collusion: An empirical analysis. Journal of Industrial Economics, 26, 21–44. Froeb, L. M., Koyak, R. A., & Werden, G. J. (1993). What is the effect of bid-rigging on prices? Economic Letters, 42, 419–423. Goyder, D. G. (1998). EC competition law (3rd ed.). Oxford: Clarendon. Graubert, J. D. (2003). Too much or too little, a summary of discussion. Washington, DC: American Bar Association’s Antitrust Remedies Forum, April. Griffin, J. M. (1989). Previous cartel experience: Any lessons for OPEC? In: L. R. Klein & J. Marquez (Eds), Economics in theory and practice: An eclectic approach. Boston: Kluwer Academic. Grossman, P. Z. (1996). The dynamics of a stable cartel: The railroad express, 1851–1913. Economic Inquiry, 34, 220–236. Harding, C., & Joshua, J. (2003). Regulating cartels in Europe: A study of legal control of corporate delinquency. Oxford: Oxford University Press. Hausman, W. J. (1980). A model of the London coal trade in the eighteenth century. Quarterly Journal of Economics, 94, 1–14. Hausman, W. J. (1984). Cheap coals or limitation of the vend? The London coal trade, 1770– 1845. Journal of Economic History, 44, 321–328. Hay, G. A., & Kelley, D. (1974). An empirical survey of price-fixing conspiracies. Journal of Law and Economics, 17, 13–38. Herling, J. (1962). The great price conspiracy: The story of the antitrust violations of the electrical industry. Washington, DC: Robert D. Luce. Hexner, E. (with the assistance of Adelaide Walters). (1946). International cartels. Chapel Hill, NC: University of North Carolina Press.
126
JOHN M. CONNOR
Hirst, F. W. (1905). Monopolies, trusts, and kartells. London: Methuen. Hovenkamp, H. (1999). Federal antitrust policy (2nd ed.). St. Paul: WestGroup. Howard, J. H., & Kaserman, D. (1989). Proof of damages in construction industry bid-rigging cases. Antitrust Bulletin, 34, 359–393. Hudson, H. (1890). The Southern Railway and Steamship Association. Quarterly Journal of Economics, 5, 70–94. Jenks, J. W. (1888). The Michigan Salt Association. Political Science Quarterly, 3, 78–98. Jenks, J. W. (1900/1901/1903). The trust problem (three editions 1900, 1901, 1903). Garden City, NY: Doubleday, Page. Jenks, J. W., & Clark, W. E. (1917/1929). The trust problem (4th and 5th editions 1917, 1929). Garden City, NY: Doubleday, Doran. Jones, E. (1914). The anthracite coal combination in the United States. Cambridge: Harvard. Jones, E. (1921). The trust problem in the United States. New York: Macmillan. Klawiter, D. (2001). After the deluge: The powerful effect of substantial criminal fines, imprisonment, and other penalties in the age of international criminal enforcement. George Washington Law Review, 69, 745–765. Kleinwa¨chter, F. (1883). Die Kartelle. Austria: Innsbruck. Kovacic, W. E., & Shapiro, C. (2000). Antitrust policy: A century of economic and legal thinking. Journal of Economic Perspectives, 14, 43–60. Kuhlman, J. M. (1972). Theoretical issues in the estimation of damages in private antitrust actions. Antitrust Bulletin, 17, 548–558. Kwoka, J. E. (1977). Pricing under federal milk market regulation. Economic Inquiry, 5, 367–384. Kwoka, J. E. (2003). The attack on antitrust policy and consumer welfare: A response to Crandall and Winston (Working Paper 03-008). Boston: Department of Economics, Northeastern University, March. Landes, W. H. (1983). Optimal sanctions for antitrust violations. University of Chicago Law Review, 50, 652–678. Lanzillotti, R. F. (1996). The great school milk conspiracies of the 1980s. Review of Industrial Organization, 11, 413–458. Lean, D. F., Ogur, J. D., & Rodgers, R. P. (1985). Does collusion pay y does antitrust work? Southern Economic Journal, 51, 828–841. Lee, I. K. (1999). Non-cooperative tactic collusion, complementary bidding and incumbency premium. Review of Industrial Organization, 15, 115–134. Levenstein, M. C. (1997). Price wars and the stability of collusion: A study of the pre-world war I bromine industry. Journal of Industrial Economics, 45, 117–137. Levenstein, M. C., & Suslow, V. Y. (2006). What determines cartel success? Journal of Economic Literature, 44, 43–95. Levenstein, M., & Suslow, V. (2004). The changing international status of export cartel exemptions (Working Paper), November. http://ssrn.com/abstract=618201 Levy, H. (1968). Monopoly, cartels and trusts in British industry (2nd ed.). London: Cass & Co. (1927, reprinted 1968). Lieber, J. B. (2000). Rats in the grain: The dirty tricks and trials of Archer Daniel Midland. New York: Four Walls Eight Windows. Liefmann, R. (1897). Die Unternehmerverbande, Konventionen, Kartelle. Liefmann, R. (1932). Cartels, concerns and trusts. New York: Dutton. Low, D. M. (2004). Cartel enforcement, immunity, and jurisdiction: Some recent Canadian developments. Speech at the International Bar Association, Rome, Italy, May 17–18.
Price-Fixing Overcharges
127
MacGregor, D. H. (1927). International cartels. Geneva: League of Nations. Marlio, L. (1947). The aluminum cartel. Washington, DC: Brookings Institution. Martin, S. (1994). Industrial economics (2nd ed.). New York: Macmillan. Michels, R. K. (1928). Cartels, combines and trusts in post-war Germany. New York: Columbia University Press. Notz, W. F. (1920). International private agreements in the form of cartels, syndicates, and other combinations. Journal of Political Economy, 28, 658–679. Notz, W. F. (1929). Representative international cartels, combines, and trusts: Trade promotion series no. 81. Washington, DC: U.S. Department of Commerce. OECD. Competition law and policy: 2002–2003 annual reports. http://www.oecd.org/document/ 49/0,2340,en_2649_34715_2509617_1_1_1_1,00.html (National antitrust authorities’ annual reports appended; site goes back to 1994–1995). Oualid, W. (1938). International raw commodity price control. Paris: International Studies Conference, League of Nations. Pesendorfer, M. (2000). A study of collusion in first-price auctions. Review of Economic Studies, 67, 381–411. Peterson, E. B., & Connor, J. M. (1995). A comparison of welfare loss estimates for U.S. food manufacturing. American Journal of Agricultural Economics, 77(May), 300–308. Piotrowski, R., O’Brien, D. P., Maunder, W. P. J., & Howe, W. S. (1933). Cartels and trusts. London: George Allen & Unwin. Plummer, A. (1934/1951). International combines in modern industry. London: Pittman. Porter, R. H. (1983). A study of cartel stability: The Joint Executive Committee, 1880–1886. Bell Journal of Economics and Management, 14, 301–314. Porter, R. H. (2001). Detecting collusion among bidders in auction markets. In: Fighting cartels – Why and how? Stockholm: Konkurrensverket. Porter, R. H., & Zona, J. D. (1999). Ohio school milk markets: An analysis of bidding. RAND Journal of Economics, 30, 263–288. Posner, R. A. (1975). The social costs of monopoly and regulation. Journal of Political Economy, 83, 807–828. Posner, R. A. (1976). Antitrust law: An economic perspective. Chicago: University of Chicago Press. Posner, R. A. (1979). The Chicago School of Antitrust Analysis. University of Pennsylvania Law Review, 127, 925–948. Posner, R. A. (2001). Antitrust law (2nd ed.). Chicago: University of Chicago Press. Pribram, K. (1935). Cartel problems. Washington, DC: Brookings. Pyndyck, R. S. (1979). The cartelization of world commodity markets. American Economic Review, 69, 154–158. Ripley, W. Z. (Ed.) (1916). Trusts, pools and corporations: Revised edition. Boston: Ginn. Sayous, A. E. (1902). Cartels and trusts in Holland in the seventeenth century. Political Science Quarterly, 17, 381–393. Seager, H. R., & Gulick, C. A. (1929). Trust and corporation problems (second printing). New York: Harper. Smith, R. A. (1963). Corporations in crisis. Garden City, NY: Doubleday. Spratling, G. R. (2001). Detection and deterrence. George Washington Law Review, 69, 817–823. Stevens, W. S. (1912a). The powder trust, 1872–1912. Quarterly Journal of Economics, 26, 444–481.
128
JOHN M. CONNOR
Stevens, W. S. (1912b). A group of trusts and combinations. Quarterly Journal of Economics, 26, 593–644. Stocking, G. W., & Watkins, M. W. (1946). Cartels in action. New York: Twentieth Century Fund. Stocking, G. W., & Watkins, M. W. (1948). Cartels or competition? New York: Twentieth Century Fund. Sullivan, L. A., & Grimes, W. S. (2000). The law of antitrust: An integrated handbook. St. Paul, MN: West Group. Sultan, R. G. M. (1974/1975). Pricing in the electrical oligopoly (Vols. I and II). Boston: Harvard Business School. Suslow, V. Y. (2001). Cartel contract duration: Empirical evidence from international cartels. Unpublished manuscript that updates a 1988 Hoover Institution Working Paper. University of Michigan Business School, October. Swann, D., O’Brien, D. P., Maunder, W. P. J., & Howe, W. S. (1974). Competition in British industry: Restrictive practices legislation in theory and practice. London: Allen & Unwin. Symeonidis, G. (2002). The effects of competition: Cartel policy and the evolution of strategy and structure in British industry. Cambridge: MIT Press. Tosdal, H. R. (1916). The German potash syndicate. In: W. Z. Ripley (Ed.), Trusts, pools and corporations: Revised edition (pp. 795–832). Boston: Ginn. U.S. Congress (1938–1940). Hearings of the temporary National Economic Committee (Vols. 1–11). Washington, DC: U.S. Government Printing Office. U.S. Congress. (1965). Staff study of income tax treatment of treble damages payments under the antitrust laws. Washington, DC: Joint Committee on Taxation, U.S. Government Printing Office. Uctum, M. (1998). Why have corporate profits declined? An international comparison. Review of International Economics, 6, 234–251. USSG. (1986). Unpublished public hearings. Washington, DC: United States Sentencing Commission, July 15. USSG Advisory Group. (2003). Report of the ad hoc advisory group on the organizational guidelines. Washington, DC: U.S. Sentencing Commission, October 7. von Beckerath, H. (1930). Modern industrial organization: An economic interpretation. New York: McGraw-Hill (1930, translated 1933). Walker, F. (1906). The German steel syndicate. Quarterly Journal of Economics, 20, 353–398. [Reprinted as in Ripley, W. Z. (Ed.). (1916). Trusts, pools and corporations: Revised edition (pp. 833–868). Boston: Ginn.] Wallace, B. B., & Edminster, L. R. (1930). International control of raw materials. Washington: Brookings. Wells, W. (2002). Antitrust and the formation of the postwar world. New York: Columbia University Press. Werden, G. J. (2003). The effect of antitrust policy on consumer welfare: What Crandall and Winston overlook, EAG 03-2. Washington, DC: Economic Analysis Group, Antitrust Division, U.S. Department of Justice, January. Werden, G. J. (2004). Economic evidence on the existence of collusion: Reconciling antitrust law with oligopoly theory. Antitrust Law Journal, 71, 719–800. White, L. J. (Ed.) (1988). Private antitrust litigation: New evidence, new learning. Cambridge, MA: MIT Press. Wiedenfeld, K. (1927). Cartels and combines. Geneva: League of Nations.
Price-Fixing Overcharges
129
Wils, W. P. J. (2005). Is criminalization of EU competition law the answer? World Competition, 28(June), 117–159. Zimmerman, J. E. (2005). Determinants of cartel duration: A Cross-sectional study of modern private international cartels. M.S. Thesis, Purdue University, May.
APPENDIX. THE EARLY LITERATURE ON CARTELS Interest in collusive organizations began well before industrial-organization economics was recognized as a distinct discipline. Prior to World War II, fewer than a dozen archival articles treat the economics of cartels, but scores of books were published on the economic and political aspects of ‘‘pools,’’ ‘‘trusts,’’ ‘‘combines,’’ ‘‘syndicates,’’ and all the other terms that were used at the time to encompass various monopolistic business arrangements.143 Consistent use of these terms was not well established until the mid 20th century.144 Bullock’s (1901) seminal paper tends to regard all of them as roughly equivalent terms for monopolistic business entities with market power over price (p. 183).145 By 1916, Ripley could differentiate these phenomena using terms that became commonly accepted jargon by the 1940s. Pools or corners were contractual joint-profit-increasing agreements by independent sellers over prices or quantities; today these are called cartels (Ripley, 1916, p. xiv).146 Ripley cites the U.S. cordage cartel, formed in 1860, as the first documented U.S. pool. Other 19th century cartels include cotton bags, distilling, iron pipes, steel, salt (Jenks, 1888), wire nails (Edgerton, 1897), and a patent pool for porcelain bathtubs.147 Trusts proper were legal instruments used in the United States from about 1879 to 1902 for combining companies under a single board of directors; beginning in the late 1890s, trusts were supplanted as a means of industrial merger by the holding company (Ripley, 1916). Thus, trusts, combines, and holding companies refer more to the outcomes of mergers and acquisitions than to cartels. Yet, the word ‘‘trust’’ was used loosely and popularly throughout the early 20th century to cover both cartels and mergers intended to increase market power. As late as the 1930s, several terms were often used interchangeably for cartels (Plummer, 1936; Curtis, 1931). Curtis considered cartel to be a term used mainly in Europe. His preferred terminology was pools for more informal and unstable cartels and trusts for cartels with strong central direction and control. Bullock (1901), a professional economist and author of an early American economics textbook, wrote the first English-language survey of cartels and trusts in the social-science literature. After noting that there was a near
130
JOHN M. CONNOR
absence of publications on the topic during 1890–1896, he finds an astonishing outpouring of 34 books and 48 serious articles in 1897–1900.148 Interest in the subject continued in the early 20th century, with most of the cartel literature from 1900 to 1940 appearing in books.149 Some of these works were written by historians and others by some of the earliest practitioners of the emerging field of industrial economics. Most of these studies contain little or no quantitative data. Bullock opines that the quantitative measurement of the market-price effects of cartels and trusts is not possible. Liefmann (1897) published one of the first economic monographs that contained the word Kartell in its title.150 The book appeared in five editions in German from 1897 to 1929. The last edition was updated, translated into English, and published in London in 1932; the Oxford University economist who wrote the book’s Introduction hailed it as the best known study of cartels and trusts ‘‘from a German perspective.’’ In many ways, Liefmann was leagues ahead of his contemporaries in the analysis of the cartel phenomenon. He coined one of the most cited and pithy definitions of cartels: ‘‘free [voluntary] associations of producers for the monopolistic control of the market’’ (Liefmann, 1932, p. ix). By this definition, he meant to include only arrangements by independent companies linked by formal or informal contractual agreements; compulsory commodity schemes enforced by government decrees or parliamentary statutes are not true cartels by his definition, though international agreements negotiated between compulsory national cartels would qualify if the negotiated agreement did not require statutory enforcement.151 He dismisses the widely accepted view of the time that cartelists are merely aiming to achieve a ‘‘reasonable profit,’’ insisting that cartels are instruments for maximizing profits. Liefmann assembles a great deal of information on German cartels and limited information on cartels outside Germany, but with one exception, he includes no useful price series that could be used to compute price effects.152 Liefmann’s positions continued to influence German economists for decades to come. However, von Beckerath (1930) opined that cartels were motivated primarily by a desire to reduce fluctuations in output or prices. To do so, durable cartels typically used their power to raise prices during slumps and restrain prices during booms. While he admits that raw-material cartels and patent pools were successful in raising prices above competitive levels in the long run, he believed that for other types, the evidence was lacking (p. 262). ‘‘y [I]t can only rarely be proved that a cartel is the only reason behind a price rise’’ (p. 263).153 Indeed, the book contains no price data. Herman Levy was a contemporary of Liefmann. Levy was a prolific writer of books on economic history. Not counting revised editions, he
Price-Fixing Overcharges
131
authored 10 books between 1900 and 1927, eight in German and two in English. While indebted to Liefmann’s concepts and definitions, Levy covers different ground than Liefmann. Unlike Liefmann, Levy is eager to quantify the economic impacts of cartels and trusts. Levy (1968) is a reprint of the second (1927) English-language edition of his book on British cartels, monopolies, and oligopolies. This work is concerned about why the British cartel movement was weaker and slower to develop than on the Continent of Europe. It contains unique information on 18th and 19th century British cartels. Another early European writer who was concerned about the lack of concrete measures of market power is a then young lawyer and economics lecturer, Hirst (1905). His book grew out of an 1899 Oxford essay that attempted to develop price-based indicators of the price effects of cartels. Noting that German cartels frequently exported significant shares of domestic output to other countries at lower prices than their fixed domestic prices, he proposes using the export prices as a yardstick. Although there is some danger of overstating the domestic overcharge if the cartel is dumping product at predatory prices or if the marginal costs of exporting are lower than comparable domestic sales, he applies this method to six German cartels using 1900–1902 prices.154 Jeremiah W. Jenks was a political science professor at Cornell University in 1900 when the first of his five editions of The Trust Problem was published, though he had already been researching pools, trusts, and monopolies for 20 years by that time.155 Jenk’s 1888 study of the Michigan salt cartel seems to be the first economic study of cartels to appear in a peer-reviewed professional journal.156 His publications display a strong empirical bent and show a deep interest in gauging the economic effects of cartels. Unusual among academics of the time, his commitment to the study of trusts seems to have been cemented by his extensive work as an advisor for the U.S. Industrial Commission, which held a series of public hearings in 1898–1899 on conditions in several oligopolistic industries. His books (1900, 1901, 1903, 1917, 1929) contain carefully constructed series of wholesale prices for refined sugar, whiskey, wire nails, barbed wire, steel, and other products controlled by cartels or dominant firms. Among his analytical advances was the creation of coterminous price series for the principal inputs for the final products (corn for whiskey, steel for nails, etc.). By correcting for changes in product prices due to input prices, he was able to determine more precisely when and how strongly prices were affected by a cartel. Harvard University seems to have been the leading campus for economic and legal studies of cartels in the early 20th century.157 One indication of its
132
JOHN M. CONNOR
preeminence is the publication of what is probably the first textbook on cartels, mergers, and monopolies in 1905.158 The revised edition is a huge (872 pages of small print) compilation of reprints from professional journals of law and economics, excerpts from briefs and court decisions, and legal commentary (Ripley, 1916).159 Ripley aimed at applying the case-study method pioneered by Harvard Law School into advanced economics courses. Eliot Jones wrote a Ph.D. dissertation at Harvard University on several episodes from 1871 to 1914 of cartelization of the U.S. anthracite coal industry, the largest U.S. mineral industry of the early 20th century. His dissertation won a University prize and was published by Harvard University Press in 1914. Jones’ first book is for its time one of the best analyses of the economic history, market structure, collusive conduct, and price effects in any industry. It may be one of the first books to combine an empirical interest in industrial concentration with attention to the antitrust laws. In addition to detailed ownership and price data from industry trade sources, Jones had available testimony and exhibits from one of the early U.S. antitrust trials. This industry case study illustrated how a concentrated, technologically dynamic industry with extensive network economies, the railroads, could leverage its market power in transportation through backward vertical integration and collusion in the coal-mining industry; after the Sherman Act was passed, the railroads adopted new strategies (mergers, cross-ownership, and interlocking directorships) to maintain their market power in coal. Along with papers in the Quarterly Journal of Economics, his writings received extensive peer review that was unusual for the period. Jones’ interest in competition and antitrust laws was extended in his 1921 book. Jones was a contemporary of Jenks, but better versed in the stillemerging concepts of industrial-organization economics. Despite his evident interest in the price effects of cartels, in his second book, quantitative data were presented on price effects for only three cartels. An issue among European writers is when and why kartells first appeared. Piotrowski (1933) delves into pre-Christian, Roman, and medieval history to find many examples of organizations that appear to resemble private cartels, but in most cases, details about their conduct and the degree of government support are lacking. However, Sayous (1902) makes a welldocumented case for the existence of cartels in the strict sense of the term in 17th century Holland.160 The Dutch Company of the North was chartered in 1614 to exploit the Greenland whale-oil industry; by 1618, the Company had adopted a supply-restraint objective to keep domestic prices above competitive levels, and by 1622, the States-General of Holland had granted it a long-lasting monopoly for whale-fishing.161
Price-Fixing Overcharges
133
Nevertheless, it is Germany that has the best claim as the birthplace of contemporary industrial cartels. Liefmann (1932) believes that the first domestic German cartel was the Neckar Salt Union, an 1829 combination of salt mines in three German states. Five more were formed prior to 1870. However, Liefmann and other writers point to the German depression of the mid 1870s as a peak for cartel formation. A 1905 German government survey found 385 industrial cartels operating; the number rose to 3,000 by 1925.162 As for international cartels, he identifies the 1867 merger of the Neckar Salt Union in Germany with the Eastern French Salt Works Syndicate as the first of its kind. By 1897, there were at least 40 international cartels with German companies as members, most of them in chemical or nonmetallic mineral product markets. Notz (1920) quotes a German book that found 114 international cartels in 1912; by 1920, he could identify 11 international cartels with participation of U.S. companies. The 1870s were also a formative period for U.S. cartels. Seager and Gulick (1929) trace the earliest documented U.S. pools to the cordage industry, which began making agreements on prices at least as early as 1861, but cordage manufacturers did not begin a formal association until 1878. The Michigan Salt Association, formed in January 1876, may be the first recorded formal U.S. cartel. Because of the high costs of transporting salt, an elaborate organizational structure, and the highly inelastic demand for salt, this cartel was successful in dominating the Midwest market for 25 years. Two lengthy reports from analysts in the U.S. Department of Commerce presage the triumph of the more precise German usage of the term cartel (Domeratzky, 1928; Notz, 1929). Notz (1929) accepts Liefmann’s classic definition of a private cartel: a voluntary association of two or more independent business organizations in the same line of business with the aim of increasing joint profits by controlling markets or reducing competition.163 Essential is an overt agreement to divide market territories, set or stabilize prices, limit or allocate industry supply, establish a common sales agency, pool intellectual property, or some combination of these five strategies. If the organizations are registered in at least two countries, then it is an international cartel. While the Department of Commerce reports are strong in detailing cartel membership and industry supply conditions, they have little to offer by way of price effects. Cartels, mergers, trade, and foreign direct investment were major concerns of the League of Nations, which sponsored a major conference on the subjects in 1927. Papers prepared by some of the leading European cartel scholars of the day were published as part of the conference proceedings (de Rousiers, 1927; MacGregor, 1927; Wiedenfeld, 1927; Economic and
134
JOHN M. CONNOR
Financial Section, 1927).164 These papers dwell on conceptual and organizational issues surrounding cartels and contain little of interest on price or welfare impacts. Indeed, the near absence of empirical detail in these reports and other studies by European scholars active in the interwar period provide a striking contrast with the industrial analyses emerging in the United States. The final report of the 1927 conference revealed a deep split between those participants who believed that cartels harmed national economies and international trade and those who believed that cartels stabilized prices, investment, and employment. Perhaps to rectify these ambiguities, the League later sponsored cartel studies with more empirical content (Benni et al., 1930; Oualid, 1938). Relatively few books were written about cartels in the 1930s, a period during which antitrust was in eclipse in the United States and cartels took on distinctly political roles as tools of economic planning in Europe and Japan. In this decade, cartels were often embraced because they were perceived as antidotes to the worldwide depression and, in some industries, deflation. Indeed, the Brookings Institution sponsored a series of books during this time to assist policy makers in implementing the National Recovery Act (e.g., Pribram, 1935). U.S. Supreme Court decisions quickly restored the antitrust laws by 1938 (Wells, 2002). When President Roosevelt and his advisors became apprised of the intimate connections between national socialism and compulsory cartels in Germany in the 1930s, they rejected using cartels to foster economic recovery. Although most books written prior to 1945 lacked empirical analyses of cartel performance, a small number of U.S. economists published a few welldocumented case studies of price effects. Many were written during the heady times (1885–1920) during which state and federal antitrust laws were being debated and first enforced, though none of these works suggested that their approaches had forensic value.165 Among the most useful papers for overcharges are Jenks (1888), Andrews (1889), Edgerton (1897), Hudson (1890), Walker (1906), Stevens (1912a, 1912b), Tosdal (1916), and Allen (1923). Jenks’s study of the Michigan Salt Association of the 1880s is a classic example of a well-researched history of the methods used by a mining cartel to control a market that incorporates substantial information on costs and prices.166 Edgerton’s (1897) paper on the U.S. Wire Nail Association is a superb analysis of the evolution, operation, and price effects of a short-lived but tightly structured, highly effective manufacturers’ cartel which was written with the help of insider interviews just a year after the cartel dissolved. This study is notable because the conspiracy is the first U.S. work on
Price-Fixing Overcharges
135
a U.S.-based international conspiracy.167 Andrews (1889) drew upon contemporary business publications to recount what is quite possibly the world’s first global cartel, the infamously scandalous Secre´tan copper syndicate of 1887–1889. Stevens’ (1912a, 1912b) study of the gunpowder trust is notable for focusing on what was believed to be the longest-running discovered cartel in the Nation’s history; Stevens carefully delineated three distinct phases of the cartel, and he drew upon the records of a 1911 antitrust trial to document the final episode. Tosdal (1916) and Walker (1906) provide competent analyses of the earlier episodes of two highly durable domestic German cartels, potash and steel, respectively; subsequent scholars have repeatedly returned to these cases. Ripley (1916) reprints a fascinating court decision of the U.S. enameled bathtub cartel, which used patent licenses on a new machine to achieve effective collusion. Allen’s (1923) account of the 18th century English copper-smelting cartel is the first quantitative assessment of cartel effectiveness by a European economist to appear in a peer-reviewed academic journal. The absence of cartel studies in professional journals in the 1920s and 1930s is striking.
Name
Alphabetic List of Cartelized Markets. Characteristics
International Location
Bid Rigging
Found Guilty, Liable for Civil Penalties, or ExtraLegalc
Number of Episodes
Average Observations
1+ 2 1 6
1 2 2 28
172 235 205 18
US X EUR US X EUR
EC fines Legal cartel US consent decree
199 276 82
X INTL X JP US
JFTC probe US guilty pleas
1 1 1
1 1 2
204 7 42 53
US US X US+UK US
X X X
US settlement Jury trial decision US pleas, EU fines US trial
1 1 1 1
1 1 6 1
52
US
X
Civil settlement
1
2
25
US
1
2
1
1
115
X FR
France, fines
JOHN M. CONNOR
Airlines, US passenger Air routes, Danish Almonds, US and export Aluminum, metal (interwar) Aluminum, metal (1990s) Aluminum foil, JP Aluminum phosphide, US Asphalt, Alabama, US Asphalt, Oklahoma, US Auction houses, fine art Auctions, houses in DC, US Auctions, used police cars, NY City Automobile manufacture, US Ball and roller bearings, France
Code No.
136
Table A1
X DE
EU fines
9
9
63 239 251 167 45 252 264
UK US KO UK US KO US
Legal cartel US trial KFTC fines Legal cartel US trial KFTC fines US trial
1 1 1 1 1 1 1
1 1 1 1 1 1 1
153 37 244 6 246 59
US US China X US X INTL UK
US Appeals Court China AMB fines US guilty pleas US guilty pleas Legal cartel
1 1 1 3 1 1
1 1 1 4 1 1
243
Taiwan
TWFTC fines
1
1
124 188 152 202 198 203
DE US US US US X US
1 2 1 1 1 1
1 2 1 1 1 2
39 142
X EUR US
Germany, fines Legal cartel Legal export cartel US civil settlement US investigation US FTC and civil trial EU fines US trials
1 1
1 8
X
X
137
216
Price-Fixing Overcharges
Banks, Euro-Zone fees, DE and NL Bath tubs, iron, UK Bath tubs, enameled, US Batteries, auto, Korea Bedsteads, metal, UK Beef in California, US Beer brewing, Korea Blueberries, wild, purchases, ME Bond underwriting, US Bread, white pan, US Bricks, China Bromine, US Bromine Cable, rubber and plastic, UK Cable TV operators, Taiwan Cable, power, Germany Carbon, arc lighting, US Carbon black, US exports Carbon dioxide, US Carbon fiber, US Cardizem heart medicine, US Carton board, EU Cartons, corrugated, US
Name
Characteristics International Location
Bid Rigging
Found Guilty, Liable for Civil Penalties, or ExtraLegalc
Number of Episodes
Average Observations
224
US
US trial
1
1
62
UK
Legal cartel
1
1
271
US
Jury trial
1
1
212 70 106 277 144 26 76 81 155 160
NO ZA X DE X ROM US US X INTL X INTL DE US
1 1 1 1 1 1 1 1 2 10
1 2 1 1 1 1 7 4 7 16
X
Legal cartel X X X
Germany, fines Romania fines US trial and decree US pleas, EU fines US jury trial Legal cartel US trial
JOHN M. CONNOR
Carpets, polypropylene, US Carpets, woven, UK Cathode ray tubes (see electronic radio and TV tubes) Cattle, fed, US Cell phones (see telephone) Cement, Norway Cement, South Africa Cement, Germany Cement, Romania Chicken, US Cigarettes, US Citric acid Choline chloride Coal, Ruhr, Germany Coal, anthracite, eastern US
Code No.
138
(Continued)
Table A1.
AU UK
6 5
9 17
Coconut oil, Philippines Coffee, Hungary Coke Concrete, Denmark Concrete, Germany Construction and other industries, US Construction and procurement, JP Construction, concrete, NY, US Construction, 8,000 buildings, Germany Construction, electric wiring contractors, Denmark Construction, Normandy Bridge, FR Construction, electrical, France Construction, university, France Construction, roads, Colorado, US Construction, roads, France
206 248 147 51 114 196
PL X HU X EUR DK X DE US
1 1 1 1 1 1
1 1 2 1 1 1
X X X
Germany, fines US convictions
213
JP
X
JFTC actions
21
5
261
US
X
US trial
1
1
174
DE
X
Germany fines
1
1
122
DK
X
Denmark, fines
1
1
247
FR
X
French fines
1
1
175
FR
X
1
1
176
FR
X
1
1
222
US
X
France consent decree France consent decree Civil settlement
1
2
177
FR
X
1
1
Parliamentary inquiry Legal cartel Hungary, fines Legal export cartel
France consent decree
139
179 166
Price-Fixing Overcharges
Coal, black, Australia Coal, Newcastle, England
Name
Characteristics International Location
Bid Rigging
Found Guilty, Liable for Civil Penalties, or ExtraLegalc Korea FTC fines
193
KO
X
211
US
X
163
JP
X
162
JP
108 107 161 32
Number of Episodes
Average Observations
1
1
1
2
Japan trial
1
1
X
JFTC fines
1
1
X NL
X
Netherlands, fines
1+
1
X NO JP
X X
Norway, probe A few civil actions
1+ 1+
1 2
KO
X
Korea, fines
1+
1
Korea FTC fines
1
1
278
X KO
101
X EGY
X
US trial
1
1
255
US
X
US trial
1
1
JOHN M. CONNOR
Construction, roads, Korea Construction, roads, seal coating, US Construction, kitchen, Japan Construction, US Navy shipyard, Japan Construction, Netherlands Construction, Norway Construction, public, Japan Construction projects, Korea Construction machinery manufacturing, Korea Construction, USAID in Egypt Construction, roads, Colorado, US
Code No.
140
(Continued)
Table A1.
US
X
Trials, settlements
1+
1
123
DE
X
Germany, fines
1+
1
260
US
X
US trial
3
3
34
US
X
Trials, settlements
1+
2
195 245
US China
X X
Trials, settlements China AMB fine
1+ 1
1 1
33 22 88 225 54 71
US X US+INTL X INTL UK US X ZA
X
Trial
1+ 9 1 4 1 1
3 30 1 4 1 1
229
TW
TFTC fines
2
2
164
JP
Japan trial
1
1
159 263
X EUR US
Legal cartel US trial
1 1
2 1
21 189
X EUR US
1 1
15 1
US, EU Probes X US consent decree Legal cartel
X
141
1
Price-Fixing Overcharges
Construction, roads, Florida, US Construction, roads, Germany Construction, roads, NY, US Construction, roads, SD and NC, US Construction, roads, US Construction, school building, China Construction, sewers, US Copper metal Copper concentrate Copper smelters, UK Dairy processing, US Diamonds, gem, South Africa Distributors, natural gas, TW Dredging, river, Japan Drugs (see pharmaceuticals) Dyestuffs Education, bar review preparation, GA Electric light bulbs Electric light bulbs, US
Name
Characteristics International Location
Bid Rigging
Found Guilty, Liable for Civil Penalties, or ExtraLegalc
184 61 60 249 48
UK UK UK IL US
116
X NO
X
UK Commission Legal cartel Legal cartel Israeli fines, prison US pleas, settlements Norway, fines
129 183
X EUR UK
X X
267
US
X
192
X UK
98 –
X US
121
AU
X
X
X
Number of Episodes
Average Observations
1 1 1 1 5
5 1 1 1 18
1+
1
US conviction UK Commission
2 1
3 4
US trial
1
1
UK Commission
1
5
US guilty pleas
1
1
Australia, fines
1
1
JOHN M. CONNOR
Electric light bulbs, UK Electric meters, UK Electric motors, UK Electric pipes, Israel Electric power equipments, US Electric power equipments, Norway Electric power equipment Electric power equipments, UK Electrical subcontracting, GA, US Electronic radio and TV tubes, UK Explosives, US Fertilizer (see nitrogen, phosphate, potash) Fire protection installation, AU
Code No.
142
(Continued)
Table A1.
X US TW X EUR
US. pleas TW FTC fines EU, fines
1 1 1
1 2 1
279
X KO
Korea FTC fines
1
1
120 36 112 266
AU US X KO US
Australia, fines US guilty pleas Korea, fines US trial
1 1 1 1
1 3 1 1
104 233
X UK US
UK, fines NYC convictions
1 2
1 2
253
MEX
Mexican FCC fine
1
1
242
Taiwan
TWFTC fines
1
1
221
Taiwan
Taiwan FTC fines
1
1
109 110 111 102
X X X X
Italy, fines France, fines Swedish court, fines Canada, fines
1 1 1 1
1 1 1 2
274 237 113
X NL X EUR X US
EU fines EU fines US settlement
4 1 1
5 1 1
IT FR SE CA
X X
X
X
143
100 220 41
Price-Fixing Overcharges
Ferrosilicon, US Flour imports, Taiwan Ferry services, English Channel Forklift trucks manufacturing, Korea Frozen foods, Australia Frozen fish, US Fuels, military, Korea Futures contract, ME potatoes, US Games and toys, UK Garbage collection, NY and NJ Gas, liquid propane, Mexico Gas, liquid propane, Taiwan Gas, natural, distribution, Taiwan Gasoline, retail, Italy Gasoline, retail, France Gasoline, retail, Sweden Gasses, compressed, Canada Gasses, compressed, NL Glass, flat, Benelux Glass, flat, US
Name
Graphite electrodes
Characteristics International Location
84
X INTL
158 2 269 240
US US US Taiwan
197
US
125 83
ES X CA
250 227
KO X EUR
40 69 256 210 201
X EUR X INTL US US US
Bid Rigging
Found Guilty, Liable for Civil Penalties, or ExtraLegalc
Number of Episodes
Average Observations
US, EU, Korea, fines First episode legal US settlements US trial TWFTC fines
1
13
2 1 1 1
2 1 1 1
US settlements
1
1
X
Spain, fines Canada pleas
1 1
1 1
X
KFTC fine EU fines
1 1
1 1
Legal export cartel Legal export cartel US trial Legal cartel US civil settlement
1 4 1 1 1
1 4 1 2 1
X
JOHN M. CONNOR
Gunpowder, US Gymnasium seats, US Gypsum wallboard, US Harbor loading services, Taiwan High fructose corn syrup, US Hotel association, Spain Insecticide, forest, Canada Insurance, auto, Korea Iron and steel rolls, cast, EU Iodine Lead Legal aid fees, DC, US Lemons, California Linerboard, US
Code No.
144
(Continued)
Table A1.
X EUR UK US X INTL JP
Legal export cartel UK Commission US trial US pleas, EU fines JFTC actions
1 1 1 1 14
1 2 1 14 1
55 38 28 94 31
UK US X US X EUR US
Legal cartels US pleas, fines US pleas, fines US prosecution US settlements
40 57 2 1 1
1 1 4 1 1
72 78
X EUR X INTL
3 3
6 3
85 9
X INTL US
X
Legal cartel EU fines, US settlements EC, Canada fines US state convictions
1 1
1 1
10
US
X
11
US
12
1
1
X
US. state convictions US state convictions
1
1
US
X
US state convictions
1
1
13
US
X
US state convictions
1
1
14
US
X
US state convictions
1+
1
145
Methyl glucamine Milk, manufacturing, three counties, Kentucky Milk, manufacturing, two counties, Florida Milk, manufacturing, three counties, Florida Milk, manufacturing, Danville, KY Milk, manufacturing, Owensboro, KY Milk, manufacturing, core area, Kentucky
137 180 258 75 214
Price-Fixing Overcharges
Linoleum exports Linoleum, UK Liquor, retail, TX, US Lysine Manufacturing basic materials, JP Manufacturing, UK Manufacturing, US Magnesium metal Magnesite Market makers, NASDAQ, US Mercury Methionine
Name
(Continued)
Characteristics International Location
Bid Rigging
Found Guilty, Liable for Civil Penalties, or ExtraLegalc
15
US
X
US state convictions
19
US
X
262
US
30
US
226
Number of Episodes
Average Observations
109
1
US settlement
1
1
US trial
1
1
US trial
1
1
US
US trial
1
1
207
US
Legal cartel
1
2
259
US
US trial
1
1
230
X INTL
EC fines
1
1
186
DE
Legal cartel
1
1
X
JOHN M. CONNOR
Milk, manufacturing, Southeastern US Milk, manufacturing, Dallas, Texas Milk, manufacturing, North Texas Milk, manufacturing, Cincinnati, Ohio Milk, manufacturing, AMPI cooperative Milk, US marketing orders Mobile/cell phones (see telephone) Movie rentals, first run, MN, US Mushrooms, canned, Germany Nails, Germany
Code No.
146
Table A1.
16
X INTL
Legal cartel
2
7
217
Chile
Legal cartel
5
8
181
UK
UK Commission
1
2
209 89 228
US X EUR X INTL
Legal cartel EC fines EC fines
1 1 2
3 3 4
138
X EUR
1
1
99 157 24 190
X US JP X US US
1 1 1 1
1 1 1 1
154
US
1
1
35
US
US trial
1
1
134
CA
Legal cartel
3
3
105 141 118
X UK US X IT
UK probe US trial Italy, fines
1 1 1
1 1 1
US pleas and trial Legal cartel Legal cartel X
X X
Price-Fixing Overcharges 147
Nitrogen (sodium nitrate) fertilizer Nitrogen, nitrate, ammonium sulfate for fertilizer, Chile Nonferrous metals, UK Oil (see petroleum) Oranges, California navel Paper, carbonless, EEC Paper pulp, bleached sulfate Paper pulp, mechanical sulfite Paper, thermal fax, US Paints, export, Japan Petroleum, US Petroleum, TX and Oklahoma Petrol., offshore leases, US Petroleum refining, Midwest Petroleum, lamp oil, Ontario Pharmaceuticals, UK Pharmaceuticals, US Pharmaceuticals, respiratory, Italy
Name
Quebracho extract Quinine
Characteristics International Location
119
X IT
135
X US
132 23 143 47 156 145 178 232 272 57 73
X EUR US US X EUR US US JP X EUR X EUR UK X EUR
50 131
X ARG X EUR
Number of
Bid Rigging
Found Guilty, Liable for Civil Penalties, or ExtraLegalc
X
Italy, fines
1
1
US indictment
2
2
1 1 1 3 1 1 1 1 1 1 4
1 1 2 7 1 1 1 1 1 1 19
3 1
8 2
X
X
US trial US trials US trial US trial JFTC fines EC fines EC fines Legal cartel Last episode 1935– 1939 US conviction US conviction US pleas, fines
Episodes
Average Observations
JOHN M. CONNOR
Pharmaceuticals, cholesterol, Italy Phosphate rock exports, US Phosphorus, red Pipes, cast iron, SE US Pipes, concrete, US Platinum Plumbing fixtures, US Plywood, US Plywood, Japan Polyvinyl chloride plastic Polypropylene, EU Porcelain, sanitary, UK Potash
Code No.
148
(Continued)
Table A1.
US
Legal US cartel
1+
7
133 208 136 257
US US X EUR US
Legal US cartel Legal US cartel US trial
1 1 1 1
1 1 1 0
268
US
US trial
1
1
219 236
BL SW
EU fines
1 2
1 2
20 194 3 168
X EUR US US UK
Legal export cartel
2 2 1 4
4 4 1 9
215
UK
173
X
X
US convictions Legal cartel
1
1
US
Commission decision DOJ consent decree
1
1
43 171
X EUR X EUR
EU fines Legal cartels
1 6
1 2
86
X INTL
US convictions
1
1
Legal US cartel
1
2
127
US
X
149
Scholarships, graduate, US Shipping, France-Africa Shipping, three conferences Shipping, chemical tankers Shipping, express packages, US
49
Price-Fixing Overcharges
Railroad, Chicago to East, US Railroad, US South Raisins, US Rayon Realtors’ listing service, CA, US Realtors’ sales commissions, MD Roofing felt, Belgium Roundwood buying, Sweden Rubber, crude Salt, Michigan Salt, rock, US Salt, white, Salt Union, UK Salt, white, duopoly, UK
Name
Code No.
150
(Continued)
Table A1.
Characteristics International Location
Bid Rigging
165 27 79 77 74
JP US X EUR X INTL X EUR
X
254 64 187 238 270
US UK DE DE JP
X
Steel pipes, sewage, UK Steel pipes, insulated, EU Steel rails, US Steel rails, Europe Steel, seamless tubes, EU Steel tubes, US Steel, flat rolled Steel, flat stainless, EU
58 93 150 169 91 151 218 92
UK X EUR US X EUR X EUR US X INTL X EUR
US and EU fines Legal cartel US trial Legal UK cartel Legal cartel Legal cartel Governmenttolerated Legal UK cartel EU fines First episode legal Legal cartel EU fines Legal cartel EU fines
Episodes
Average Observations
1 1 1 1 2
1 1 1 5 6
2 1 1 4 1
2 1 1 6 2
1 1 1 1 1 1 1 1
1 2 1 3 3 1 1 3
JOHN M. CONNOR
Soil and gravel, Japan Soft drinks, US Sodium chlorate Sorbates Steel, bulk metal, European Steel road culverts, US Steel drums, UK Steel girders, Germany Steel and iron, Germany Steel, integrated, Japan
Found Guilty, Liable for Civil Penalties, or ExtraLegalc
Number of
4
US
X
US convictions
1
2
5
US
X
US convictions
2
4
EU fines
1 4 1 1
X EUR X INTL US X US+CA
Legal export cartel DOJ probe
1 3 2 1
44 17 67 126 96 128 223 170
US X INTL US ES UK X EUR US DE
US trial Legal export cartel US trial Spain, fines EU, fines Legal cartel Civil settlement Legal cartel
1 2 2 1 1 1 1 1
1 3 8 1 1 1 1 1
29 146 139 97
US X INTL US X IT
Legal export cartel US trial EC probe
1 1 1 1
1 4 1 1
117 241 200 273 65
X IT X Taiwan US X UK UK
Italy, fines TWFTC fines US settlement UK OFT decision Legal UK cartel
1 1 1 1 1
1 1 1 2 1
X X
X X X
151
95 87 191 103
Price-Fixing Overcharges
Steel, structural, buildings, US Steel, structural, bridges, US Steel, structural, EU Sulfur, international Sulfur, crude, US exports Sulfuric acid, US and Canada Sugar beets, US Sugar, cane Sugar refining, US Sugar, Spain Sugar refining, UK Tea Tetracycline, US Thorium nitrate, Germany Timber, US auctions Tin Titanium metal, US Telephone fees, UK and Germany Telephone fees, Italy Thread, surgical, Taiwan Tobacco leaf, US Toys and games, UK Transformers, large, UK
Name
Transformers, system, UK Travel brokers’ fees, UT, US Tungsten carbide Uranium metal
Characteristics International Location
Number of
Bid Rigging
Found Guilty, Liable for Civil Penalties, or ExtraLegalc
X
Legal UK cartel
1
1
Episodes
Average Observations
66
UK
234
US
US trial
1
1
8 130
X INTL X INTL
2 1
4 6
46 265
US US
US trial US pleas, settlements US jury trial US trial
1 1
1 1
80
X INTL
US and EU fines
14a
64
278
X China
US civil case
1
1
140
US
Patent abuse trial
1
1
231
BL
EC fines
1
1
185 149
DE X US
Legal cartel Legal cartel
1 1
1 2
JOHN M. CONNOR
Vanadium ore, US Vegetable parchment manufacturing, US Vitamins and carotenoids, bulka Vitamin C, China exports to US Vitamin D, US Vitamin B4 (see choline) Wallpaper manufacturing, BL Wire, Germany Wire nails, US
Code No.
152
(Continued)
Table A1.
Total 279 markets
56 148 182 68 90 –
UK
Legal UK cartel
US UK X INTL X EUR 103 International
83
1
1
First episode legal UK Commission Legal export cartel Fined by EC
5 3 5+ 1
6 3 8 2
192 guilty or liabled
816b
57 known ‘‘legal’’ 30 presumed legale
770 aver-age
Price-Fixing Overcharges
Wire rope, non-marine, UK Whiskey alcohol, US Wire and cable, UK Zinc metalf Zinc phosphate
270 peak 1,040 total
Source: Appendix Table 2 and references. One for all vitamins, one for the three carotenoids, and 12 individual vitamins. b This total counts five multiple cartel summaries (see cartel numbers 15, 38, 55, 213, and 214 above) as 245 episodes. Counting each of these entries as one episode reduces the total to 576. In contrast, most bid-rigging cartels could in principle count each contract as an episode, but are treated as one here; for example, in cartel number 211 more than 3,500 contracts were overtly collusive bids. c Fines, trials, consent decrees, settlements, commission decisions, parliamentary inquiries, and known official investigations are all considered adverse sanctions for cartels. Adverse parliamentary and commission decisions resulted in changes in conduct similar to consent decrees. d Includes seven markets (88, 97, 103, 105, 107, 198, and 276) that in 2004–2005 were being investigated by antitrust authorities; a high proportion will be legally sanctioned. e Counts blank entries in column above. Blank entries are cases without information about any criminal sanctions or adverse civil proceedings and are presumptively legal or extra-legal. f This cartel was fined at the end of its life by the EC (8/6/1984) but operated openly in the belief that it was legal for most of its existence. a
NB. Appendix Tables 2–5 can be found at: http://www.agecon.purdue.edu/directory/details.asp?username=jconnor
153
This page intentionally left blank
PREDATORY PRICE CUTTING AND STANDARD OIL: A RE-EXAMINATION OF THE TRIAL RECORD James A. Dalton and Louis Esposito ABSTRACT John McGee’s 1958 paper, ‘‘Predatory Price Cutting: The Standard Oil (NJ) Case,’’ has had an astonishing influence on both antitrust policy in the United States and economic lore. McGee argued that predatory pricing is irrational and his analysis of the Standard Oil Company Matter, decided in 1911, led him to conclude that the Record in this case does not show that Standard Oil engaged in predatory pricing. This single publication appears to serve as a foundation of the U.S. Supreme Court’s position on the issue of predatory pricing, as well as the assertion by many economists that predatory pricing is irrational and rarely occurs. Numerous arguments have been advanced during the past 25 years that predatory pricing can be a rational strategy. As to McGee’s empirical findings, there has been no re-examination of the Record of the Standard Oil case to determine the validity of his finding that the trial ‘‘Record’’ does not support the claim that Standard Oil engaged in predatory pricing.
Research in Law and Economics, Volume 22, 155–205 Copyright r 2007 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0193-5895/doi:10.1016/S0193-5895(06)22005-0
155
156
JAMES A. DALTON AND LOUIS ESPOSITO
We examined this Record and have found that the trial Record contains considerable evidence of predatory pricing by Standard Oil. Therefore, the Record does not support McGee’s conclusion that Standard Oil did not engage in predatory pricing. Thus, the decisions of the Supreme Court in recent years, as well as the opinions of many economists, concerning predatory pricing are not consistent with either current theory or the empirical record.
1. INTRODUCTION The study of history sometimes demands a re-examination of the facts of a matter in order to evaluate the basis of key conclusions, especially when those conclusions appear to have had a substantial effect on a key issue of social policy. John McGee’s 1958 paper, ‘‘Predatory Price Cutting: The Standard Oil (NJ) Case,’’ has had an astonishing influence both on antitrust policy in the United States and on economic lore. McGee argued that predatory pricing is irrational and his analysis of the Standard Oil Matter1 led him to conclude that the trial Record does not support the assertion that Standard Oil engaged in predatory pricing. This single publication appears to serve as a foundation of the U.S. Supreme Court’s position on the issue of predatory pricing, as well as the basis for the assertion by many economists that predatory pricing is irrational and rarely occurs. In recent years, the Supreme Court cited McGee’s paper and its conclusions when addressing predatory pricing in two important cases. In Brooke Group Ltd v. Brown and Williamson Tobacco Corp., 61U.S.L.W. 4669 (1993), ‘‘the Supreme Court found that predatory pricing was speculative and ‘inherently uncertain’ and noted its general implausibility’’ (Bolton, Brodley, & Riordan, 2000, p. 2243). In Matsushita Electric Industrial Corp., Ltd., et al. v. Zenith Radio Corp., 475 U.S. 574 (1986), the Court stated that ‘‘there is a consensus among commentators that predatory pricing schemes are rarely tried and even more rarely successful’’ (Bolton et al., 2000, p. 2243). The Court explicitly cited McGee’s paper in support of this latter statement.2 The influence of McGee’s paper continues to this day. David Friedman wrote in 2000: Despite the widespread belief that Rockefeller maintained his position by selling oil below cost in order to drive competitors out of business, a careful analysis of the record of the antitrust case that led to the breaking up of Standard Oil found no evidence that
Predatory Price Cutting and Standard Oil
157
he had ever done so. The story appears to be the historian’s equivalent of an urban myth (Friedman, 2000, p. 250).
Mankiw, in his best-selling introductory economics textbook (2004), wrote: Although predatory pricing is a common claim in antitrust cases, some economists are skeptical of this argument and believe that predatory pricing is rarely, and perhaps never, a profitable business strategy (Mankiw, 2004, p. 365).
And in 2005 Carlton and Perloff, in their best-selling industrial organization textbook, wrote: For example, one of the most widely cited examples of price predation was the creation of Standard Oil. Supposedly, Rockefeller bought small, independent oil refineries after having lowered price to drive them out of business. McGee (1958), in his careful examination of this historical period, rejects that view and concludes that Rockefeller’s rivals were bought out on rather favorable terms (Carlton & Perloff, 2004, p. 360).
In summary, McGee’s paper is arguably one of the most influential papers ever published by an economist in terms of its impact on American antitrust policy and economic lore.3 Regarding the influence of McGee’s paper on economic thinking, Posner wrote in 1976 that the literature on predatory pricing had been ‘‘excessively’’ influenced by McGee’s ‘‘pathbreaking’’ article on the Standard Oil Trust (Posner, 1976, p. 185).4 At least four factors justify a careful re-examination of the Standard Oil Record to determine whether there is direct evidence of predatory pricing. These factors, especially in combination, increase expectations regarding the likelihood of finding evidence of predatory pricing in the trial Record. First, economists have challenged McGee’s contention that predatory pricing constitutes irrational behavior repeatedly during the past 25 years; Section 2 reviews these findings. Second, some economists argue that Standard did engage in predatory behavior. Granitz and Klein (1996), for example, argued that Standard’s participation with railroads in a transportation cartel caused Standard’s rivals to experience relatively higher transport costs. Such behavior was predatory because it excluded equally or more efficient competitors from the market.5 Third, Standard Oil’s business philosophy and organization, coupled with the industry’s structure, made selective use of predatory pricing a potentially effective mechanism for eliminating some rivals. Fourth, historians and biographers have provided numerous anecdotes from John D. Rockefeller’s papers regarding the predatory nature of Standard’s behavior.6,7 Section 3 discusses the third and fourth factors. The purpose of this paper is to determine, by re-examining only cases cited by McGee in his 1958 paper, whether there is sufficient evidence in the
158
JAMES A. DALTON AND LOUIS ESPOSITO
Trial Record of the Standard Oil matter to conclude, as McGee did, that Standard Oil did not engage in predatory pricing. It is worth noting that in our empirical analysis, we ignore considerable evidence of likely predatory pricing by Standard Oil presented in the report by Commissioner Garfield (Garfield, 1906)8 of the U.S. Department of Commerce, Bureau of Corporations (1906) and the Rockefeller papers. Nevertheless, the major finding of this paper is that the Record contains considerable evidence of predatory pricing. Simply stated, the Record does not support McGee’s conclusion that Standard Oil did not engage in predatory pricing. To the contrary, the evidence in the Record, at the very least, would have provided the Attorney General of the U.S. with a strong justification in 1906 for conducting a more complete and separate investigation into the extent and effects of Standard’s predatory pricing behavior. Section 4 presents the methodology and framework for our examination of the Standard Oil Record. Section 5 presents the core of the paper, namely, the presentation and analysis of the evidence from our assessment of the Record for cases that McGee analyzed in his paper. Section 6 offers concluding comments as well as our application of a legal rule that Bolton et al. (2000) have recently proposed for use in predatory pricing cases.
2. RATIONAL PREDATORY PRICING 2.1. McGee’s Model and Critique by Economists McGee argued that predatory pricing is irrational for three reasons: given its much larger market share, the losses of the predator would greatly exceed the losses of the prey; an efficient prey could survive losses in the short run because the capital market would supply it with the necessary funds; and even if the prey were forced to close down in the short run, either the prey or a new purchaser would reopen the plant when the predator subsequently raised prices (pp. 139–142).9 McGee’s analysis assumes: (1) the predator reduces the price on all its output; (2) there are perfect information and perfect capital markets (i.e., common knowledge, symmetric information, and certainty); and (3) there are no signals to potential entrants about the possible responses of the dominant firm to new entry, i.e., McGee ignored strategic considerations in his analysis. After reviewing McGee’s argument and other analyses, Posner stated that a persuasive case that predatory pricing was irrational had not been made.
Predatory Price Cutting and Standard Oil
159
With reference to the Standard Oil case in particular, Posner argued that since Standard operated in numerous markets, it did not have to cut its price on all its output. Standard instead could cut or threaten to cut its price in selective geographic markets. Further, even an implicit threat of price cutting could deter entry (Posner, 1976, pp. 186–187). Posner reiterated his position in 2001: while documented cases of predatory pricing are rare, it is not always an irrational method of deterring entry. Especially when it takes the form of area price discrimination or ‘fighting brands’, so that the predator does not have to lose more money than the new entrant by lowering his prices throughout the market, a monopolist may maximize his profits by investing in a reputation for predatory response to threatened entry (Posner, 2001B, p. 936).
Posner also challenged McGee’s assumption of perfect capital markets when he introduced the ideas of uncertainty and asymmetric information in real markets (Posner, 1976, p. 185).10 Indeed, loan terms tighten and the cost of capital increases with risk. Several recent Nobel Laureates have underscored the impact of uncertainty and imperfect information on decision-making by smaller rivals. Stiglitz argued that small deviations from the assumption of perfect information in markets lead to substantial deviations in actual from ideal market performance (Stiglitz, 2002, pp. 460–501). Akerlof posited asymmetric information as a ‘‘major determinant of market structure (Akerlof, 2002, p. 413).’’ Spence suggested that the demonstration effect of a dominant firm’s behavior can contribute to the development of a monopoly structure in the absence of perfect information (Spence, 2002, p. 453).11 In summary, when viewed in the context of strategic behavior, asymmetric information, and imperfect capital markets, price cutting (actual or threatened) may increase or sustain market power if it deters entry or disciplines rivals.12 2.2. A Rational Strategy of Predatory Pricing13 Given considerable uncertainty in the petroleum industry in the 30-plus years prior to 1906, a small competitor probably could not have fully known the purpose of Standard Oil’s strategies, the precise nature of its cost structure, and the payoff matrix.14 Consequently, Standard Oil could have employed a strategy that was designed to convince potential entrants, as well as the capital markets, that it was the low-cost firm and could profitably operate at low prices, even if it were not a low-cost firm.15 This situation
160
JAMES A. DALTON AND LOUIS ESPOSITO
would be further complicated by the uncertainty faced by potential entrants regarding what their actual costs would be, how much market demand was growing, and the accuracy of their forecast of the price elasticity of demand. The resolve of the potential entrant, as well as the willingness of the capital markets to supply funds, diminishes with the increase in uncertainty. This diminishing resolve is especially strong when a large portion of the investment is a sunk cost, as it was in the petroleum refining industry. The resolve is even weaker when the length of time before Standard would raise its price is unknown. The small rival will act rationally while Standard, by its actions and reactions (for example, by cutting its price substantially in a particular geographic area), would send a signal to the entrant, all potential entrants, and the capital markets that Standard was the low-cost firm, was willing to sustain losses in the short run to protect its dominant market position, or was irrational.16 An exit by the smaller rival would reinforce this signal.
3. FURTHER JUSTIFICATION FOR RE-EXAMINATION OF THE RECORD This section presents the business philosophy and internal organization of Standard Oil before 1906, the structure of the oil industry in the late 1800s, and information contained in the published work of historians and biographers during the past 65 years on the pricing practices of Standard. Taken as a whole, this body of information indicates that Standard met competition with deep price cuts and had a philosophy consistent with a policy of predatory pricing. Such information provides strong justification for a re-examination of the Record to determine whether McGee’s conclusions can be sustained. 3.1. Business Philosophy and Internal Organization Rockefeller aggressively pursued orderly markets and stable prices, disdaining competition, from as early as 1869 or 1870.17 Cooperation was the major dimension of Rockefeller’s organizational philosophy and, for him, ‘‘cooperation’’ meant ‘‘such loaded items as trust, monopoly, oligopoly, or cartel’’ while he ‘‘scorned the concept of free markets in textbooks.’’ Rockefeller acknowledged that ‘‘The idea was mine’’ to organize the oil industry under the control of the Standard Oil Company (Nevins, 1953, v. 2, p. 9).
Predatory Price Cutting and Standard Oil
161
For Rockefeller then, an orderly market translated into a monopoly or a cartel in which price competition is absent. He proceeded to ‘‘make the oil business safe and profitable’’18 by vertically integrating to form a giant company.19 Standard Oil monopolized the oil industry through acquisition and managerial control of companies that it either wholly owned or controlled. These shares were placed in the trust that was formed in 1882 and then in 1899 were transferred to form the Standard Oil Company of New Jersey.20 Rockefeller ‘‘knew everything that was going on.’’ His committee system was a ‘‘top-down hierarchical system’’ that effectively coordinated the companies in this ‘‘confederation,’’ including partially owned subsidiaries. The Executive Committee included the top managers who ‘‘set the overall policies and directions’’ (Chernow, 1999, pp. 228–229, 250).21 Standard also employed a sophisticated and extensive intelligence network. Its employees identified the shipments and customer destinations of wholesalers of competing refiners using information obtained from agents of the railroads, retailers of refined oil products, and other employees of Standard. Standard maintained this information in an elaborate card catalogue that was then used to direct its sales force to capture or recapture the customers of rival refiners.22
3.2. Structure of the Oil Industry Rockefeller succeeded in eliminating the chaos of competitive markets and developed the monopoly that became Standard Oil. Standard controlled only 10 percent of the refining industry in the United States in 1870; by 1880 it had 90 percent of refining capacity and controlled the gathering lines in the crude fields, the pipelines, and rail transportation. From that time forward, Standard persistently controlled 80–90 percent of refining, shipping, and marketing in the United States, while similarly dominating the export of refined oil products.23 As late as 1907, shortly after the United States filed its monopolization complaint against Standard Oil on November 15, 1906, Standard refined and marketed more than 87 percent of domestic and exported kerosene, and the company was more than twenty times the size of Pure Oil, its next largest rival (Chernow, 1999, p. 537).24 The demand for petroleum products grew rapidly from 1870 to 1906, and a high rate of growth should have facilitated entry (Scherer & Ross, 1990, pp. 119–120). Nevertheless, Standard maintained its dominant share of both refining and marketing for 30 years in spite of this growth, providing
162
JAMES A. DALTON AND LOUIS ESPOSITO
additional weight for an inquiry into whether Standard employed predatory pricing. Thus, Standard was a well-organized combination whose leadership dominated the oil industry during the years prior to the Supreme Court’s decision in 1911 to dissolve Standard Oil. The Supreme Court concluded that the means and effects of the combinations comprising the company clearly displayed the anticompetitive intent of Standard Oil of New Jersey.25 In summary, the hypothesis that Standard Oil used predatory pricing as a tool for acquiring and maintaining its monopoly is consistent with Standard’s philosophy, its organization, and the market dominance that it pursued and achieved. 3.3. Information Available on Price Cutting by Standard Oil By 1958, historians and biographers of John D. Rockefeller had provided at least a prima facie case for the likelihood of predatory pricing by Standard Oil. For example, Rockefeller biographer Allen Nevins characterized attacks on ‘‘the price-slashing tactics used by the Standard’’ as having been ‘‘well justified’’ (Nevins, 1953, v. II, p. 370).26 Indeed, Scherer implied that McGee should have looked at sources outside the Record before he rendered his conclusion. Citing evidence in Nevins (1953, v. II, pp. 62–67), Scherer concluded that ‘‘Had McGee bothered to cast his net further into historical works drawing upon the Rockefeller papers, he would have found that Standard tried in some instances to drive out its rivals through price cutting’’ (Scherer, 1970, p. 275). Historical research, published both before and after 1958, illustrates that Standard used deep price cuts when confronted by new rivals in the marketplace. The findings of this research may not be sufficient to show actual price predation but they certainly invite further analysis of the Record.27
4. THE FRAMEWORK FOR THE INVESTIGATION OF THE RECORD To this point, our paper shows that there is strong justification for reexamination of the Record to determine whether McGee’s findings in 1958 are reasonable. Specifically, Section 2 shows that economists believe that predatory pricing can be a rational strategy. Section 3 shows that the policy and practices of Standard were consistent with a strategy of predatory
Predatory Price Cutting and Standard Oil
163
pricing. This section of the paper specifies the hypothesis to be investigated, the empirical methodology, the measure of predatory pricing employed in the analysis, and the criteria used in the selection of the cases analyzed in the re-examination.
4.1. Hypothesis and Methodology In essence, McGee examined the Record and rejected his hypothesis that Standard Oil used predatory pricing to acquire its monopoly in refining either by compelling refiners to sell out or by eliminating refiners and nonintegrated marketers (McGee, 1958, pp. 143, 153, 157). We proceed by analyzing four of the five ‘‘major’’ cases that McGee analyzed in his paper to investigate his hypothesis. This analysis relies only on information in the trial Record. If, after our re-examination of these four cases, we find substantial evidence that contradicts McGee’s interpretations, we can conclude that McGee’s rejection of the null hypothesis was in error. The weak database contained in the Record severely constrains our empirical analysis, as it did McGee’s work in 1958. This constraint presents several obstacles to determining unequivocally whether or not Standard employed predatory pricing. First, the cases that can be selected for analysis occurred mainly after Standard had achieved a 90 percent market share. For example, McGee did not supply any detail on the matter of the acquisitions in the early 1870s and the Record did not have much, if any, relevant information for that time period. Second, oral testimony provided most of the evidence, including validation of trial exhibits. By the time of the testimony in 1908–1909, some of the alleged behavior was already 36 years old. Indeed, McGee recognized the toll of time on the stock of relevant witnesses and the detail in the memories of witnesses to hand (p. 145). It is not surprising that the bulk of McGee’s cases address Standard’s behavior only in the last 15 years prior to 1906.28 Finally, the Record does not often provide a clear description of market behavior. The parties in 1906 did not employ databases that expert economists and accountants use today to analyze pricing behavior.29 Thus, although the evidence in the Record may in some cases be insufficient to conclude either that Standard did or did not engage in predatory pricing, the evidence may be sufficient to justify further investigation by the thenAttorney General of the United States, assuming better data were available. In such cases, one cannot accept McGee’s conclusion that the Record does not contain evidence that Standard engaged in predatory pricing because
164
JAMES A. DALTON AND LOUIS ESPOSITO
such a conclusion clearly implies that the Record reasonably represents the nature of Standard’s pricing behavior. This acceptance would be especially inappropriate and undesirable since so many observers use McGee’s article to argue that Standard Oil did not employ predatory pricing.
4.2. Measure of Predatory Pricing McGee evidently employed an ad hoc definition of predatory pricing. He did not define a measure of predatory pricing in his 1958 paper. In his 1980 article, McGee stated that to establish predatory pricing ‘‘it will have to be a cost-based rule’’ and that ‘‘My 1958 article refers to marginal and average variable costsy (p. 292).’’30 Yet McGee did not specify an explicit rule, let alone price–cost margins, as his measure of predatory pricing in his 1958 paper. Regardless, McGee could not have applied a cost-based rule to the information in the Record on anything close to a consistent basis because the Record does not contain sufficient information to construct a comprehensive database on prices and costs for his universe of cases. The testimony on price– cost margins, however, at times is clear enough to analyze price–cost margins and at other times allows strong inferences to be drawn about the margins. Regardless of McGee’s statement that he adopted a cost-based measure, there is no consensus among economists about an acceptable definition of predatory pricing. According to Scherer, ‘‘‘Predatory’ must be enclosed in quotation marks, because there is not a universally accepted definition (Scherer, 1996, p. 107).’’ We employ a two-step approach to determining whether predatory pricing was used by Standard Oil. 4.2.1. Step One We employ the following definition in our empirical analysis of the trial Record: Predatory pricing occurred if, in specific instances: (a) Standard Oil lowered its price in the short run below the price of the entrant; (b) the apparent effect of the price cut either eliminated a rival, disciplined a rival, coerced a rival into a merger or acquisition, or discouraged the entry of new firms; and (c) Standard Oil then increased the price to its original level or higher. This non-cost-based approach to predatory pricing is similar, in many respects, to the concepts of Scherer (1976a, 1976b), Williamson (1977), Baumol (1979), Kahn (1991), and Edlin (2002).31 In point (a) above, the initial step for predatory pricing arises in the primary market when the dominant firm reduces its price below the
Predatory Price Cutting and Standard Oil
165
entrant’s price.32 Consumer surplus rises when sellers reduce prices.33 With points (b) and (c), consumer surplus declines when competition declines and price rises. The dominant firm’s behavior actually reduces consumer surplus below the level caused by the entrant’s price cut. Indeed, the entrant may be economically efficient but financially weaker (Kahn, 1991, p. 137). Regardless, when initially an entrant is less efficient the dominant firm prevents a further gain in allocative efficiency by preventing the entrant from achieving efficiency in a reasonable period of time.34 This three-part measure focuses on the impact of the dominant firm’s pricing on competition.35 The entrant must reduce the price below the price of the monopolist in order to sell a product. When the dominant firm reduces its price below the rival’s price and either eliminates or disciplines an entrant before raising the market price, this behavior is not consistent with aggressive competition on the merits. The dominant firm need only match the rival’s lower price to prevent loss of market share. With this behavior, the monopolist can charge a higher price without providing a better product or improved services over the long run. This monopolization is not consistent with ‘‘greater skill, foresight, or industry.’’36 Further, this monopolizing behavior sends signals to other actual and potential cost-efficient competitors. In effect, the success of the dominant firm’s aggressive behavior does not contribute to economic efficiency; competitive forces are reduced, supply is lower, and capacity utilization is not improved. Indeed, the fact that the dominant firm responds to entry by reducing its price below the entrant’s price supports the existence of the competitive challenge posed by actual and potential entry and, hence, the welfare-reducing impact of the dominant firm’s behavior. In summary, our measure for determining whether Standard Oil engaged in predatory pricing does not require either evidence that Standard Oil priced below its cost or evidence of Standard’s intent to predate. Our measure, on the other hand, does allow us to distinguish between predatory pricing and hard or aggressive price competition. Predatory pricing always involves a temporary decrease in price that increases quantity demanded in the short run followed by a decrease in supply and a higher price in the long run. On the other hand, aggressive price competition should result in greater supply and a lower price in the long run, not a higher price and restricted supply. Also, aggressive price competition will not eliminate cost-efficient rivals. However, although one can theoretically distinguish between predatory pricing and aggressive price competition, in analyzing the Record one would expect to see a lot of ‘‘noise’’ in the data. This noise may, in some instances, make it more difficult to determine whether Standard engaged in aggressive price competition or predatory pricing.
166
JAMES A. DALTON AND LOUIS ESPOSITO
We also employ a second step when analyzing whether predatory pricing occurred. 4.2.2. Step Two Given the limitations of the Record, we present, where possible, other specific information from the Record that is relevant to the analysis. In effect, we apply a stricter measure by introducing factors such as Standard’s price structure during the period of its ‘‘aggressive pricing,’’ Standard’s intent, and its price–cost margins after the price reductions.37 These factors are neither necessary nor sufficient for an analysis of the nature of aggressive pricing behavior. They can, however, underscore conclusions reached under the three-part measure defined above. Standard’s Price Structure. Section 5 shows that Standard typically did not cut prices throughout a market when combating a rival. Rather, Standard generally cut prices only to customers of its rivals, leaving unchanged the prices to the customers that it served. Two factors make this fact relevant for an analysis in the fuller context of Standard’s aggressive pricing behavior. First, avoiding price cuts across the board limits sacrifice (lost profits) in the short run; the rationality of predatory pricing is inversely related, in part, to the cost of the behavior. Second, targeting only the customers of Standard’s rivals implies a strategy of long-run profit maximization based on raising the prices to customers regained by Standard to the higher level of prices that Standard had been charging its own customers.38 Standard’s Intent. Some researchers have argued that finding evidence of intent to predate is important to establish predatory pricing.39 Indeed, Greer has argued that the aforementioned selective price cutting can ‘‘contain striking evidence of predatory intent (Greer, 1980, p. 354).’’ We will analyze evidence that relates to intent to predate only where the Record is clear for the specific cases analyzed herein and where the evidence does not derive from possibly biased statements by ‘‘hothead’’ competitors of Standard.40 Standard’s Price–Cost Margins. One cost-based measure of predatory pricing requires that the alleged predator set its prices below its own average variable costs.41 Such a rule is deemed by some economists to be a more definitive measure of predatory pricing than the measure that we have defined. We will incorporate and analyze evidence, where available, as to whether Standard set prices below its variable costs.
Predatory Price Cutting and Standard Oil
167
4.3. Criteria for Selection of Cases Our objective is to analyze the Record of cases that McGee cited in his paper as ‘‘principal examples’’ of alleged predatory pricing (p. 144).42 McGee mentioned 12 cases of ‘‘price cutting against competing refiners’’ and seven cases of ‘‘price cutting involving jobbers and retailers.’’ Of the 12 cases involving refiners, only five cases involved one or more pages of fact and analyses. The facts and analyses of the other seven refining cases were presented by McGee in one-half of a page or less. Of the seven cases involving jobbers and retailers, only four of the cases discussed specific pricing behavior of specific jobbers or retailers and in each of these cases the facts and analyses were presented by McGee in one-third of a page or less. The five ‘‘major’’ cases (i.e., cases involving one or more pages of fact and analysis) discussed by McGee in his 1958 paper were (1) the Cleveland Acquisitions, (2) the Rocky Mountain Oil Company, (3) Scofield, Shurmer, and Teagle (SS&T), (4) Fehsenfeld (the Red C Interests), and (5) Cornplanter Refining Company. With one exception (the Cleveland Acquisitions), we analyze each of these cases in detail.43 Granitz and Klein analyzed the Cleveland Acquisitions in detail and concluded ‘‘there is considerable testimonial evidence that Standard used the South Improvement Company to induce Cleveland refiners to sell and that many of these sales were at distress prices (Granitz & Klein, 1996, p. 15).’’ Given their extensive analysis and conclusions, we saw no need to further analyze the Cleveland Acquisitions.44 And this trial Record contains little coherent information on the specific acquisitions occurring 34 years before the government sought to break up Standard Oil. If we find substantial evidence in each of the four cases that we analyze indicating that Standard Oil engaged in predatory pricing, then an analysis of the other ‘‘minor’’ cases analyzed by McGee is unnecessary. In other words, a finding in four of the major cases analyzed by McGee that there is sufficient evidence to suggest that Standard engaged in predatory pricing is more than sufficient to contradict McGee’s conclusion.
5. ANALYSES OF CASES OF ALLEGED PREDATORY PRICING Each of our case studies in this section uses the following format: (1) a summary of the facts contained in the Record; (2) an evaluation of McGee’s
168
JAMES A. DALTON AND LOUIS ESPOSITO
analysis and conclusions; and (3) a presentation of an alternative analysis and conclusion, where appropriate.
5.1. The Red C Case According to McGee, the Red C case was a ‘‘suspect’’ case of predatory pricing that did not involve either a merger or an acquisition (pp. 153–154). The Red C Oil Manufacturing Company marketed refined oil products in the south; contrary to its name, it did not refine petroleum.45 The Record contains testimony on a significant number of incidents that are relevant, many of which occurred before 1901.46 5.1.1. The Facts Red C and Standard Oil of Kentucky competed against each other in the marketing of refined oil products primarily in the states of Maryland, Virginia, and the Carolinas. From 1878 to 1897, Standard and Red C sold and shipped refined oil products in barrels by rail. After 1897, Standard ‘‘increasingly’’ used tank cars for shipping its refined oil products while Red C continued to use barrels, except in the Baltimore area where they used tank wagons.47 Standard had the volume to justify shipping in carload lots. Consequently, Standard had a significant cost advantage against Red C when Red C shipped less than a full carload because the freight rate for lessthan-carload shipments was ‘‘destructive of our business,’’ according to W. H. Fehsenfeld, the President of Red C.48 Red C competed for customers in markets that Standard had monopolized. Whenever Red C entered a geographic market dominated by Standard, it initially sold its products at prices below the existing price for Standard’s products. According to C. T. Collings, Second Vice President of Standard Oil of Kentucky, Standard was the pioneer in ‘‘establishing the business’’ and any new entrant would have to initiate a price cut in order ‘‘to wean the trade’’ away from Standard.49 Standard responded to the threat of losing customers by selectively reducing its prices. Standard Oil of Kentucky had developed a customer database for the geographic markets in which it operated and used it to identify customers that had defected to Red C.50 When Standard dispatched a salesman to the customers that had placed orders with Red C, the salesman would offer the defector a price below the price they were receiving from Red C. This often caused Red C customers to countermand their orders with Red C and/or cease buying from Red C in the future.
Predatory Price Cutting and Standard Oil
169
Standard Oil used as many as seven bogus companies during its competition with Red C to implement this type of selective price cutting: Eureka, Eagle, Southern Oil Company of Richmond, Dixie Oil Works, Davidson Oil Company, Paragon Oil, and Home Safety Oil Delivery.51 Generally speaking, a bogus wagon was owned by Standard Oil but was perceived by customers as representing a marketing company independent of Standard.52 The purpose of a bogus wagon was to undercut the prices to customers of Standard’s rivals while allowing Standard to sell at higher prices to its own customers in the same geographic market. McGee acknowledged that bogus wagons were used because ‘‘secrecy was useful’’ in order for Standard not to antagonize its own customers who were paying higher prices (p. 159). Testimony from a senior executive of Standard further reinforces the manner in which Standard reacted to competition in territory that included Red C specifically and competitive wholesalers in general. H. P. Westcott had been First Vice President of Standard of Kentucky since 1900 and for several years before that time had been in the sales department at Standard’s headquarters in New York. He testified that he had been receiving data on shipments by competitors, by source, for years throughout his territory. He would then turn these data over to the ‘‘statistical department’’ in New York. Further, under his direction, Standard Oil of Kentucky operated bogus companies, including Dixie Oil, Banner Oil, and Consumers Oil; these companies had not been operated as Standard-named businesses in order to get the business from the rivals but as secret Standard businesses that later would turn over sales taken from competitors to Standard (v. 2, pp. 704–728).53 These bogus companies had several characteristics that enabled effective price discrimination by Standard: they presented themselves as independent of Standard;54 they had employees in common with Standard;55 they did not pursue Standard’s customers (Fehsenfeld, pp. 2312–2320); Standard used the intelligence system and bogus companies to recapture customers that Red C had taken and to eliminate Red C from specific geographic markets;56 Standard priced below Red C’s carload price causing Red C to lose customers and to exit from particular geographic markets;57 and the bogus companies turned their customers over to Standard.58 This clearly implies that Standard increased prices to those customers that the bogus companies had regained from Red C, once Red C exited the market; after all, Standard had not reduced prices to its own customers. Lowering price to customers until a rival is eliminated and then increasing the price is consistent with our definition of predatory pricing. The Record contains considerable additional testimony of the impact on Red C of price cutting by the bogus companies. Standard’s response to the
170
JAMES A. DALTON AND LOUIS ESPOSITO
entry of Red C into its markets had the intended effect of preventing Red C from obtaining enough sales to justify receiving the full carload rate for its shipments. Standard’s selective price cutting placed Red C at a cost disadvantage relative to Standard, which received the substantially lower freight rate.59 5.1.2. McGee’s Explanation of the Case McGee’s analysis of the Red C case essentially relied only on the testimony of Fehsenfeld and Collings.60 He argued that Red C did the price cutting. McGee quoted Fehsenfeld who testified that he did cut price below Standard’s price when entering a territory dominated by Standard because ‘‘in going into a territory we would have to offer some inducement.’’61 McGee then accepted, without reservation, the testimony of Collings that Red C was always the price cutter, Standard never initiated price cuts, and, with one exception, Standard never reduced its prices as low as Red C’s prices. Finally, McGee clearly implied that Standard’s pricing behavior could not have injured Red C’s effectiveness as a competitor because ‘‘Red C interests had grown steadily and prosperous since their modest beginning in 1878 (p. 154).’’62 McGee accepted without equivocation Collings’ assertion that only in one case did Standard meet Red C’s lower prices. Yet McGee failed to note that Fehsenfeld contradicted McGee on the same page of testimony cited by McGee. Fehsenfeld testified that Standard ‘‘very often’’ cut prices below Red C after it undercut Standard’s prices (pp. 2340–2341). Moreover, Standard successfully signaled Red C that immediate re-entry was not a feasible strategy. According to Fehsenfeld, Red C might begin to sell again only when ‘‘the merchants had forgotten to a large extent, the experience of their previous purchases of me.’’ And these conditions prevailed for the entire time that Fehsenfeld was on the road, that is, 1878–1897 (p. 2301). McGee also ignored the implications of contradictions in Collings’ own testimony that showed Standard cutting price. At one point, Collings testified that he knew nothing about Red C’s cancellation of carload shipments for the towns of Pelzer, Greenville, and Laurens in 1904. Yet Collings acknowledged that he wrote a letter directing sales people to prevent Red C from getting carload shipments to these same towns. At another point, Collings did acknowledge one instance in which Standard undercut Red C’s price to keep the business of Union Cotton Mills, a large Standard customer (pp. 967–970). As noted above, Standard did cut prices when competing against Red C. The real question is how substantial were the cuts and what were their
Predatory Price Cutting and Standard Oil
171
impacts on competition. Consider the testimony about the behavior of Standard from its own employees. (a) Wofford had been employed by Standard during most of the period covered by Fehsenfeld’s testimony. He covered Alabama and Georgia, territories in which Red C operated (v. 5, pp. 2150–2157; Fehsenfeld, p. 2331). Wofford testified that Standard developed a database on the customers of competitors and used it to cut prices sufficiently so that rivals would make no sales (pp. 2150–2157). Collings did not contradict this testimony; he only stated that ‘‘The salesman had no authority to cut prices’’ without Collings’ authority. (b) A. N. Reed, Standard’s Special Agent in South Carolina, in 1904 instructed agent A. C. McGee to cut Standard’s price a half cent a gallon below Red C’s price in order to break up a carload.63 (c) J. D. Austin, a former salesman for Standard Oil, told Fehsenfeld that when he worked for Standard ‘‘his instructions were to stay at Yorkville [S.C.] for 30 days, if he could, to keep me from selling a carload of oil.’’64 (d) Mr. Greathead, Standard’s agent in Virginia, told Metzel, a salesman for Red C, that he had been instructed, ‘‘to undo as much of my [Metzel’s] business as he could,’’ otherwise he would lose his job.65 (e) W. Arthur, Standard’s commission agent in Atlanta, stated that after Red C sold there, Standard would make Red C sorry if it attempted to sell there again (Pet. Ex., p. 896). Standard not only engaged in pricing behavior aimed directly at diverting trade from Red C, it also evidently threatened to destroy the business of merchants buying from independent marketers like Red C by selling at a retail price that would eliminate any profits for local retailers buying from Red C. For example, in Charlotte, the merchants told Fehsenfeld that if they bought from Red C, then Standard ‘‘would cut prices so that they would not be able to make a profit (Fehsenfeld, pp. 2309–2310).’’ In summary, there is voluminous testimony in the Record that clearly contradicts McGee’s argument that Standard did not cut prices to prevent Red C from successfully marketing to Standard’s customers. The facts demonstrate that (a) when Red C entered the markets that Standard had monopolized, Standard used its salesmen and bogus companies to cut price below the price of Red C to prevent Red C from shipping in full carloads and competing on the same cost basis as Standard; (b) Standard reduced its price below its variable cost;66 (c) Standard intended to eliminate Red C with this pricing practice; and (d) Standard raised the price to the recaptured
172
JAMES A. DALTON AND LOUIS ESPOSITO
customers after Red C exited the market. Standard’s pricing behavior in the Red C case is consistent with our definition of predatory pricing and an even stricter definition that integrates intent, selective price cutting, and price– cost margins. If McGee examined the same Record that we did, why is his conclusion so different from the conclusion in the above paragraph? The answer lies in (a) his selection of the facts relevant for analysis, (b) his interpretation of the facts he selected, and (c) the implied weight that he assigned to the credibility of witnesses, including witnesses whom he did not cite. These factors reflect, at a minimum, selective use of the evidence. For example, when referring to Fehsenfeld’s statement that Red C cut prices, McGee focused only on Red C as a price cutter, an expected behavior for an entrant into a monopolist’s market. He ignored evidence regarding the entire scenario of Standard’s program to cut prices in order to eliminate Red C, and he ignored the implications of Standard’s use of bogus companies to target specific customers of Red C. McGee also ignored the testimony of numerous other witnesses, including former employees of Standard, who testified to Standard’s pricing policies and practices. In contrast, McGee accepted, without reservation, the testimony of Collings,67 an executive of Standard. Consequently, one can reasonably ask why Red C would have reduced its prices to the point that it had to exit markets because it could not generate a profit. McGee clearly did not address this fundamental question. Finally, consider McGee’s statement alleging long-time financial health for Red C. Contrary to McGee’s assertion, Red C was not a large company after 21 years in operation; it had only four wagons at the time of Fehsenfeld’s testimony in 1907 (Fehsenfeld, p. 2342). Given its relatively small size, it would be difficult to conclude that Red C had been very successful in an industry in which the production of crude oil products had grown at a compound annual rate of growth of 8.5 percent since 1878, the year in which Red C entered the marketing business, and 9.8 percent since 1888, the time that Fehsenfeld became an employee of Red C.68 In addition, McGee ignored the fact that Red C’s size and profitability derived in part from its marketing of non-petroleum products. In 1908 Red C’s product catalogue contained 84 pages (Fehsenfeld, pp. 2342–2344). Indeed, predatory pricing can be a profitable strategy simply by limiting the growth of wholesalers, thereby limiting the growth of the independent refiners supplying those wholesalers. Moreover, by focusing on Red C’s survival as a wholesaler McGee ignores the fact that Standard successfully eliminated Red C from geographic markets.
Predatory Price Cutting and Standard Oil
173
5.1.3. An Alternative Explanation A more plausible explanation of the facts of the Red C case employs a simple model of predatory price cutting in which the dominant firm intends to eliminate its competitor while minimizing its losses in the short run from any price reductions in order to charge higher prices in the long run. This simple model describes Standard’s pricing behavior in the Red C case, covering a period at least as long as 1886–1897 if not extending into the early 20th century.69 The key elements of this model are that Standard as the dominant firm: (a) prevented a rival firm from selling to the monopolist’s customers in the immediate and distant future by reducing its price below the price at which the entrant could efficiently compete, a price below Standard’s own variable cost; (b) minimized the impact on its own profits of reducing prices by using selective price cutting in order to eliminate a rival; (c) succeeded in its intent to eliminate the rival and in signaling to that rival not to re-enter the market; and (d) raised its price after the rival exited the market. 5.2. The Scofield, Shurmer, and Teagle (SS&T) Case McGee listed the SS&T case among his ‘‘suspect cases involving acquisitions’’ (p. 144). SS&T was primarily a marketer of refined oil products who competed with Standard in the Midwest. Standard acquired the company in 1901 and changed the name from Scofield, Shurmer, and Teagle to Republic Oil Company. 5.2.1. The Facts It is important to recognize that SS&T was primarily a marketer of products of independent refiners rather than a refiner (W.C. Teagle, v. 3, p. 1167). In fact, McGee noted testimony to the effect that ‘‘the capacity [of SS&T’s refinery in Cleveland] was limited as compared with the sales end of our business.’’70 If the refinery operated at its full 200,000 barrel annual capacity, it would have accounted for no more than 0.4 percent of the 49.7 million barrels of U.S. refinery output in 1899 (Williamson & Daum, 1959, p. 615). Further, since McGee cited the refinery’s manager as testifying that the refinery was ‘‘old and out of date,’’ McGee evidently accepted the fact that SS&T’s refinery was not a viable factor in the refining market (p. 152, fn. 67).
174
JAMES A. DALTON AND LOUIS ESPOSITO
Standard maintained a database of shipments by independent refiners to independent wholesalers in SS&T’s territory, and then of shipments by independent wholesalers to retailers. Salesmen in SS&T’s territory testified that Standard collected information on shipments by competitors from its own sales force and railroad employees; the ‘‘particulars’’ were filed regularly and frequently, even daily; Standard’s managers instructed salesmen either to get the orders countermanded before the competitor’s shipment arrived or to prevent the loss of orders to competitors in the future; and the salesmen were to offer the lower prices only to competitors of Standard.71 SS&T operated mainly as a wholesaler in the Missouri and eastern Kansas territories; its main competitors were Standard and National Oil.72 The competitive environment in SS&T’s geographic markets was succinctly described to be one in which a manager of Standard instructed salesmen in ‘‘an everyday affair, almost’’ to tell customers of rivals that Standard would sell for less.73 And, as discussed below, with these instructions salesmen were expected to act without timidity.74 A significant amount of testimony shows that Standard’s prices eliminated SS&T from markets, limited its market shares in other markets, and, at least in some if not many instances, were below its variable cost. Standard’s managers and salesmen, in addition to customers and competitors of SS&T, provided this information in testimony. Standard’s Managers. The extent of this price cutting was illustrated by W. H. Hawkins, Standard’s manager of tank wagons in SS&T’s territory in the 1890s. Hawkins testified that at one time he was providing rebates to a quarter of the approximate 1,000 retailers in Kansas City (v. 2, pp. 985–988). When he competed against SS&T, Hawkins received information on SS&T’s shipments because Standard ‘‘was always in touch with everything coming into Kansas City from their competitors’’ (v. 3, pp. 1242– 1245). Standard’s bogus wagons, Sutton Brothers and Excelsior, whose purpose was to regain the business lost to the competition, did ‘‘run out’’ the competition, after which the price reductions ceased and Standard raised prices (pp. 1242–1246). More generally, ‘‘It was just up to the salesman to do it. At the time when there was competition, the Standard Oil Company is pretty lenient in prices; they give a salesman pretty wide latitude’’ (p. 1247) to the extent that ‘‘while we had competition we made the price so low as to get the business y we would give a rebate of sufficient amount to get his business, that was all.’’75 The managers of Consolidated Tank Line Company (CTL) confirmed Standard’s pricing practices. CTL was an affiliate of Standard that competed
Predatory Price Cutting and Standard Oil
175
against SS&T (Hawkins, pp. 985–990). W. N. Davis, manager of CTL prior to 1901, testified that he told salesmen to get the business of competitors ‘‘In any way possible (v. 2, pp. 952–957).’’ E. J. Pratt, who in 1892 became manager for CTL in Kansas City, instructed salesmen ‘‘To make the price necessary to secure the business’’ and ‘‘The instructions were very liberal as regards [the methods of the salesmen] (v. 2, p. 959).’’ Standard’s Sales Agents. Salesmen of Standard described the practices that they implemented for Standard’s managers. They priced aggressively, without constraints, and expressed Standard’s intent to eliminate SS&T as a competitor. Musser was an agent for Standard in Kansas City during the period 1892– 1897. He confirmed that he used information in the database to identify customers of independents, and that his supervisors instructed him ‘‘to use any means’’ to prevent sales by independents (v. 2, pp. 968–970). E. M. Wilhoit had been an agent for Standard in Kansas City. He used the information from the database on competitors to reduce prices to customers of competing wholesalers while charging the higher price to Standard’s customers. He also testified that he was authorized to set his own price, ‘‘[i]n most cases – any price [to get the business].’’76 Bogus companies also used the information in the customer database to target the customers buying from Standard’s competitors; Wilhoit explicitly identified Excelsior Oil as a bogus company in the area (p. 1231). T. R. Hopkins was Standard’s agent in a territory in the Kansas City area. He testified that Waters-Pierce, Standard’s subsidiary in the area, had a virtual monopoly except for two customers which SS&T served. WatersPierce gave rebates to these customers and then raised the market price once SS&T had been suppressed (Hopkins, v. 3, pp. 1027–1031). H. J. Cohn testified that as a salesman for Waters-Pierce he reduced prices ‘‘far enough to get the business’’ from the independent (v. 2, pp. 97–98). H. C. Yungling testified that he was instructed to give rebates only to customers of SS&T (pp. 928–932) and ‘‘I got their business (pp. 931–932).’’ Standard’s manager at St. Joe instructed G. Kuenster to get SS&T’s business by reducing price, and on one occasion he reduced price from 12 cents a gallon to 5 cents a gallon when the cost of freight alone was 2 cents a gallon (Musser, v. 2, pp. 995–996). W. A. Morgan was instructed to ‘‘destroy all competition,’’ and to prevent SS&T from selling a barrel (v. 3, pp. 1004–1005); he used price cuts ‘‘as a method of getting the business’’ and, being ‘‘authorized y to cut the price whenever I saw fit,’’ he cut the price ‘‘fifty or a hundred’’ times (pp. 1021–1022). And Morgan went to the customers of the independents
176
JAMES A. DALTON AND LOUIS ESPOSITO
‘‘cutting the price, getting the business by any method I saw fit to pursue’’ (p. 1013), ‘‘I can’t recall any occasion where the [competition] met the price y because they couldn’t afford it. They didn’t make the money or profit that the trust was making (p. 1021).’’ Competitors and Customers of SS&T. A. L. Stocke, affiliated with the St. Louis Oil Company since 1878, competed against SS&T and Waters-Pierce. He bought refined product from SS&T and 15–20 other independent refiners. When St. Louis Oil opened its stations Waters-Pierce immediately began charging low prices and St. Louis Oil sometimes sold at a loss (v. 2, pp. 891–896). S. Lederer, a retailer at two locations in SS&T’s territory, bought refined product from Waters-Pierce. Whenever a competing retailer was able to reduce its price because it got a lower price from its wholesaler, WatersPierce allowed Lederer to reduce price and Waters-Pierce would rebate the difference to Lederer. He provided an example in which he reduced his retail price in several steps from 20 cents a gallon to 5 cents a gallon and was reimbursed (v. 3, pp. 1044–1046). He testified that ‘‘Every time they met my price I went still lower and when the wholesaler had eventually sold its carload over time I think they all lost money though (p. 1046).’’ The testimony presented above on the pricing behavior of Standard in SS&T’s territory clearly indicates that Standard intended to, and did, cut prices below the costs of independents, including SS&T, and below its own variable costs. In two instances, Standard or one of its companies lowered the price to 5 cents a gallon. Granitz and Klein estimated that in 1878 the cost of refined product (i.e., cost of crude oil plus the marginal refining cost) per gallon was 7.1 cents (1996, p. 20, fn. 53). Although crude prices and refining costs declined in the 1880s, it would appear that the cost of refined product during this period was not less than 5 cents per gallon.77 Since average variable cost is equal to the cost of crude oil, the marginal refining cost, and the transportation cost per gallon, it seems reasonable to conclude that on at least two occasions when Standard charged a price of 5 cents per gallon, its price was below its own variable cost. 5.2.2. McGee’s Explanation of the SS&T Case McGee stated that ‘‘there is no evidence that predatory price cutting had anything to do with the SS&T acquisition, or with terms on which Standard purchased it (p. 153).’’ In his very brief analysis, McGee characterized the SS&T case as an acquisition by Standard of a large refiner that had clashed with Standard
Predatory Price Cutting and Standard Oil
177
and in 1880 had broken a market sharing agreement with Standard into which it had entered in 1876. Obviously, neither point is relevant for an analysis as to whether Standard predated against SS&T. McGee also argued that because Standard, after the acquisition, continued to employ all but three employees of SS&T supported the conclusion that no coercion was involved in the acquisition. According to McGee, had there been coercion, then the ‘‘enemies’’ created would not have represented Republic as an independent (p. 152).78 McGee based his conclusion about the absence of predatory pricing on these several points. McGee’s argument that Standard’s retention of most of SS&T’s employees translates into the absence of coercive behavior by Standard prior to the acquisition fits neither common sense nor the simple facts. He argued that ‘‘there was no coercion surrounding the sale. For it is impossible to keep a secret when those who must be relied on to keep it can injure their enemies by betraying it (p. 152).’’ On the contrary, McGee should have asked why these employees would jeopardize their continued employment by betraying the secret; the common sense answer is that the employees had a stronger incentive not to betray Standard. This criticism notwithstanding, McGee’s point is inherently contradictory; revelations of coercive behavior by Standard reinforce the signal sent to potential entrants. McGee, moreover, neglected to mention that SS&T employees moved into very rewarding positions with Standard, reinforcing his argument that they were not enemies any longer.79 Most importantly, the facts in the Record do not support McGee’s conclusion that Standard did not predate. Whether predatory pricing caused the acquisition cannot be determined from analysis of the testimony of W. Teagle, the only witness cited by McGee. In fact, Teagle was clearly not qualified to discuss the history of pricing practices against an entire company over a long period of time because at the time of the merger in 1901 he had worked for SS&T for only two years in an untitled capacity and was only 23 years old. Indeed, Teagle essentially acknowledged his low subordinate status when he testified that around the time of the acquisition he had to assume that SS&T had sold to Standard, thereby implying that he was not a party to the negotiations regarding the sale (W. Teagle, pp. 1152–1156). McGee’s asking whether Standard’s pricing behavior affected either the decision to merge or the terms of the merger is too narrow a question; predatory pricing can also affect market shares of independents and the expectations of potential entrants. Regardless, information from the Record demonstrates that Standard priced below the costs of SS&T and probably below its own costs at times prior to Standard’s acquisition of SS&T.
178
JAMES A. DALTON AND LOUIS ESPOSITO
Finally, McGee’s reference to SS&T breaking an agreement with Standard in 1880 is puzzling. He did not explain its relevance. Our analysis of the Record suggests that SS&T entered into a market sharing agreement with Standard in 1876 in order to end Standard’s regular attempts to discipline aggressive rivals during the competitive struggle. Indeed, McGee failed to cite testimony by John D. Rockefeller that SS&T acknowledged in 1880 that the 1876 agreement restrained trade (v. 16, p. 3206), an admission consistent with the agreement that limited SS&T’s production to 85,000 barrels per year. Its refinery had a 200,000 barrel capacity. Given the market’s rapid rate of growth, an analyst should wonder why Standard guaranteed SS&T a profit to limit its output;80 evidently the refineries had excess capacity at the time.81 McGee’s analysis and discussion does not warrant the conclusion that Standard did not predate against SS&T. Two of his three justifications are irrelevant and the third justification is not consistent with his conclusion. He stated his conclusion without any recognition of the numerous pieces of evidence of below-cost pricing to eliminate competition in SS&T’s markets prior to its acquisition by Standard in 1901, and that behavior’s potential effects on the financial condition of SS&T and/or the terms of its acquisition by Standard. 5.2.3. An Alternative Explanation of the SS&T Case The Record does contain considerable information that Standard employed predatory pricing in SS&T’s geographic markets. Specifically, as in the Red C case, the available evidence in the Record with respect to Standard’s pricing behavior is consistent with a simple model of temporary selective price cutting aimed at disciplining and/or eliminating a rival after which Standard raised prices. It appears that the initial intent of Standard’s aggressive price cutting towards SS&T was to have SS&T limit its production. This goal was achieved by the 1876 agreement. SS&T’s abandonment of this agreement in 1880 was followed by a consistent and prolonged strategy by Standard of cutting price to SS&T’s customers with the intent to limit SS&T’s sales to Standard’s customers. In fact, on at least two occasions Standard priced below its own variable costs. Standard’s pricing strategy severely limited SS&T’s sales and growth during the 1880s and 1890s. In 1901 Standard acquired SS&T and operated it under the name Republic Oil.82
5.3. The Cornplanter Case McGee identified Cornplanter as a ‘‘suspect case’’ of predatory pricing that did not result either in a merger or an acquisition. Cornplanter, founded in
Predatory Price Cutting and Standard Oil
179
1888, was a refiner in Warren, Pennsylvania, and a marketer in Minnesota, New York, and Massachusetts (pp. 153, 155–156). 5.3.1. The Facts Cornplanter and Standard engaged in price wars in the three geographic areas. The Record contains considerable relevant factual detail that McGee omitted in his discussion of the Cornplanter case. This Record addresses Standard’s pricing in the three geographic areas in which Cornplanter marketed against Standard. Because of this structure of the Cornplanter case and the market-specific content of the information, we present this factual detail below in our evaluation of McGee’s conclusions regarding each of the three geographic marketing areas. 5.3.2. McGee’s Explanation of the Cornplanter Case McGee focused almost entirely on the testimony of W.D. Todd, Cornplanter’s manager (pp. 153–160). Todd testified that (a) Standard threatened to drive Cornplanter from the St. Paul market if Cornplanter did not agree to reduce output and coordinate pricing; (b) a Standard Oil executive had told him in 1898 that Standard’s goal was to put all independents out of business; and (c) Standard had started a price cutting war in New York. Todd further testified that Cornplanter had started a price cutting war in the Boston area that was resolved by a marketing agreement between Standard and Cornplanter. Prices rose from 612 cents to 10 cents a gallon within a few days of the agreement, and remained unchanged for a long period of time. The agreement lasted until 1906 when Cornplanter sold its Boston distribution facilities to the Gulf Refining Company. Todd also testified that the value of Cornplanter’s stock had increased from $10,000 to $450,000 between 1888 and 1908 (v. 6, pp. 3207–3235). McGee did not present analyses of Standard’s behavior in each of the three markets. One can infer from his statements, however, that McGee believed that Standard did not employ predatory pricing in the Cornplanter case: (a) Regarding Standard’s threat to drive Cornplanter from the market, McGee simply stated ‘‘The Standard threat never materialized (p. 159).’’ (b) He stated that the price war in Boston was costly to both sides and, consequently, Standard and Cornplanter signed a pricing agreement that lasted until Cornplanter sold its distribution facilities, but he did not specify a direct connection between this agreement and his conclusion. (c) McGee did not refer to any pricing behavior in New York State in his article. (d) Finally, McGee stated that despite ‘‘their difficulties with Standard, Cornplanter’s capital had grown, in the 20 years of its existence, from
180
JAMES A. DALTON AND LOUIS ESPOSITO
$10,000 to $450,000. Todd admitted that they were alive and, indeed, very healthy (pp. 155–156).’’ In other words, for McGee the Cornplanter case simply represented a case of aggressive price competition between two old rivals competing for market share. McGee’s analysis of the Cornplanter case was primarily a refutation of Todd’s testimony that Standard had threatened extinction and had engaged in price-cutting behavior to drive Cornplanter from the market. The additional information in the Record, however, goes beyond Todd’s testimony. In sections ‘‘Minnesota,’’ ‘‘New York,’’ and ‘‘Boston,’’ we analyze more fully the pricing behavior in the three geographic areas cited by McGee. Minnesota. That McGee ignored Crenshaw’s testimony is puzzling because Crenshaw provided significant information about Standard’s pricing behavior. Crenshaw was general manager of the sales department for Standard of Indiana and was responsible for setting prices in Minnesota for the period 1902–1905. He acknowledged that Standard lost money when it sold at 912 cents in Minneapolis at the time that its prices in Duluth were 12 cents a gallon, and that this loss was to beat the competition (v. 15, pp. 2385–2395). Crenshaw broadened the geographic scope of predatory pricing to the state of Minnesota when he acknowledged that Standard had sold oil at a loss in the state of Minnesota for most of 1903 and 1904 (pp. 2391–2393). Further, Crenshaw specifically identified Cornplanter as one of ‘‘the chief offenders’’ in Duluth to which Standard was responding when it reduced price and lost money for most of 1902 and 1904. Thus, according to Standard’s executive responsible for setting prices in Minnesota, Standard did sell at a loss in Minnesota when it encountered Cornplanter and other competitors. New York. Cornplanter operated Tiona Oil Company in Troy, New York. McGee did not present an analysis of the price war in New York. He simply noted that Todd testified that Standard had started a ‘‘price cutting war against Cornplanter in New York (p. 155).’’83 Based on our investigation of the Record, McGee either ignored or missed vital information that related directly to the issue of predatory pricing by Standard in New York. Todd testified (pp. 3215–3216) that a Mr. Jennings, purportedly a director of Standard, discussed the pricing situation in Troy with him. Todd testified that Jennings told him that Todd: had to take into consideration that the Standard Oil Company has to operate differently from what a small concern would. y We have a policy to pursue, and that is to make it just as difficult for an independent to put out oil as we possibly can; in other words, we want to drive them out of business if we can; if we can’t, why, we sometimes buy them
Predatory Price Cutting and Standard Oil
181
out, and sometimes we make a dicker; but our first move is to make it as expensive as we can. Now y you can readily see this, because if we didn’t, where would we be in a few years? The independents would have all the business.
Evidence in the Record supports Jennings’s description of Standard’s intent to meet competition with pricing behavior intended to drive its rivals from the New York market. L.J. Drake in 1903–1904 was the general sales agent for Standard in New York State and, as such, was in charge of Troy. He testified that Standard priced below its own costs ‘‘if we were forced to’’ (v. 2, pp. 762–772). L.T. Messner was the general manager of Cornplanter’s Tiona Oil Company. Before July 1900, no independents were wholesaling oil in Troy. Then Cornplanter’s Tiona Oil Company entered Troy in July 1900 charging 712 cents per gallon. Profit provided the motivation because Standard was charging 9 cents a gallon in Troy when it was also charging 712 cents a gallon in Binghamton despite identical freight costs for refined oil delivered to both locations. Standard and its bogus company, Troy Oil Works, went after the customers of Tiona with two one-half cent price cuts to 612 cents. Tiona matched these cuts. The viability of 612 cents per gallon is rejected by Messner. He described a meeting with a Mr. McKenzie, the manager of Standard’s Troy Oil Works, in which McKenzie told Messner that since the cost of kerosene delivered to Troy was 6 cents, he could not understand how a wholesaler could sell at a price that netted only a half cent over the cost of goods sold.84 This testimony clearly implies that Standard priced below its variable costs. Todd also testified that Cornplanter lost money in Troy.85 Messner also testified that Tiona lost customers to Standard’s secret price cuts (v. 20, pp. 44–69). In summary, Standard implemented a policy of predatory pricing in New York when it set secret price cuts, below its own variable costs, to regain customers lost to Cornplanter. This policy caused Cornplanter to incur substantial losses and to lose customers. Boston. McGee implied that there was no predation by Standard in Boston when he noted that Cornplanter had tripled its gallonage in Boston and then entered into an agreement with Standard to coordinate price and output in Boston (p. 155, fn. 91). McGee failed to note that this agreement caused Cornplanter to substantially reduce its gallonage. These facts, taken as a whole, suggest that Standard’s price cutting drove Cornplanter’s prices so low that Cornplanter realized that independent competition was no longer in its best interest.
182
JAMES A. DALTON AND LOUIS ESPOSITO
Todd testified that shortly after Cornplanter entered Boston, Standard reduced prices by one-half cent a gallon continuously from 10 cents a gallon until the tank wagon price fell to 612 cents a gallon (pp. 3216–3218). Standard continued to reduce its prices when Cornplanter began to compete in the communities outside Boston.86 Todd testified that Cornplanter lost $30,000 to $35,000 during this period when it achieved its maximum sales of 125 tank cars a month. At this point Cornplanter entered into the agreement with Standard. Under the agreement: (a) Cornplanter surrendered pricing autonomy by having its price linked directly to the price of crude oil; and (b) Cornplanter could not receive more than 26 tank cars a month in Boston and could not ship to areas outside of Boston. Prices then jumped to 1012 cents a gallon in the Boston area. In summary, Cornplanter lost money in its attempt to compete with Standard’s prices in the Boston area, and the arrangement with Standard forced Cornplanter to surrender 80 percent of its sales along with its pricing autonomy in the Boston area in return for higher stable prices profitable to both firms. These facts suggest that one cannot reject the hypothesis that Standard engaged in predatory pricing in the Boston area in order to eliminate price competition and achieve higher prices and higher profits in the long run. Indeed, the evidence indicates that Cornplanter did lose money and Standard probably priced below its own variable costs. Depending on the assumption about marginal refining costs, the variable cost of producing and transporting refined oil product in 1899 was between 5.1 cents and 6.1 cents per gallon.87 Further, the fact that Cornplanter acquiesced to the settlement that required Cornplanter to surrender significant sales clearly suggests that Cornplanter and Standard were losing money when Standard kept reducing its prices in the markets in and surrounding Boston. Finally, once its rival had been disciplined, Standard raised prices in the Boston market. Value of Cornplanter’s Capital. McGee stated without establishing relevance that Cornplanter’s capital grew from $10,000 to $450,000 between 1888 and 1908. He evidently implied that Cornplanter was not hurt by Standard’s pricing behavior. The real question, however, is whether Cornplanter would have experienced even greater growth in capital and profits during this period had Standard not employed its pricing policy in the three extensive geographic areas. This time period experienced substantial growth in worldwide oil sales. Finally, whereas McGee noted Cornplanter’s ‘‘numerous other oil interests (p. 156, fn. 95)’’ he did not analyze their contribution to Cornplanter’s value in 1908; this raises the question of how much of the
Predatory Price Cutting and Standard Oil
183
increase in Cornplanter’s value was the result of the success of their other oil interests outside of the competition against Standard in the wholesale market. 5.3.3. An Alternative Explanation of the Cornplanter Case Our analysis of the evidence in the Record indicates that Standard did engage in predatory pricing against Cornplanter in Minnesota, Troy, and Boston. Standard’s executive in charge of sales testified that Standard priced below its costs when it encountered Cornplanter in Minnesota. Standard’s agent in Troy testified that Standard used selective pricing below its costs after Cornplanter tried to enter the Troy market. Standard disciplined Cornplanter when Standard priced below its costs when Cornplanter entered Boston-area markets, causing Cornplanter to yield significant sales to Standard after which Standard raised the price. Once again, Standard’s pricing behavior in the Cornplanter case can best be understood using a simple model of selective price cutting aimed at disciplining or eliminating a rival.
5.4. The Rocky Mountain Oil Case According to McGee, ‘‘this is the nearest thing to predatory price cutting that I found in the Record. But there are reasons to suppose that there was something more here than meets the eye (p. 149).’’ 5.4.1. The Facts Continental Oil, a Standard subsidiary, had a virtual monopoly on the marketing of refined oil products in Colorado, Montana, Wyoming, New Mexico, Utah, and Idaho in the late 1880s and early 1890s.88 Continental Oil purchased virtually all of its refined oil products from the Florence Oil & Refinery and the United Oil Company, independent refineries in Florence, Colorado (W. Tilford, v. 1, p. 155).89 Several stockholders of the United Oil Company planned to build a new refinery in Colorado in about 1892. The owners of the planned refinery asked Continental Oil if it would be interested in purchasing the output of this new refinery.90 Wesley and Henry Tilford, representing the Continental Oil Company, informed them on two different occasions that, given the current size of the geographic market, Continental’s purchases of refined oil products from Florence Oil & Refinery and United Oil Company was ‘‘all the oil they could market in that vicinity (H. Tilford, pp. 730–731).’’
184
JAMES A. DALTON AND LOUIS ESPOSITO
Nevertheless, the refinery was built and operated as the Rocky Mountain Oil Company. Rocky Mountain Oil began marketing its refined oil products in competition with Continental Oil in 1892. According to H. Tilford, President of Continental Oil, Rocky Mountain cut prices below those of Continental as soon as it entered market (v. 2, pp. 730–731). The two pre-existing refiners made deep price cuts to Continental, the Standard-owned wholesaler, as a consequence of Rocky Mountain’s entry into a market virtually monopolized by Continental Oil.91 Prices apparently fell at least by 50 percent during the price war.92 According to W. Tilford, Treasurer of Standard Oil of New Jersey, Rocky Mountain sometimes ‘‘sold oil at less than the cost of freight and packages.’’ He denied that Continental ever sold at prices below the cost of freight and packages because, as he asserted, ‘‘our marketing facilities were much better than theirs (v. 1, p. 157),’’ an assertion that he did not attempt to support. W. Tilford further stated ‘‘that whenever they [Rocky Mountain] cut prices at points below cost, we didn’t follow at those points (p. 179).’’ W. Tilford further claimed that Rocky Mountain, unlike Continental, lost money as a consequence of this price war. Rocky Mountain Oil ceased selling refined oil products in the market and apparently disappeared after approximately two years. The price of refined oil products to Continental immediately increased, apparently back to 15 cents a gallon (p. 156). W. Tilford also testified that, during the price war with Rocky Mountain, the refineries had lowered their prices to Continental (p. 180). E. M. Wilhoit was the sales agent for Continental Oil in Kansas City, Kansas at this time and competed against Rocky Mountain Oil in marketing (v. 3, p. 1216). Wilhoit offered a different perspective on Continental’s pricing behavior during its competition with Rocky Mountain, namely that Continental may have reduced its price below Continental’s cost (v. 3, pp. 1215–1217). He testified that he was authorized to set any price to obtain the business and explicitly referred to Rocky Mountain Oil in Kansas City, Kansas. Wilhoit was not constrained by Continental’s costs or profits in his setting of prices when competing against Rocky Mountain. For example, Wilhoit testified that he was expected to tie up a jobber in Kansas City, Kansas to a contract with Standard Oil for a dollar a barrel less than Standard’s regular price in the area; this price represented a discount of 212 cents a gallon when ‘‘usually a cent a gallon secured the business (p. 1216).’’ When he did not secure this contract, Wilhoit then received a letter from Continental’s district office stating ‘‘you were sent to this point to contract with this party. There was no string tied to you. Make some price to get that business. Return to this point and tie this party under contract.’’93
Predatory Price Cutting and Standard Oil
185
5.4.2. McGee’s Explanation of the Rocky Mountain Case McGee briefly discussed some of the facts of the case and concluded that it did not represent an episode of predatory pricing but rather a temporary breakdown in a cartel-like arrangement between Continental Oil, a wholly owned marketing subsidiary of Standard Oil, and the local refiners from whom they purchased their refined oil products (p. 149). For McGee, wholesaler Continental Oil and the producers of refined oil products in the region had essentially formed a cartel. Continental monopolized the sale of refined oil products and shared those profits with its cartel partners, namely the refiners who supplied Continental. In exchange for their share of the monopoly profits, the refiners limited their production and did not market their output directly in competition with Continental. A subset of the owners of one of the cartel members increased refinery output when it formed Rocky Mountain Oil, which then attempted to sell its output to Continental. Since Continental was already selling the cartel’s profit maximizing output, it refused to purchase the additional output. Rocky Mountain then began to sell its product in direct competition with Continental at a price lower than Continental’s price. A price war ensued that lasted approximately two years before Rocky Mountain Oil left the market. McGee believed that the facts were consistent with his cartel hypothesis. For example, ‘‘Tilford did not regard Rocky Mountain as having failed, but simply as having been reabsorbed by those who started it.’’ And ‘‘after the reabsorption of Rocky Mountain,’’ the refiners increased the prices at which they sold their products to Continental. Thus, according to McGee, ‘‘the net result of the conflict is that United and Florence apparently emerged from the conflict with more favorable contract terms vis a vis Standard (p. 150).’’ McGee implies that, as a result of the conflict, United and Florence had increased their share of the cartel’s profits at the expense of Standard. McGee’s conclusion that the Rocky Mountain case did not represent a case of predatory pricing seems to be based primarily on this assertion and the fact that, according to McGee, other independent refiners entered the market following Rocky Mountain’s exit. The Record simply does not support McGee’s contention that the refiners emerged with more favorable contract terms. According to the Record, the price of illuminating oil in the market was 15 cents per gallon prior to Rocky Mountain’s entry into the market. The refiners from whom Continental purchased their refined oil products reduced their selling price to Continental during the price war and then raised their selling price to Continental after Rocky Mountain disappeared from the market (W. Tilford, p. 180). In fact, the market price returned to 15 cents per gallon after Rocky Mountain
186
JAMES A. DALTON AND LOUIS ESPOSITO
exited the market, the exact same price that existed prior to Rocky Mountain’s entry into the market. Thus, United and Florence, as well as the other refiners who were supplying Continental, do not appear to have emerged from the conflict with more favorable contract terms and thus did not increase their share of the cartel’s profits. With respect to the question of entry, it is true, as McGee asserts, that two new independent refineries were built in the region. However, those refineries, in Spring Valley and Boulder, were built in 1905–1906, 11 years after the price war that eliminated Rocky Mountain Oil (H. Tilford, pp. 731–732). 5.4.3. An Alternative Explanation of the Rocky Mountain Case Assume, as McGee suggests, that a cartel had existed between Continental and the oil refiners in the region. Then the attempt by Rocky Mountain Oil to sell to wholesalers competing against Continental, Standard’s wholly owned wholesaler, would have destroyed the cartel absent a response from Continental. If Continental had let the market price fall permanently to absorb the increased output of Rocky Mountain, it would have been encouraging other members of the cartel to increase their output and enter into direct competition with Continental in marketing their increased output. Thus, if Continental were to maintain the cartel arrangement with its suppliers and protect its monopoly position in the marketing of refined oil products, then it had to move swiftly to eliminate this new rival. Swift action would also serve to discourage any output expansion by other members of the cartel. Regardless of who initiated the first price cut, Continental continuously cut prices during the period from 1892 to 1894 until they may have fallen below cost for both Continental and Rocky Mountain. The fact that Rocky Mountain cut price first is to be expected given that they were entering a market monopolized by Continental and therefore is irrelevant to the issue of predatory price cutting. As stated earlier, W. Tilford testified that Rocky Mountain cut its prices below its costs but that Continental’s prices never fell below its costs during the price war. He claimed Continental did not have to cut prices below their cost because of better and more efficient marketing facilities. As a consequence, according to Tilford, Rocky Mountain lost money but Continental did not (p. 156). W. Tilford never mentioned during this part of his testimony that Continental’s suppliers had lowered their prices to Continental during the price war. Thus, it would appear that once the price war began, Continental decided that all the members of the cartel would share in any decrease in
Predatory Price Cutting and Standard Oil
187
profits that resulted. By forcing its suppliers to lower their selling prices, Continental effectively lowered its costs. Thus, Tilford’s claim that Continental never sold below cost or lost money during the price war with Rocky Mountain is probably more logically explained by the lower prices it forced upon its suppliers rather than his unsupported assertion that Continental had more efficient marketing facilities. In summary, the Record suggests that Standard Oil probably did engage in predatory pricing behavior during the period from 1892 to 1894 in Colorado and Wyoming where it was in direct competition with Rocky Mountain Oil. Standard’s pricing behavior was aimed (a) at eliminating a rival from the market; and (b) at signaling other potential entrants of Standard’s likely reaction to their entry. The Record shows that Standard did indeed succeed in eliminating Rocky Mountain as a competitor in the marketing of refined oil products in Colorado and Wyoming. The Record, especially Wilhoit’s testimony, indicates that Standard may well have priced below its cost during this period in order to protect its monopoly position in marketing.94 Thus, Standard reached its two-pronged goal by lowering the price temporarily and then increasing it after the elimination of Rocky Mountain. The Record also suggests that Standard’s pricing behavior discouraged new entry for a substantial period of time. Standard’s pricing behavior and its effects are consistent with modern strategy-based theories of predatory pricing.
6. CONCLUDING COMMENTS This paper examined McGee’s hypothesis, presented in his 1958 paper, that Standard Oil engaged in predatory pricing to acquire and maintain its monopoly in refining. McGee concluded that there was no evidence in the Record that Standard Oil engaged in predatory pricing in the refining and wholesale markets. We carefully analyzed four of the five major cases analyzed by McGee and found more than sufficient evidence to reject McGee’s conclusion that Standard Oil did not engage in predatory pricing. Indeed, we believe that the Record provides substantial evidence in each case that we analyzed supporting the hypothesis that Standard Oil did engage in predatory pricing. However, even if one were to conclude that the evidence in the four cases that we analyzed is not sufficient to conclude unequivocally that Standard Oil did engage in predatory pricing, one clearly cannot accept McGee’s conclusion that Standard did not engage in predatory pricing.
188
JAMES A. DALTON AND LOUIS ESPOSITO
We recognize that the data and information in the Record generally does not meet the current standard for evidence in predatory pricing cases, particularly with respect to price–cost margins and potential recoupment. Nevertheless, we believe that the evidence in the Record does support, at least, a strong prima facie argument that Standard Oil did engage in predatory pricing. As a consequence, an Attorney General today would find the evidence more than sufficient to inaugurate a full investigation into Standard’s pricing behavior. Nevertheless, in all four cases that we examined, we found that Standard Oil’s pricing behavior was consistent with strategic behavior aimed at increasing or sustaining market dominance by eliminating competition, deterring entry, and/or disciplining rivals. Bolton, Brodley, and Riordan have recently proposed a legal rule for the courts to use in predatory pricing cases (2000, p. 2264). The proposed rule would augment existing legal rules and practices. Their rule, which explicitly incorporates modern economic theories of strategic behavior and interaction, ‘‘would require proof of the following elements: (1) a facilitating market structure; (2) a scheme of predation and supporting evidence; (3) probable recoupment; (4) price below cost; and (5) absence of a business justification or efficiencies defense.’’ Table 1 summarizes our application of Bolton, Brodley, and Riordan’s legal rule for each of the cases we analyzed. Applying Bolton, Brodley, and Riordan’s proposed rule for determining whether a firm has engaged in predatory pricing, we conclude that Standard Oil clearly engaged in predatory pricing in the Red C, SS&T, and Table 1.
Application of Bolton, Brodley, and Riordan’s Legal Rule.
Elements
Facilitating marketing structure Scheme of predation
Probable recoupment Price below cost Bus. Just. Or efficiency defense
The Red C Case
SS&T Case
Cornplanter Case
Rocky Mountain Case
Dominant firm
Dominant firm
Dominant firm
Dominant firm
Selective price cutting – bogus wagons Yes
Selective price cutting
Selective price cutting
Select. price cutting
Yes
Yes
Yes
Yes None
Yes None
Yes None
Probable None
Predatory Price Cutting and Standard Oil
189
Cornplanter cases and probably engaged in predatory pricing in the Rocky Mountain case. In conclusion, our re-examination of the Record demonstrates that the Standard Oil case cannot and should not be used by economists and courts as evidence that predatory pricing is irrational and therefore rarely occurs. To the contrary, we believe that the Record provides ample evidence that Standard Oil did indeed engage in predatory pricing.
NOTES 1. Standard Oil Company of New Jersey v. United States, 221 U.S. 1 (1911), hereinafter called the Standard Oil Matter. The United States filed the complaint on November 15, 1906. The U.S. began taking testimony before a Special Examiner in 1908 and defendant Standard Oil Co. presented its direct case in 1909. On November 20, 1909, the District Court ruled for the U.S. The appeal then went directly to the Supreme Court; at that time, the special ‘‘Expediting Act’’ allowed the Department of Justice to bypass the Circuit Court in an appeal. Standard Oil argued its appeal in March 1910 and re-argued it in January 1911. On May 15, 1911, the Supreme Court upheld the decision of the lower court and ruled that Standard Oil had six months to divest its subsidiaries by transferring the stock from the trust back to the stockholders of the companies as originally constituted when acquired by Standard. In December 1911, 38 companies were spun off from Standard. 2. As cited in Bolton et al. (2000, p. 2243). In this same place, the Court also cited a paper by Koller (1971) who examined 40 cases of alleged predatory pricing. In the 23 cases that he judged to contain sufficient information, Koller concluded that predation was actually attempted in seven cases but succeeded in only four of them. Zerbe and Cooper (1982) challenged Koller’s findings. They examined the same 40 litigated cases as Koller and concluded that predatory pricing occurred in 27 cases. See also Zerbe and Mumford (1996) for a discussion of other empirical studies of predatory pricing, including a reclassification of Elzinga’s (1970) cases in the Gunpowder Trust. Whereas Elzinga found that a quarter of the eight cases for which adequate data were available involved predatory pricing. Zerbe and Mumford, using ‘‘a more generous definition of predatory pricing,’’ found that 47 percent of the 11 cases with data involved predatory pricing (p. 961). Zerbe and Mumford also point to the Supreme Court’s opinion that predatory pricing is not rational and rarely happens could have served as a barrier to such cases making their way into the courts (1996, pp. 951–956). The 10th Circuit in United States v. AMR Corp., 335 F. 3d 1109, 1115 (10th Cir. 2003) referred to post-McGee research that ‘‘challenged the notion that predatory pricing schemes are implausible and irrational’’ and acknowledged that it would ‘‘approach the matter with caution’’ [but] not y with the incredulity that once prevailed (as cited in Cross, 2003, p. 9).’’ See also the Court’s reliance on McGee’s work in Advo Inc. v. Philadelphia Newspapers, Inc., 51 F.3rd 1191, 1196 (1995) as cited in Zerbe and Mumford (1996, p. 957.) 3. In addition to the above examples in the text, consider the statements of other economists. Lott states: ‘‘McGee’s 1958 and 1980 papers probably present the
190
JAMES A. DALTON AND LOUIS ESPOSITO
best-known refutation of the frequent claim that Standard behaved like a predator, acquiring companies under threats of predation (1999, p. 8).’’ Armentano states: ‘‘Unfortunately, for lovers of legend, this one has been laid theoretically and empirically prostrate. In a now classic article, John McGee argued y there are serious logical weaknesses in the assumption that large firms are motivated to engage in predatory practices y The empirical evidence with respect to Standard Oil’s practices reinforces these theoretical predictions’’ (1996, p. 63). Boudreaux and Folsom state: ‘‘Did Rockefeller, once he had his large market share, engage in predatory price cutting? Absolutely not, says John McGee, who is an authority on this issue. According to McGee, Rockefeller avoided predatory strategies because they would have been costly and ineffective. The reasons for this conclusion are today well-known and need not be reviewed here (1999, p. 559).’’ 4. Elzinga and Mills attributed the ‘‘genesis’’ of the Chicago School’s position on predatory pricing to McGee’s articles (1958, 1980) and McGee’s teacher, Aaron Director (2001, p. 2476, fn. 11). 5. Posner essentially employed this definition of predation (1976, p. 188). Mariger (1978) examined the matter of Standard’s alleged predatory pricing and he produced ambiguous conclusions. He performed two empirical tests. First, Mariger found a simple positive correlation between price and market share for 31 cities in 1904. He noted that this result is consistent both for predatory pricing and for the dominant firm pricing model (pp. 345–346). Second, data limitations caused him to simulate pricing only for 11 cities in 1904. He concluded that he could not reject the dominant firm pricing model in that ‘‘our results cast some doubt on the traditional interpretation of Standard Oil’s pricing behavior’’ (p. 361). This test uses too few data points for a year long after Standard had acquired its dominant position in refining and marketing. Mariger also included two short summary paragraphs that convey his sense after reading the Record (p. 346). Then he opined that ‘‘It is not clear that Standard cut prices with the intention of either driving the competitor to bankruptcy or to a non-remunerative sale’’ (p. 347). He did not present any specific information to support his statement and omitted any consideration of the impact of Standard’s pricing as a signal to competitors and potential competitors. 6. A study by the U.S. Department of Commerce, Bureau of Corporations (1907) documented what it considered as numerous instances of predation. 7. Historians recognize the benefits of re-opening study of the past. Renowned historian James M. McPherson (2003) stated in an address as President of the American History Association: ‘‘Interpretations of the past are subject to change in response to new evidence, new questions asked of the evidence, new perspectives gained by the passage of time. There is no single, eternal, and immutable ‘truth’ about past events.’’ 8. See reports by the Bureau of Corporations (1906, 1907). 9. See also the concise summary of McGee’s model in Bolton et al. (2000, p. 2244). McGee also asserted that predatory pricing would not affect the acquisition price in a merger (1958, pp. 139–141). Burns (1986), however, found that predatory pricing by tobacco companies reduced the acquisition prices of tobacco firms. 10. In this context, Posner noted that whereas settlement is always cheaper than litigation in resolving legal disputes, litigation occurs because both parties to the dispute have different expectations as to the outcome of the litigation process (1976, p. 185).
Predatory Price Cutting and Standard Oil
191
11. Other economists have also recognized that a strategy of predatory behavior can indeed be successful in deterring entry and disciplining rivals when uncertainty and market imperfections are present (American Bar Association, 1996, p. 44; Salop, 1981; Klevorick, 1993). 12. For an analytically rigorous analysis of strategic interaction among firms that incorporates the effects of asymmetric information and capacity constraints on competition see Tirole (1988). 13. The discussion in Section 2.2 draws heavily on concepts and ideas contained in Milgrom and Roberts (1992, pp. 128–162); Bolton et al. (2000); Pepall, Richards, and Norman (2002, Ch. 6); and Baird, Gertner, and Picker (1994, pp. 178–187). Elzinga and Mills (2001), in contrast, have argued that whereas strategic theories of predatory pricing are plausible but fragile, the necessary supportive facts ‘‘are difficult, if not impossible, for courts to observe (2001, p. 2475).’’ 14. According to Baird, Gertner, and Picker: ‘‘Entrants in later markets observe the actions of Incumbent. They can infer that if Incumbent ever accommodated, Incumbent must be rational, but they cannot be certain when Incumbent predates whether Incumbent is aggressive or Incumbent is rational but mimicking the actions of the aggressive type in order to deter entry (1994, p. 181).’’ 15. Crawford recently argued that a rational player can exploit a bounded rational player by active misrepresentation or lying (Crawford, 2003, pp. 133–149). 16. Granitz and Klein (1996) demonstrated that the railroads gave Standard, and not its rival refiners, rebates on the product it shipped and drawbacks on the product shipped by rivals. Overall then, Standard probably had lower total monetary unit costs during the time that it enjoyed these rebates. However, as to the wholesale markets in which Standard operated, its real economic costs may not have been lower until it switched to delivery in bulk rather than in barrels and if its rivals could not generate the volume necessary to justify delivery in bulk. 17. Chernow (1999, p. 130). Rockefeller attributed ‘‘ruinous competition’’ in the industry to ‘‘the overdevelopment of the refining industry’’ [RAC, Inglis notes, 4.8, ‘‘Cleveland’’ (as cited in Chernow, 1999, p. 130, fn. 5)]. Consequently, Rockefeller argued that ‘‘The day of combination is here to stay. Individualism has gone, never to return’’ [RAC, Inglis notes, 4.8, ‘‘Genesis of the Standard Oil’’ (as cited in Chernow, 1999, p. 149, fn. 82)]. By 1873, ‘‘We proved that the producers’ and refiners’ associations were ropes of sand (Nevins, 1953, v. II, p. 16).’’ Historian Daniel Yergin illustrated this drive for order by noting that, whereas Rockefeller vertically integrated from refining into transportation and marketing, he stayed away from production of crude oil until the late 1880s because ‘‘It was too risky, too volatile, too speculative (1991, pp. 51–53).’’ 18. Reference is to an interview with John D. Rockefeller as reported in David F. Hawke, John D (as cited in Yergin, 1991, p. 42, fn. 7). 19. Nevins (1953, v. I, p. 370) and Yergin (1991, pp. 52–53). According to an unidentified successor to Rockefeller, a person who had worked as a young man with Rockefeller, the latter sought ‘‘orderly, economical efficient flow’’ and that this ‘‘orderliness would only proceed from a centralized control of large aggregations of plant and capital with the one aim of an orderly flow of products from the producer to the consumer’’ (Halliday, Rockefeller, p. 10 as cited in Yergin, 1991, p. 54, fn. 19).
192
JAMES A. DALTON AND LOUIS ESPOSITO
20. Chernow characterized the willingness of rivals to ‘‘cooperate’’ as follows: ‘‘The purchase of his competitors’ firms had not been the benevolent act that Rockefeller suggested, but he had a powerful selective memory (1999, p. xxii).’’ 21. Rockefeller portrayed the focus of his Executive Committee as follows: ‘‘You gentlemen on the ground can judge better than we on the matter, but let us not drift into arrangements where we cannot control the policy’’ (Hidy and Hidy, 1955, v. 1, p. 6, as cited in Yergin, 1991, p. 46, fn. 10). The senior management included 17 men, five of whom controlled 57 percent of the shares (Chernow, 1999, p. 46). 22. Chernow (1999, p. 256) and Yergin (1991, pp. 45–46). Section 5 below uses the Record to show how Standard developed and used this intelligence operation. 23. Yergin (1991, pp. 38–43, 51); Chernow (1999, p. 243); and Granitz and Klein (1996, p. 42, fn. 115). 24. Kerosene was the most important product. Other products included naptha, gasoline, fuel oil, lubricants, and petroleum jelly and paraffin. 25. The Standard Oil Company of New Jersey v. The United States, 221 U.S. 1 (1911). The list of defendants included the nine individual former trustees, and approximately 40 companies. 26. Nevins offered a similar conclusion in 1940 in his first two-volume biography of Rockefeller when he stated that Rockefeller’s ‘‘basic condition,’’ which was to establish and maintain order, ‘‘required him to warn off or drive out every possible entrant (v. II, p. 67).’’ 27. For example: (a) When an interviewer asked about deep price cuts when competing against rivals, Rockefeller stated: ‘‘These people did not want cooperation. They wanted competition. And when they got it they didn’t like it (RAC, Inglis Interview, 1 and 407 as cited in Chernow, 1999, p. 258, fn. 45).’’ (b) Colonel W. P. Thompson, an executive with Standard Oil, informed Rockefeller in 1886 that not only did Standard sell at cost where competitors appeared, but also that the practice was widespread. Thompson also implored the executives to understand that ‘‘The system is working well, better than any other we can devise and our feeling is to hold along on this basis. I beg you and the other gentlemen will keep in mind that we are selling 1/4 [sic] of all the oil we handle without a farthing of profit to the department.’’ Letter from W. P. Thompson to J. D. Rockefeller, RAC, 1.2B69F513, February 2, 1886 (as cited in Chernow, 1999, p. 258). (c) Chernow concluded from his study of the Rockefeller papers that ‘‘y Rockefeller’s files are so rife with references to this practice of predatory pricing’’ as to refute Allen Nevin’s verdict that only 37 of 37,000 towns serviced by Standard exhibited predatory pricing (1999, p. 257). (d) Rockefeller wrote that when refiners refuse ‘‘to sell out [to Standard] at good high prices y [t]hey will be sick unto death now having failed in their wicked scheme. A good sweating will be healthy for them and they ought to have it, and it is not money lost to us to have other people see them get it’’ Rockefeller Letter, March 11, 1878, Camden Papers (as cited in Nevins, 1940, v. II, p. 68). (e) Standard evidently also instituted ‘‘savage price wars’’ against its rivals for shares in Europe and India (Chernow, 1999, pp. 246–247, 259, 431; Yergin, 1991, pp. 61–62, 65, 68, 72, 75, 120, 126, 128). According to Chernow, ‘‘Standard ruled foreign markets no less dictatorially than domestic ones (1999, p. 244).’’ 28. The beginning of the time period also appears constrained by the formation of the Trust and the date of passage of the Sherman Antitrust Act.
Predatory Price Cutting and Standard Oil
193
29. Scherer has observed that even relatively recent antitrust cases ‘‘labor under such severe evidentiary difficulties that they frequently fail to elicit proof of the main point at issue (1970, p. 275).’’ 30. McGee did not present empirical evidence from the Record in his 1980 paper. 31. For example, Kahn has stated that ‘‘Predation is usually conceived as involving the successful attempt to eliminate competition by driving a competitor from the market. It ought equally to embrace attempts to discipline a price-cutting competitor and convert it to a follower y. In either event, the essential ingredient is the same: a successful investment in the preservation or achievement of monopoly power (1991, p. 144).’’ 32. In a general equilibrium analysis, consumer surplus across all markets will fall (Just & Rausser, 2007). 33. An anonymous referee reminded us that deadweight loss occurs when price falls below costs. 34. In this context, Edlin and Farrell have opined: ‘‘If an incumbent is strategically making (or threatening to make) limited good offers with the purpose and effect of stifling competition – competition that would give more in the long run and/ or enhance efficiency – that’s bad (2004, p. 525).’’ 35. See also Edlin (2002), Edlin and Farrell (2004), Kahn (1991), Shepherd and Shepherd (2004), and Scherer (1976a, 1976b) for a similar emphasis on competition for a determination of whether a dominant firm employed predatory pricing. 36. With the elimination of the price competition, the dominant firm recoups a return on its investment that occurred when it sacrificed short-run profits. That sacrifice is consistent with a strategy of maximizing profits over the long run. Edlin and Farrell (2004) have argued that the sacrifice of short-run profits presumes expected recoupment. 37. Scherer (1976a, 1976b) prescribed a flexible, detailed analysis of alleged predatory pricing that is reflected in a combination, in part, of our measure and these supplemental considerations. 38. In addition, according to Shepherd and Shepherd, ‘‘Selective price action is anticompetitive when it gives dominant firms a catalogue of weapons that are unavailable to little rivals (Shepherd & Shepherd, 2004, p. 217).’’ 39. Greer (1979). Kahn has argued that ‘‘I would have the antitrust laws place heavy emphasis on the intent that can be inferred from the actions and policies under examination. Heavy but not decisive. I would not condemn on the basis of intent alone. (1991, p. 142).’’ In her study of the British shipping cartel’s pricing behavior between 1879 and 1929, Morton examined limited information on pricing behavior across various cases to determine whether ‘‘the combination of events is more consistent with a deliberate price war than with the interaction of supply and demand.’’ She argued that ‘‘Establishing predatory intent is an important part of the qualitative argument’’ (1977, pp. 692–693). McGee did not opine about the relevance of intent for his analysis in 1958. Twenty-two years later, however, he stated that ‘‘the intent of the price cutter and the mental state of his rivals are difficult to establish and are ambiguous. They are largely irrelevant; it effects that matter (1980, p. 292).’’ 40. The majority opinion of the Supreme Court in this case concluded that Standard intended to monopolize the refining and marketing of refined products. The Court did not address specific cases as McGee did. Rather, it addressed
194
JAMES A. DALTON AND LOUIS ESPOSITO
Standard’s basic methods for the formation and operation of the Trust. According to the Court, theses methods are a clear indication of Standard’s intent to monopolize. Chief Justice White, in delivering the opinion of the Court, wrote: ‘‘Recurring to the acts done by the individuals or corporations who were mainly instrumental in bringing about the expansion of the New Jersey corporation during the period prior to the formation of the trust agreements of 1879 and 1882, including those agreements, not for the purpose of weighing the substantial merit of the numerous charges of wrongdoing made during such period, but solely as an aid for discovering intent and purpose, we think no disinterested mind can survey the period in question without being irresistibly driven to the conclusion that the very genius for commercial development an organization which it would seem was manifested from the beginning soon begot an intent and purpose to exclude others which was frequently manifested by acts and dealings wholly inconsistent with the theory that they were made with the single conception of advancing the development of business power by usual methods, but which on the contrary necessarily involved the intent to drive others from the field and to exclude them from their right to trade and thus accomplish the mastery which was the end in view’’ (Standard Oil Company of N.J. et al v. U.S., 221 U.S. 1; S. Ct 502; 1911). 41. Bolton et al. (2000) discuss the various proposed cost-based measures of predation. 42. The lower court and the Supreme Court ruled only on the issue of Standard’s acquisition and maintenance of its monopoly through the formation and use of the Trust; the courts did not address the government’s allegations regarding ‘‘unfair’’ pricing. Consequently, the respective decisions did not contain a ‘‘finding of facts’’ regarding Standard’s alleged selective price cutting. 43. McGee also addressed the role of ‘‘bogus’’ companies, although he did not treat the topic as a specific case. We analyze the role of bogus companies in Standard’s pricing practices for the four cases analyzed herein. The use of ‘‘bogus’’ companies to maintain market power by a dominant firm is not an uncommon practice. Brevoort and Marvel (2004) concluded that NCR priced knock-offs of rivals’ machines (‘‘knockers’’) below its costs. See also Morton (1997) and Yamey (1972) for the use of bogus ‘‘fighting ships’’ by the shipping conference cartel to price below costs of shippers who cheated on the cartel’s pricing agreements. 44. Regardless, the information in the Record regarding these acquisitions is scanty, at best. McGee (1958) argued that merger was a lower cost alternative to predatory pricing and that Standard paid a premium to purchase rival refineries . As to the asserted use of premiums, selective price cutting that leaves unaffected the price for the bulk of the dominant firm’s sales can protect the value of those sales and make the acquired capacity of rivals more valuable to the dominant firm than to the smaller, threatened rival. Further, Granitz and Klein (1996) found that Standard was able to reduce the value of rival refineries when the rebates it received in return for its support of the railroad cartel effectively increased the transportation costs of its rivals thereby reducing the market value of acquired refineries. 45. McGee incorrectly identified Red C as an integrated refining company. The Red C, established in 1878, marketed refined oil products in Maryland, Virginia, the Carolinas and, to a limited extent, Georgia. The Island Petroleum Company, established as a refining company in 1901, owned some of the stock in Red C and other
Predatory Price Cutting and Standard Oil
195
marketing companies; W. H. Fehsenfeld was the only common stockholder in Red C and Island Petroleum. Before 1901, the Red C had bought its product from independent refiners in Baltimore; after 1901, Red C purchased product from Island Petroleum and other independents. Red C did not purchase product from Standard Oil. (Fehsenfeld, v. 5, pp. 2326–2343) 46. The style of citations presents the volume number and page number from the Record with the first cite of the witness. Subsequent cites of the respective witnesses contain only page numbers. 47. McGee (1958, p. 153); and Fehsenfeld , p. 2303. 48. Fehsenfeld, p. 2303. A carload was charged at the 6th class rate while the lessthan-carload rate was charged at the 3rd class rate (Fehsenfeld, pp. 2303–2305, 2343). Red C also may have been disadvantaged for not having storage facilities near customers but this also meant that Red C had lower marketing costs; the Record did not reveal the net effect, if any, of these factors (Fehsenfeld, pp. 2338–2343). After 1897 when Standard began to increase shipments in tank cars, Red C faced a ‘‘disadvantage’’ when it shipped in barrels (Fehsenfeld, p. 2333). 49. Collings (v. 5, pp. 895–896) and Fehsenfeld (pp. 2340–2341). 50. Based on his ‘‘own knowledge’’ from 16 to 20 years of experience, Fehsenfeld knew about these data from a combination of his own observations, his customers, reports from his salesmen in the markets, freight people, and Standard’s own documents. Red C employed Fehsenfeld as a traveling salesman from 1886 to 1892 and then as head salesman and assistant manager until 1897, when he became secretary and treasurer. In 1903, he became the president of Red C. In all, he had worked for Red C for 21 years prior to his testimony. The content of Standard’s information was such that Fehsenfeld’s customers could tell him the details of time and terminals of Red C’s shipments, such incidents being ‘‘common, almost universal’’ (pp. 2303– 2311, 2326–2329; see also Petitioner’s Exhibits, pp. 715–750). For a first-hand description of representative illustrations of Standard’s database, see Fehsenfeld (p. 2308) and W. J. Metzel (v. 5, p. 2407). Collings, too, acknowledged that upon learning that Red C had cut prices in South Carolina, Standard sent a salesman to ascertain these price cuts (p. 895). 51. Fehsenfeld discussed at length the activities of the first five companies (pp. 2311–2327); Metzel identified all seven bogus companies (p. 2406). McGee implied that Metzel ‘‘erroneously accused Standard of owning and operating [bogus] firms that it did not have anything to do with (1958, p. 158).’’ McGee cited to Metzel’s testimony on p. 2412, yet McGee did not support this assertion. In fact, Metzel testified at that point that he saw Paragon’s bills that proved Paragon’s wholesale business was owned by Standard. 52. The Record shows that the bogus company was generally perceived to be independent of Standard regardless of whether the operator of the bogus company deliberately represented it as independent. The focus should be on the role that a bogus company played in Standard’s strategy and its impact on pricing and on Red C, regardless of any instructions to deliberately hold a bogus company out as independent of Standard. 53. Standard abandoned this program around 1905–1906 when legal challenges caused Standard to stop holding out affiliated companies as independents (Westcott, p. 728).
196
JAMES A. DALTON AND LOUIS ESPOSITO
54. In general, see McGee (1958, pp. 158–159), Fehsenfeld, pp. 2311–2325 and Metzel, p. 2407–2415. 55. One employee, Blaustein, worked with Eagle, Eureka and Southern. (Fehsenfeld, pp. 2311–2315; and Metzel, pp. 2407–2414). 56. Fehsenfeld, pp. 2326–2329; see also Petitioner’s Exhibits, pp. 715–750 and Metzel, pp. 2406–2411, 2416. 57. Fehsenfeld, pp. 2311–2325 and Metzel, pp. 2407–2415. In addition, Fehsenfeld testified that bogus companies caused Red C to increase compensation for its employees thereby driving up their operating costs; and Metzel testified that Dixie induced customers to switch by offering Red C’s customers inducements such as purchases of merchandise from the retailers. 58. Fehsenfeld, p. 2319 and Metzel, pp. 2415–2416. 59. Fehsenfeld, p. 2303. Metzel also testified to this behavior and the impact of less than carload rates. He had worked for Red C as a salesman from 1890 to the time of his testimony. Metzel identified Standard as his principal competition. He covered Maryland, Virginia, and parts of Pennsylvania, Delaware, and West Virginia (p. 2406). 60. In 1886, Collings joined Standard Oil of Kentucky, a marketing subsidiary of the Trust, as a manager of traveling salesmen, progressing to second vice president in 1900. 61. Fehsenfeld, p. 2341. This ‘‘inducement’’ helped cover the cost of returning barrels in contrast to purchasing from a tank car. 62. McGee did not discuss the allegations regarding the behavior and impact of the bogus companies with respect to Red C. However, he did address bogus companies later in his article without reference to the Red C case (pp. 158–159). At that later point he merely acknowledged that Standard did operate bogus wagons and implied that by appearing as independents when they charged lower prices to customers of Standard’s rivals, the bogus wagons would not antagonize Standard’s customers who were paying higher prices. Standard had the capacity to increase its output when it priced aggressively. The bogus wagons delivered the product to the customers they took away from Standard’s rivals. And the Record did not contain any information indicating that Standard could not increase its supply when it aggressively priced its own branded product in order to take business from its rivals. 63. Collings, pp. 959–962, and Petitioner’s Exhibits, pp. 686 and 690. In the latter exhibit, Reed instructed McGee to cut price a half cent below Red C’s price in order to break up a carload. Collings asserted that Reed was only calling for a price reduction that still left Standard’s price higher than Red C’s price. Yet Collings did not attempt to explain how Standard’s higher price would keep a customer from choosing Red C’s product. Moreover, the calculations introduced in crossexamination show that if Collings were correct about the higher price in this instance, the true difference would have left Standard’s price 3 cents a barrel higher (pp. 966–967). 64. Petitioner’s Exhibits, pp. 751–752, contain the communications from Austin as cited by Fehsenfeld (p. 2308). 65. Metzel, p. 2407. Metzel was a traveling salesman for Red C from 1890 through his testimony in 1908; his territory covered Maryland, Virginia, and parts of Pennsylvania, Delaware, and West Virginia.
Predatory Price Cutting and Standard Oil
197
66. At the time of entry, Red C was an efficient competitor because it succeeded in obtaining a full carload, thus matching Standard’s transport costs. Standard reduced its price below Red C’s price and its own variable costs when it recaptured customers when Red C priced in expectation of obtaining the full carload tariff. Red C had been an efficient entrant when it priced to obtain a full carload. Standard cut its price until Red C could not match it thereby leaving Red C with less than a full carload, which raised Red C’s costs above the efficient level. And then there is the additional matter of less-than-full cost pricing by the bogus companies. 67. Indeed, Collings’ testimony was not credible. For example, Collings testified: ‘‘I don’t know’’ when asked whether merchants would buy from Red C when Red C matched the price of Standard (p. 964). Also he had earlier testified that a competitor would have to set price below Standard’s price in order to make the sale (pp. 896–897). And, with the exception of the competition for the business of Union Cotton Mills, Collings denied that Standard had cut Red C’s prices without rebutting the specific testimony of Fehsenfeld and Metzel who had testified prior to Collings. 68. Total production of crude oil in barrels was 15.4 mm in 1878, 27.6 mm in 1888, and 178.5 mm in 1908 (U.S. Bureau of the Census, 1976). 69. The testimony does not go back prior to 1886 at which point Red C employed Fehsenfeld. 70. Teagle as cited in McGee, p. 152, fn. 67. 71. Ackert,v. 3, pp. 1106–1111; Cohn, v. 2, pp. 906–909; Davis, v. 2, pp. 952–954, 955–967; Hawkins, v. 2, pp. 85–993; Mayer, v. 3, pp. 1239–1240; Morgan, v. 3, pp. 1005–1006, 1009–1110, 1021; Musser, v. 2, pp. 969–970; Pratt, v. 2, pp. 959–960; Stewart, v. 2, pp. 994–995; Von Harten, v. 2, pp. 927–928; Wilhoit, v. 3, pp. 1037, 1212–1228, 1255–1264; Jungling, v. 2, pp. 928–932. 72. Nevins characterized SS&T as a persistent competitor of Standard (Nevins, 1953, v. II, pp. 54–58). 73. Cohn, v. 2, pp. 909–910. Cohn worked for Waters-Pierce, an affiliate of Standard that operated extensively in Missouri. 74. The remainder of this section discusses price cutting initiated by Standard that clearly rebuts the testimony of C. P. Ackert who had been with Waters-Pierce from 1889, becoming General Manager in 1899. Ackert testified that he instructed his salesmen not to undercut the prices of SS&T, only to meet SS&T’s prices (pp. 1090– 1094). 75. Hawkins, pp. 1243–1244. Hawkins noted that Standard used a rebate ‘‘For the purpose of hiding it.’’ The customer paid the full market price but Standard later rebated the amount of the price cut. 76. Wilhoit, v. 3, pp. 1215–1216, 1033, 1037. Drake, who was in overall charge of Wilhoit’s territory, testified that Wilhoit did not have such control of pricing. However, Drake did not direct Wilhoit. Mayer, the local manager in Kansas City, directed Wilhoit when Drake operated out of Chicago and Omaha (Drake, v. 13, p. 1073; Wilhoit, v. 3, p. 1215). Moreover, Morgan, Standard’s salesman in this territory in 1898–1901, testified that he had been instructed to cut prices against the competitors there (v. 3, pp. 1003–1014). 77. This estimate is based on data on crude prices and refining costs in Williamson and Daum (1959), pp. 483, 484, 566.
198
JAMES A. DALTON AND LOUIS ESPOSITO
78. Standard changed the name from SS&T to Republic after the acquisition, and the sales people then represented Republic as a company that was independent of Standard. 79. After the acquisition, Walter Teagle became General Manager of Republic and benefited from the acquisition even more so when he became a Director at 31 before becoming CEO of Standard Oil. Walter was in his early twenties at the time of the acquisition and had been working without a title under SS&T’s general manager. Henry Teagle became a manager in Republic; H. D. Whelan, moved from cashier to acting manager; and A. G. Shiers became an assistant manager. W. C. Teagle, v. 3, pp. 1149, 1157; C. L. Nichols, v. 3, pp. 1158–1159; H. D. Whelan, v. 2, pp. 972–973; and H. G. Shiers, v. 2, p. 978. 80. Standard guaranteed a profit of $35,000 a year to SS&T (Rockefeller, p. 3204). Standard entered into similar secret agreements with Pioneer Oil and other independent refiners (Rockefeller, pp. 3207, 3211–3212). 81. Rockefeller biographer Nevins stated that SS&T regularly exceeded the 85,000 barrel limit and that Standard was concerned that the increase in supply would threaten its profits at a time in which the market had considerable excess capacity (Nevins, 1940, v. I, pp. 366–367, 373–374; v. II, pp. 45–49; 1953, v. II, pp. 69ff). In the ensuing court case, the Court agreed that the agreement restrained trade (Chernow, 1999, p. 196). 82. McGee stated, ‘‘There is no real evidence that the bogus Republic waged predatory pricing campaigns, either (p. 153, fn. 73).’’ McGee did not present information to support his conclusion that Republic Oil did not use predatory pricing. McGee also argued that Standard did not operate Republic Oil as a bogus firm after Standard acquired SS&T and changed its name (p. 152). The question as to whether Standard used the Republic Oil Company as a bogus company in order to predate does not bear on the question of whether Standard’s pricing behavior either drove SS&T to merge or otherwise affected the terms of the acquisition. The decision by SS&T to sell out occurred before Republic came into existence as the successor to SS&T. There was sufficient information from our study of the SS&T case to analyze the Republic matter. Our analysis is presented in the Appendix. 83. McGee only implied that Cornplanter retaliated to Standard’s behavior in New York by starting a ‘‘costly’’ price war in Boston. McGee did not justify this assertion and the Record does not shed any light on this connection. 84. Messner, v. 20, pp. 44–69. See also McMillan, v. 12, p. 720. 85. This corroborated Todd’s separate testimony that Cornplanter lost money when Standard cut prices in Troy, Binghamton, Oneonta, Utica, and five towns around Binghamton. Subsequently, Cornplanter sold five stations around Binghamton to Standard (pp. 3220–3221). 86. Todd further testified that (1) Cornplanter was wrong not to expect Standard to compete aggressively outside Boston, an area in which Cornplanter entered with two to three tank wagons (p. 3229); and (2) Cornplanter did not cut below Standard’s prices (p. 3230). 87. This estimate is based on data on crude prices, refining costs, and transportation costs in Williamson and Daum (1959, pp. 483 and 623). 88. Testimony of Wesley Tilford, v. 1, p. 179. According to the government, Continental had 98–99 percent of the market (Brief of the United States, v. 2, p. 157).
Predatory Price Cutting and Standard Oil
199
89. McGee discussed the case in his section dealing with refiners, however it is a case involving Standard’s marketing at wholesale in a geographic market in which Standard did not operate a refinery. 90. McGee stated that Rocky Mountain Oil was started by United and Florence. According to the Record, however, Rocky Mountain Oil was started by several stockholders of the United Oil Company (W. Tilford, pp. 155–156). 91. Brief of the United States, v. 2, p. 157; and testimony of H. Tilford v. 2, pp. 729–731. 92. The price of refined oil fell from 15 cents per gallon to 7 cents a gallon, or even lower, during this period (W. Tilford, p. 179). Note also that the government attorney in cross-examination stated that where Standard Oil or one of its subsidiaries monopolized a geographic market, the average price was 15 cents per gallon or higher and in markets where it faced competition, its prices were 8 cents a gallon or lower (as asked in testimony of W. Tilford, p. 180). 93. Ibid., p. 1217. Wilhoit also stated: ‘‘Before the Refinery got in operation, the Standard Oil made our fight on the Rocky Mountain Company and the Union Pacific went into the hands of a receiver and the Refinery was afterwards dismantled – went broke of course (p. 1216).’’ 94. Continental’s costs fell after the entry by Rocky Mountain because refiners reduced the price to Continental. 95. See the testimony of executives and managers of Standard: Teagle, v. 3, pp. 1154–1156; Nichols, v. 3, pp. 1161–1164; Cochrane, v. 3, p. 1181; Finlay, v. 3, pp. 1133–1139; Heyer, v. 3, p. 1165; Drake, v. 2, pp. 753–756; Steigerwald, v. 3, pp. 1041–1043; Northrup, v. 3, pp. 1047–1048; Whelan, v. 2, pp. 973–976. Other employees of Standard testified to the bogus status of Republic: Turrell, v. 3, pp. 1051–1054; Jockel, v. 3, pp. 1062–1063; and Hardcastle, v. 3, pp. 1063–1064. For corroboration by competitors of Standard, see Gardner, v. 2, pp. 943–949; and Wilhoit, v. 3, pp. 1033–1035. For corroboration by customers, see Phipps, v. 2, pp. 997–998; Grace, v. 2, pp. 998–999; and Weinberger, v. 2, pp. 999–1001. 96. Of course, to operate as an independent that did not seek the trade from Standard’s customers does not mean that Republic would have rejected the business of Standard’s customers who approached this seeming independent. It is also reasonable for Republic to have tried to keep the customers that had been served by SS&T. And it was reasonable for Standard to have used Republic Oil to price discriminate in order to protect the revenues of Waters-Pierce, Standard’s affiliate in the territory, and Standard’s own wholesaler, the former Consolidated Tank Line Company. 97. Cytron, v. 2, p. 942; Gardner, v. 2, pp. 945–952; Stocke, v. 2, pp. 892–895; Shiers, v. 2, pp. 979–984; Turrell, v. 3, pp. 1051–1052; Whelan, v. 2, p. 976; L. J. Drake, v. 2, p. 755 [Drake was in charge of sales for Standard Oil of Indiana for which Republic sold refined product (pp. 746–753)]. See also Chernow (1999, p. 255); Yergin (1991, p. 97); Nevins, (1940, v. II, pp. 531–532). 98. Most of the testimony regarding this territory was read into the Record from an earlier case – State of Missouri v. Standard Oil, et al. (218 Mo., 1). The interrogation in this case elicited even less information on price–cost relationships than the case brought by the United States. 99. Gardner, v. 2, pp. 946–950; and Crenshaw, v. 3, pp. 1173–1174.
200
JAMES A. DALTON AND LOUIS ESPOSITO
100. W. C. Teagle testified that National had started the war when it reduced a price to obtain a Standard customer. An initial price cut is not a price war and it can be a sign of competition. Gardner indicated that Republic had driven the price down even further. According to Gardner, Crenshaw acknowledged that Teagle, the General Manager of Republic, had driven prices too low (v. 2, pp. 947–951). 101. G. W. Mayer, a manager for Standard in Kansas City, set up the meeting between Teagle and Gardner because Mayer wanted to end the price war that had ‘‘demoralized the trade’’ and caused Standard to lose customers. Mayer did not hear Crenshaw say that had he known about these deep cuts he would have stopped them. However, the Record did not establish that Mayer was present when Crenshaw made this observation. Gardner testified that Crenshaw delivered this statement when he and Crenshaw were on their way to meet Mayer and Teagle; Mayer confirmed that he did not meet with them until they arrived at the room where the meeting took place and that he did not know what had been discussed as the two men proceeded to the meeting place (pp. 1178–1180).
REFERENCES Akerlof, G. (2002). Behavioral macroeconomics and macroeconomic behavior. American Economic Review, 92(3), 411–433. American Bar Association. (1996). Section on Antitrust Law, Monograph 22. Predatory Pricing. Chicago: American Bar Association. Armentano, D. T. (1996). Antitrust and monopoly: Anatomy of a policy failure (2nd ed.). New York and London: Holmes and Meier. Baird, D. G., Gertner, R. H., & Picker, R. C. (1994). Game theory and the law. Cambridge, MA: Harvard University Press. Baumol, W. J. (1979). Quasi-permanence of price reductions: A policy for prevention of predatory pricing. Yale Law Journal, 89(1), 1–26. Bolton, P., Brodley, J. F., & Riordan, M. H. (2000). Predatory pricing: Strategic theory and legal policy. Georgetown Law Journal, 88(8), 2239–2330. Boudreaux, D. J., & Folsom, B. W. (1999). Microsoft and standard oil: Radical lessons for antitrust reform. Antitrust Bulletin, 44(3), 555–576. Brevoort, K., & Marvel, H. P. (2004). Successful monopolization through predation: The national cash register company. In: J. Kirkwood (Ed.), Research in Law and Economics, (Vol. 21, pp. 85–126). Amsterdam: Elsevier. Burns, M. (1986). Predatory pricing and the acquisition cost of competitors. The Journal of Political Economy, 94(2), 266–296. Chernow, R. (1999). Titan: The life of John D. Rockefeller, Sr.. New York: Random House. Crawford, V. P. (2003). Lying for strategic advantage: Rational and boundedly rational misrepresentations of interactions. American Economic Review, 93, 133–149. Cross, W. D. (2003). What’s up with Section 2? Antitrust Magazine, 18(Fall), 8–11. Edlin, A. (2002). Stopping above-cost predatory pricing. Yale Law Journal, 111, 941–991. Edlin, A., & Farrell, J. (2004). The American airlines case: A chance to clarify predation policy (2001). The Antitrust Revolution. New York: Oxford University Press. Elzinga, K. G. (1970). Predatory pricing: The case of the gunpowder trust. Journal of Law and Economics, 13, 223–240.
Predatory Price Cutting and Standard Oil
201
Elzinga, K. G., & Mills, D. E. (2001). Predatory pricing and strategic theory. Georgetown Law Journal, 89, 2475–2494. Friedman, D. D. (2000). Law’s order: What economics has to do with law and why it matters. Princeton, NJ: Princeton University Press. Garfield, J. R. (1906). Report of the commissioner of corporations on the transportation of petroleum, May 2, 1906. Washington: G.P.O. Granitz, E., & Klein, B. (1996). Monopolization by ‘Raising Rivals’ costs’: The standard oil case. Journal of Law and Economics, 39(1), 1–47. Greer, D. F. (1979). A critique of Areeda and Turner’s standard for predatory practices. The Antitrust Bulletin, 24, 233–261. Greer, D. F. (1980). Industrial organization & public policy. New York: Macmillan. Hidy, R. W., & Hidy, M. E. (1955). Pioneering in Big Business, 1882–1911. New York: Harper & Brothers. Just, R. E., & Rausser, G. C. (2007). General equilibrium in vertical market structures: Monopoly, monopsony, predatory behavior and the law. Paper presented at University of Washington Cost-Benefit Conference, May 18, 2006. Kahn, A. E. (1991). Thinking about predation: A personal diary. Review of Industrial Organization, 6, 137–146. Klevorick, A. K. (1993). The current state of the law and economics of predatory pricing. Antitrust and Industrial Organization, 83(2), 162–167. Koller, R. H. (1971). The myth of predatory pricing: An empirical study. Antitrust Law and Economics Review, 4(Summer), 105–123. Lott, J. R. (1999). Are predatory commitments credible?: Who should the courts believe? Chicago and London: University of Chicago Press. Mankiw, N. G. (2004). Principles of microeconomics (3rd ed.). Cincinnati: South-Western College Publications. Mariger, R. (1978). Predatory price cutting: The standard oil of New Jersey case revisited. Explorations in Economic History, 15(4), 341–367. McGee, J. (1958). Predatory price cutting: The standard oil (N.J.) case. Journal of Law and Economics, 1(1), 137–169. McGee, J. (1980). Predatory pricing revisited. Journal of Law and Economics, 23(2), 289–330. McPherson, J. M. (2003). Revisionist historians. AHA Perspectives, (Summer), September Online Edition. Milgrom, P., & Roberts, J. (1992). Economics, organization and management. Saddle River, NJ: Prentice Hall. Morton, F. S. (1997). Entry and predation: British shipping cartels 1879–1929. Journal of Economics and Management Strategy, 6(4), 679–724. Nevins, A. (1940). John D. Rockefeller, The heroic age of American enterprise (Vols. I–II). New York: Scribner’s Sons. Nevins, A. (1953). Study in power: John D. Rockefeller, Industrialist and Philanthropist (Vols. I–II). New York: Scribner’s Sons. Pepall, L., Richards, D. J., & Norman, G. (2002). Industrial organization (2nd ed.). Cincinnati: South-Western. Posner, R. A. (1976). Antitrust law: An economic perspective. Chicago: University of Chicago Press. Posner, R. A. (2001B). Antitrust in the new economy. Antitrust Law Journal, 68(3), 925–943.
202
JAMES A. DALTON AND LOUIS ESPOSITO
Salop, S. C. (Ed.) (1981). Strategy, predation, and antitrust analysis. Washington, DC: Federal Trade Commission. Scherer, F. M. (1970). Industrial market structure and economic performance. Chicago: Rand McNally. Scherer, F. M. (1976a). Predatory pricing and the Sherman Act: A comment. Harvard Law Review, 89, 883–889. Scherer, F. M. (1976b). Some last words on predatory pricing. Harvard Law Review, 89, 901–903. Scherer, F. M. (1996). Industry structure, strategy, and public policy. New York: Harper Collins. Scherer, F. M., & Ross, D. (1990). Industrial market structure and economic performance (3rd ed.). Boston: Houghton Mifflin. Shepherd, W. G., & Shepherd, J. M. (2004). The economics of industrial organization (5th ed.). Long Grove, IL: Waveland Press. Spence, M. (2002). Signaling in retrospect and the informational structure of markets. American Economic Review, 92(3), 434–459. Stiglitz, J. E. (2002). Information and the change in the paradigm of economics. American Economic Review, 92(3), 460–501. Tirole, J. (1988). The theory of industrial organization. Cambridge, MA: The MIT Press. U.S. Bureau of the Census. (1976). The statistical history of the United States from colonial times to the present. U.S. Department of Commerce, Bureau of Corporations. (1906). Report of the Commissioner of Corporations on the transport of petroleum, House of Representatives Document No. 812. Washington: G.P.O. U.S. Department of Commerce, Bureau of Corporations. (1907). Report of the Commissioner of Corporations on the petroleum industry, Part I – The position of Standard Oil Co. in the petroleum industry; Part II – Prices and Profits. Washington: G.P.O. Williamson, H. F., & Daum, A. R. (1959). The American petroleum industry, 1859–1899. Evanston, IL: Northwestern University Press. Williamson, O. E. (1977). Predatory pricing: A strategy and welfare analysis. Yale Law Journal, 87(2), 284–340. Yamey, B. S. (1972). Predatory price cutting: Notes and comments. Journal of Law and Economics, 15(1), 129–142. Yergin, D. (1991). The prize: The epic quest for oil, money and power. New York: Simon and Schuster. Zerbe, R. O., & Cooper, D. S. (1982). An empirical and theoretical analysis of alternative predation rules. Texas Law Review, 61, 655–715. Zerbe, R. O., & Mumford, M. T. (1996). Does predatory pricing exist? Economic theory and the courts after Brooke Group. The Antitrust Bulletin, 41, 949–985.
LEGAL CASES CITED Advo Inc. v. Philadelphia newspapers, Inc., 51 F.3rd. (1995). Bathke v. Casey’s General Stores, Inc., 64 (8th Cir.). (1995). Brooke Group Ltd. v. Brown & Williamson Tobacco Corporation. 509 U.S. (1993). Matsushita Electric Industrial Corp., Ltd., et al. v Zenith Radio Corp. et. al., 475 U.S. (1986).
Predatory Price Cutting and Standard Oil
203
Standard Oil Co. et. al. v. United States, 725 (Mo. Ct. App.). (1909). Standard Oil Company of N.J. et .al. v. U.S., 221 U.S. 1. (1911). State of Missouri v. Standard Oil et al. (218 Mo., 1). (1909). United States v. Standard Oil Co. NJ et. al. 173F. 177 (Mo. Ct. App.). (1909). United States v. Standard Oil Co. NJ et. al. 173F. 177 (Mo. Ct. App.). (1909) – Brief of Facts and Argument for petitioner, Volumes 1 and 2 (1908). United States v. AMR Corp., 335 (10th Cir.). (2003). Vollrath Company v. Sammi Corp., 9 (9th Cir.). (1993).
Further Reading Areeda, P., & Turner, D. F. (1975). Predatory pricing and related practices under Section 2 of the Sherman Act. Harvard Law Review, 88, 697–733. Bolton, P., Brodley, J. F., & Riordan, M. H. (2001). Predatory pricing: Response to critique and further elaboration. Georgetown Law Review, 89, 2495–2529. Brodley, J. F., & Hay, G. A. (1981). Predatory pricing: Competing economic theories and the evolution of legal standards. Cornell Law Review, 66(4), 738–803. Bruckman, B. O., & Ling, M. C. (Eds) (1995). Predatory pricing law: A circuit-by-circuit survey. Chicago, IL: Section of Antitrust Law, American Bar Association. Carlton, D. W., & Perloff, J. (2005). Modern industrial organization (4th ed.). Boston: Addison Wesley. Dewey, D. (1964). Monopoly in economics and law. Chicago: Rand McNally. Eckert, A. (2002). Predatory pricing and the speed of antitrust enforcement. Review of Industrial Organization, 20(June), 375–383. Gifford, D. J. (1994). Predatory pricing analysis in the Supreme Court. Antitrust Bulletin, 39(2), 431–483. Greene, W. L., Adelstein, R. D., Farmer, D. A., Jr., Goldberg, M. R., McKeown, J. T., Ordover, J. A., Quick, L. M. (1996). Predatory pricing. American Bar Association, Section on Antitrust Law, Monograph 22. Chicago, IL: American Bar Association. Hidy, R. W. (1952). The standard oil company. Journal of Economic History, 12, 411–424. Manns, L. D. (1998). Dominance in the oil industry: Standard oil from 1865 to 1911. In: D. I. Rosenbaum (Ed.), Market dominance: How firms gain, hold or lose it and the impact on economic performance (pp. 11–37). Westport, CT and London: GreenwoodPraeger. McCall, C. W. (1988). Rule of reason versus mechanical tests in the adjudication of price predation. Review of Industrial Organization, 3(3), 15–44. Milgrom, P., & Roberts, J. (1982). Predation, reputation, and entry deterrence. Journal of Economic Theory, 27, 280–312. Posner, R. A. (1998). Economic analysis of law (5th ed.). New York: Aspen Law and Business. Posner, R. A. (2001A). Antitrust law: An economic perspective (2nd ed.). Chicago: University of Chicago Press. Rosenbaum, D. I., & Ye, M.-H. (1992). Attempts to monopolize and the determination of specific intent. Quarterly Review of Economics and Finance, 32(1), 50–70. Telser, L. G. (1966). Cutthroat Competition and the Long Purse. Journal of Law and Economics, 9(October), 259–277.
204
JAMES A. DALTON AND LOUIS ESPOSITO
APPENDIX. REPUBLIC OIL COMPANY The Facts Standard operated Republic as a bogus independent after Standard changed the name from SS&T to Republic Oil Company. Nothing changed after the acquisition except the name on the letterhead of stationary and on the tank wagons. Republic operated as if it were an independent, and instructed its sales and office employees to represent that Republic was independent of Standard.95 Republic used Standard’s database that identified the customers and suppliers of wholesalers where independents provided stiff competition to Standard. Republic’s sales force was instructed to seek only the business of Standard’s independent competitors.96 The sales force then used rebates to obtain the business of these independents and did not approach Standard’s customers with any rebate offers.97 One question then is the extent to which Republic reduced prices. The Record does not provide specific data on price–cost relationships.98 Nonetheless, the testimony of A. G. Shiers, who managed Republic’s tank wagon business in Kansas City, and A. H. Gardner, the manager of National Oil, a competitor of Standard, clearly suggest that Republic engaged in predatory pricing. Shiers gave 20–25 rebates a month that set prices below Republic’s regular market price. He captured the two largest customers of National Oil when he started a price war (v. 2, pp. 979–984). The loss of these customers clearly implies that National Oil simply could not meet the prices of Republic Oil because they were National’s most important customers in that area. Gardner and P. C. Crenshaw, a Standard executive, testified that Standard arranged for a meeting in 1902 to end a price war between Republic and National,99 a price war that Republic had initiated.100 He testified that Republic drove down the price from 13 cents a gallon to 8 cents a gallon in two months. According to Gardner, Crenshaw said that if he had known about the depth of these price cuts, he would have stopped Teagle from engaging in the practice (Gardner, v. 2, pp. 947–951).101 Standard could not keep its own customers ignorant of the rebates that Republic was offering. National had informed Standard’s customers that they could get the same rebate from Republic, thereby eliminating the discriminatory advantage that Standard enjoyed when Republic, the bogus company, only went after National’s customers.
Predatory Price Cutting and Standard Oil
205
McGee’s Explanation of the Republic Oil Case McGee cited testimony by P. C. Crenshaw as further proof that Republic had not engaged in predatory pricing (p. 153, fn. 73). Crenshaw’s testimony, however, did not address Republic’s specific pricing practices. Crenshaw himself ruled out his ability to testify about Republic’s pricing behavior when he acknowledged that he had ‘‘no personal knowledge of rebates being paid in Missouri or elsewhere (p. 1173).’’ Rather, Crenshaw only described his effort in 1902 to settle a price war between National and Republic. That this event occurred clearly implies that Republic was involved in deep price cutting, an event that McGee neglected to explore. Gardner testified that Teagle, Republic’s General Manager, had driven prices too low, a reaction consistent with Republic pricing below its variable costs. An Alternative Explanation of the Republic Oil Case Republic used Standard’s database to target only the customers of independent wholesalers who were competing with Standard. Republic discriminated by granting substantial rebates to the customers of the independents. The loss by National Oil of its two prime customers, as described by Shiers, and the deep price cuts by Standard that led to a meeting to settle the price war clearly suggest behavior consistent with predatory pricing by Standard. As a consequence, McGee’s conclusion that there is no evidence that Republic engaged in predatory pricing is highly suspect, if not erroneous. The evidence in the Record shows opportunity, intent, behavior, and outcomes consistent with a model of predatory pricing that uses selective price cuts to eliminate or discipline a rival. A deeper investigation of Republic’s pricing behavior is necessary before it can be concluded that Republic did not engage in predatory pricing.
This page intentionally left blank
ON THE OPTIMAL NEGLIGENCE STANDARD IN TORT LAW WHEN ONE PARTY IS A LONG-RUN AND THE OTHER A SHORT-RUN PLAYER Henrik Lando ABSTRACT It is well established that courts should and in fact do require a higher level of care by people working within their profession than by amateurs. Adequate care is simply more within reach for the professional than for the amateur (less ‘costly’). This article analyzes whether a further distinction between the professional and the amateur should influence the way courts set negligence standards: the professional is more likely to invest in acquiring information concerning negligence standards, and the professional is hence more likely than the amateur to be influenced by the standards. This issue is analyzed for the case where the professional is the injurer and the amateur is the victim. The amateur is assumed not to acquire any information concerning standards, and the behavior of the amateur is taken as exogenously fixed. Under this assumption, the negligence standard applied to the professional may be either higher or lower than first best, depending on whether care levels by the injurer and the victim are
Research in Law and Economics, Volume 22, 207–216 Copyright r 2007 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0193-5895/doi:10.1016/S0193-5895(06)22006-2
207
208
HENRIK LANDO
substitutes or complements and on whether, in the absence of information, the amateur over- or under-estimates the standard applied to him or her.
INTRODUCTION The following situation occurs often: a professional party and an amateur enter into a relationship and the amateur incurs a loss for which both parties are partly to blame. For example, a bank employee provides inadequate advice to a customer who invests imprudently and thereby incurs a loss that would have been averted if better advice had been given or if the customer had acted more prudently. For such situations, it is well established that the negligence standards of the two parties must be set with a view to their respective abilities as professional and amateur; exercising care may be more within reach for the professional than for the amateur, and negligence standards should reflect this. This article introduces a further distinction between the two. To the extent that the professional is a long-run player, he or she will be more influenced by the incentives established by the legal system than the amateur, since the professional will have an incentive to invest in acquiring the information concerning how the Court assigns liability, while the amateur, if he or she is a one-time player, will be unlikely to invest in acquiring this information.1 This article analyzes how a court should set negligence standards when the amateur’s behavior can be viewed as exogenously fixed, i.e. when negligence standards can affect only the behavior of the professional party.2 This assumption will be shown to imply that: – The optimal negligence standard applied to the professional’s conduct may then be either higher or lower compared to what is optimal when both parties can be relied on to become informed about the Court’s verdict.3 Important determinants of the optimal negligence standard will be whether amateurs will be likely to over- or under-estimate the standards that will be applied to them and whether care by the professional and by the amateur are substitutes or complements. – Whether amateurs over- or under-estimate the standard applied to them, applying strict liability to the professional injurer will be optimal. In the following, three real world cases will illustrate the situation under study. A model will then be used to derive the implications of the assumption that the amateur cannot be supposed to know the negligence standard set by the Court, and a simplified numerical example will illustrate the forces
On the Optimal Negligence Standard in Tort Law
209
at work in the model. Some qualifications to the analysis are then discussed on the background of the real world examples. A conclusion ends the article.
EXAMPLES First Example: Professional Advice In a case decided by the Danish Supreme Court (U.2000.577/2 H), a group of ‘consumers’ (the legal term for what was referred to above as ‘amateurs’, i.e. people acting outside their profession and not on an ongoing basis) invested in projects that turned out to be a scam. The projects were suggested to them by a firm of brokers, who misled the group in various ways, e.g. by making them believe that insurance had been drawn covering losses. The broker firm went bankrupt when the real nature of the project and the absence of insurance were eventually discovered. As part of the project, a bank provided financing to the group of investors. The bank did not investigate the nature of the projects involved but simply provided the loans based on an assessment of the investors’ personal finances. Furthermore, the bank did not mention to the group of investors that it had not investigated the projects. Considering the relative degrees of negligence by the group of investors and by the bank, the Supreme Court remarked as follows: In marketing the projects, loans from the bank were a central part of the projects, and this the bank must have understood. The bank had to consider that some investors assumed that the bank, which offered loans to the project, had found the project to be trustworthy, and that such investors would borrow the money trusting this to be the case and without seeking other advice. On this background, y the bank ought to have informed the investors that it had not investigated the investment projects.
However, the Court further reasoned: Investors who could not themselves judge the risk of participating ought to have sought professional advice or at least ought to have addressed the bank to make sure it had investigated the prospect.
The Court decided to find the bank not liable on the basis of a balancing of the relative degrees of negligence. The question to be analyzed is whether this verdict ought to have been reversed on the theory that most ordinary citizens are uninformed about standards and therefore are not affected by the establishment of negligence in cases of this nature, while banks will be likely to note the outcome and adjust their routines accordingly (when doing so is worth the cost).
210
HENRIK LANDO
Second Example: Insurance Law A standard issue in insurance law concerns negligence standards when the insured party (the consumer) provides information to the insurance company concerning the nature of risks involved, while the insurance company asks questions of a more or less specific nature. When some issue relevant to the size of the risk has not been taken into account by the insurance contract, the question arises whether the insurance company should be held responsible for not asking specifically about the given issue or whether the consumer should be held responsible for not bringing up the issue in response to a generally worded question such as whether there are any circumstances that increase the risk. In establishing the relative standards of negligence, it is well established that account should be taken of the fact that the insurance company is a professional party who can invest in acquiring appropriate skills and who may therefore take proper precautions, while at least some consumers may find it difficult to fill in questionnaires in a satisfactory way. Naturally, these differences in costs should affect negligence standards. The question raised in this article is how standards of negligence should be set when they are much more likely to be known by the insurance companies than by the consumers.4
Third Example: General Tort/Contract Law The third example concerns a case on the boundary of contract and tort law, which has reached the media but not yet the Court. A small shopkeeper meant to deliver the earnings of the day ($80.000) into her bank-box. The usual procedure was to deliver the earnings in a bag into a particular slot in the wall of the bank, but on that day, thieves had put up a sign indicating that the slot was out of order, and that money should be deposited in the bank’s ordinary letter-box above the slot. The shopkeeper deposited the money and it was then lifted out of the letter-box by the thieves. It may be mentioned that the bank’s letter-box was not well protected from theft and that the special slot had been out of order before. Supposing that there are precautions banks can take to prevent such losses (such as putting up a sign on the wall warning its customers never to deposit money anywhere else than in the slot), the question is whether the bank should be held liable on the grounds that banks in general may be more likely than shopkeepers to take note of a strict standard of negligence applied to them.
211
On the Optimal Negligence Standard in Tort Law
THE MODEL Consider a model of a professional (the injurer) and an amateur (the victim) who may both lower the probability of a loss l. The probability of the loss equals p(a, b), where a is the precaution of the injurer and b is the precaution of the victim. The cost to the injurer is c(a), while the cost to the victim of exercising precaution is d(b). If we assume (as we will throughout) that firstorder conditions are sufficient and necessary for an optimum, the negligence standards (a*, b*) that minimize total losses are given by ca ða Þ ¼
pa ða ; b Þl
and
d b ða Þ ¼
pa ða ; b Þl
These are the optimal standards when both parties are well informed about the situation (see e.g. Shavell, 1987). As is well known, if both parties are well informed about these standards, they will have an incentive to abide by them, and optimality will be ensured. The question is whether the standard applied to the injurer should be altered when the victim cannot be relied upon to know the standard applied to him or her. Consider first the case where the victim over-estimates the likelihood of having to carry own losses, i.e. thinks that the Court will be harsh on him or her. In this case, he or she is likely to choose some level of precaution b1 greater than b*. Then the standard that should be applied to the injurer, a1*, assuming that the first-order condition yields the global minimum, solves: ca ða1 Þ ¼
pa ða1 ; b1 Þl
from which it can be seen that if a and b are substitutes, i.e. if pab(a,b)o0, so that at a higher level of b, an increase in a induces less of a decrease in the probability of a loss, then a1*oa*. If the standard for the victim will continue to be set at b*, the injurer will choose as his or her level of care the now optimal level a1*. The idea behind requiring a lower standard of care by the injurer is that when the victim takes more than the optimal degree of care, there is less reason to demand care by the injurer. Conversely, if a and b are complements, the negligence standard applied to the injurer should be stricter when the victim believes to be treated more harshly than is in fact the case. In this case, if the standard applied to the victim remains b*, the injurer will choose the higher level as the level of care.5 On the other hand, if the victim believes that he or she will be treated more leniently than is in fact the case, results are turned around: if a and b are substitutes (complements), the negligence standard applied to the injurer should then be stricter (more lenient) than if the victim can be assumed to invest in information concerning the standards applied by the Court.
212
HENRIK LANDO
One complication is worth mentioning: in the case of substitutes, it may not be optimal for the injurer to live up to the higher standard if the standard applied to the victim will be maintained at b*, since the loss will be shared by the parties (under the rule of comparative negligence) and if a high fraction is borne by the victim, it may conceivably be optimal for the injurer to act with less than the higher level of due care, despite the discontinuity of his or her loss at that point. However, should this problem exist, a simple solution would be to lower the standard of care applied to the victim to the level b1 in order to induce the injurer to exercise due care. Hence, this complication is not worth further attention. The intuition behind the result that it is optimal in the case of substitutes to demand more care by the knowledgeable party when the victim cannot be relied on to take due care is clear: care by the injurer will then prevent more instances of harm from occurring. Note finally that the rule of strict liability will tend to be optimal when victims are unaffected by the rulings of the Court. When victims’ actions are ~ the injurer will under strict liability minimize exogenous, equal to b; ~ þ cðaÞ; which induces the socially optimal level of care, given the pða; bÞl level of care exercised by the victim, whether that level is too high or too low (as long as the injurer knows the actual level of care exercised by the victim). Hence, strict liability is optimal not only for unilateral care situations but also when the victim does act but does not acquire information concerning the standard or the kind of rule applied by the Court.
AN ILLUSTRATION The basic principle may be illustrated in the case where efforts by the injurer and the victim are substitutes and where some fraction of victims is optimistic while others are pessimistic about outcomes for them. Consider the following situation: the professional (the injurer) can take two acts 0 or 1 and so can the amateur (the victim). Both can eliminate the loss of one unit by undertaking the more costly act 1 (Table 1). It is less costly for the victim to take precautions so the optimal outcome is for the victim to take precautions and for the injurer not to. This may be accomplished by setting the standard of due care for the injurer below the act 1 (i.e. to not consider act 0 on the part of the injurer as negligence). Whether the rule is comparative or contributory negligence, one achieves the optimal result: the victim will take care while the injurer will not.
213
On the Optimal Negligence Standard in Tort Law
Table 1.
The Pay-Off Matrix.
Injurer
Act 0 at cost 0 Act 1 at cost 3/4
Victim Act 0 at cost 0
Act 1 at cost 1/2
Loss ¼ 1 Loss ¼ 0
Loss ¼ 0 Loss ¼ 0
However, if the victim does not know the negligence standard imposed by the Court, the equilibrium just mentioned may not materialize. If we assume that some fraction a of victims believes that they will have to cover their own losses if they act imprudently (i.e. take act 0), while the fraction 1 a believes that the injurer will be liable in that case, and if we assume that this fraction is unaffected by how the Court assigns liability, it may be optimal to require the injurer to take act 1. If the fraction a of victims acts with caution and the fraction 1 a does not, the total expected costs when the injurer chooses act 1 is 3/4, while the total expected cost is 1 a when the injurer chooses act 0. Thus, the optimal rule is to assign liability to the injurer when ao1/4.6 Thus, if a sufficient fraction of victims will not take care, and the legal rule will not affect their conduct, it becomes optimal to require act 1 of the injurer. This is an instance of the principle mentioned above that if victims believe they will be treated leniently, and a and b are substitutes, the negligence standard applied to the injurer should be stricter than if victims have the incentive to become informed. However, it should be noted that it is optimal to require a higher standard of care by the injurer only if a sufficiently high fraction of victims can be expected to act without due care in the absence of information concerning what the Court will consider due care.
DISCUSSION Some qualifications to the general argument of this article are worth mentioning. First, it may be claimed that litigation costs (or conflict resolution costs more generally) should be included in the analysis. However, it is not clear how this would affect results. If one assumes that victims will become informed after the occurrence of an accident about the standard applied by the Court (as the victims contact a lawyer), a stricter negligence standard (or strict liability) will create a larger pool of valid claims and hence conceivably also a larger set of claims over which there will be divergent opinions and
214
HENRIK LANDO
conflict. On the other hand, if the injurer will avoid losses through greater care, fewer losses will be incurred, and fewer conflicts will arise. This is a well-known trade-off that leaves the issue of whether changing the standard imposed on the injurer increases conflict resolution costs as an empirical question. Second, it is clear that one should not under-estimate the incentive for private parties to acquire information concerning negligence standards, and for information to reach people who have an interest in acquiring that information. Information spreads easily through mass media and the Internet, and amateurs or consumers may become informed by more knowledgeable parties (such as lawyers, bank personnel, accountants or other advisors). And each verdict may well reach the attention of a certain number of people and thereby contribute to the general impression people have of the standards that the legal system imposes on them as amateurs or consumers. These factors may in certain cases modify the present analysis. Third, it should be noted that Courts may affect the incentive for the parties to become informed.7 Thus, in the case of the fraudulent projects, it makes sense for the Court to insist that the amateurs acted negligently by not asking for professional advice; if the Court can make clear that this is required of non-professionals,8 this requirement may spread through various channels, e.g. through bank personnel providing advice on personal finances, and if amateurs become aware of this requirement, the system may thereby be brought to work as hypothesized by the bench-mark economic models of accident law. In other words, if amateurs can be induced to seek professional advice, standards of negligence can hereby become known to them, and the Court can then set standards optimally. In the given case, the Court attempted to send two signals: to the bank that it may be found negligent in similar cases (if the victims had acted less negligently) and to future amateur investors that not seeking professional advice is an act of negligence. It is essentially an empirical question whether the Court should have sent a stronger signal to banks or whether it was correct to attempt to influence the incentive to seek advice. But the present analysis suggests that shared liability might have been the optimal verdict: imposing liability on the bank would have been rather certain to be noted by banks, while the signal to amateurs that they must seek professional advice seems less certain to reach (a majority of) future amateur investors. For this reason, since ignorant investors who do not seek professional help are likely to continue to exist, and since the cost for the bank of warning them in some way seems rather small, it might well have been advisable for the court to share the losses more equally between the two parties.
On the Optimal Negligence Standard in Tort Law
215
CONCLUSION Liability in a tort or contract case is often established on the basis of an expost evaluation of the blameworthiness of the acts of the two parties. As is well known, the law and economics approach differs by its ex-ante approach: it establishes liability on the basis of how the verdict or the standards expressed by the verdict will affect future behavior. Often, but not always, the two approaches will lead to the same result. One instance where the two approaches may differ arises when it is only optimal for the professional party to become informed about the verdict and the standards expressed by it. In this instance, only the behavior of the professional will be affected by the verdict, and the Court should set standards with a view to affecting the behavior of the professional in an optimal manner, taking as given the behavior of the amateur. The following results were derived concerning the optimal way for the Court of doing so: If the (rationally) ignorant amateur takes less care than optimal and the precautionary measures taken by the two parties are substitutes, it will be optimal to require more than the socially optimal level of care by the professional party (which may be more than appears reasonable). On the other hand, if care levels are complements as it may be when the parties need to cooperate, and the level of care by the amateur is below the socially optimal level, it is optimal to also require a lower than first-best level of precaution by the professional. Moreover, whether the ignorant party over- or under-estimates the standards imposed by the Court, as long as the victim’s behavior can be considered as exogenous, imposing strict liability on the knowledgeable party will induce the best obtainable level of precaution; this solution takes advantage of the fact that the injurer is likely to know the behavior of victims better than the Court.
NOTES 1. Verkerke (2003) discusses legal ignorance, its prevalence and how it affects existing legal doctrines and rules. 2. In Lando (2006), the case is considered where the amateur may be supposed to derive information concerning negligence standards from the acts undertaken by the professional. 3. This finding establishes one clear difference between the point made in this paper and that made by Ayres and Gertner (1992) and Verkerke (2003) in their papers on penalty default rules. These papers argue that default rules should be set to the disadvantage of the informed party to induce the informed party to reveal
216
HENRIK LANDO
information to the other party. While their point hence relates to reveal information, the present analysis is mainly about tort cases where private information cannot be revealed between the parties prior to the ‘accident’. 4. Although insurance companies attempt to provide this information to consumers. 5. Again on the sufficient but not necessary assumption that the first-order condition yields the global minimum of total cost given b; the condition is not necessary as the discontinuity of the injurer’s loss-curve under the negligence rule will tend to make it optimal for the injurer to act with due care even if the level of due care is raised to a higher level. 6. If the injurer can observe the precaution actually taken by the victim, the injurer should of course choose the high level of precaution when but only when the victim does not exercise care. In this case, the optimal standard applied to the injurer will depend on the behaviour of the victim (if it can be observed by the Court). 7. Although the incentive to become informed about a standard is not directly related to its strictness; naturally, the incentive is related to the value of correcting a mistaken belief in either direction. 8. It seems reasonable to assume that the particular requirement that amateurs must seek professional advice will more often come to their knowledge than complex standards of negligence set by the Court.
REFERENCES Ayres, I., & Gertner, R. (1992). Strategic contractual inefficiency and the optimal choice of legal rules. Yale Law Journal, 101, 729–773. Lando, H. (2006). Legal ignorance and optimal standards of negligence. Working paper http:// papers.ssrn.com/sol3/papers.cfm?abstract_id=906084 Shavell, S. (1987). Economic analysis of accident law. Cambridge, MA: Harvard University Press. Verkerke, J. H. (2003). Legal ignorance and information-forcing rules. Working Paper, University of Virginia Law & Econ. Research Paper No. 03–4. http://ssrn.com/abstract=405560
SHOULD VICTIMS OF EXPOSURE TO A TOXIC SUBSTANCE HAVE AN INDEPENDENT CLAIM FOR MEDICAL MONITORING? Thomas J. Miceli and Kathleen Segerson ABSTRACT Traditional tort law does not allow a victim of exposure to a toxic substance to seek damages without evidence of actual loss. Given the difficulty of collecting damages after a long latency period, however, we examine the desirability of granting exposure victims an independent cause of action for medical monitoring at the time of exposure. We show that such a cause of action is not necessary to induce victims to invest in efficient monitoring. It can, however, increase incentives for injurer care, but only at the cost of greater litigation costs. The general reluctance of courts to adopt a cause of action reflects their recognition of this trade-off.
1. INTRODUCTION Under the traditional law of torts, victims cannot sue for damages without evidence of actual loss. For example, a victim of exposure to a hazardous substance who has not developed symptoms of illness does not have a cause
Research in Law and Economics, Volume 22, 217–231 Copyright r 2007 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0193-5895/doi:10.1016/S0193-5895(06)22007-4
217
218
THOMAS J. MICELI AND KATHLEEN SEGERSON
of action because, as a leading case book notes, ‘‘The threat of future harm, not yet realized, is not enough.’’1 However, several commentators have noted that this requirement erects undue barriers to recovery for exposure victims because, given the long latency period of many illnesses, it is often difficult at the time of illness to establish a causal connection between the exposure and the illness. This is especially true of illnesses for which there is a background risk unrelated to the exposure. Even when causation can be established, however, statutes of limitations and statutes of repose, as well as the potential for injurer insolvency, may render the injurer judgment proof.2 These problems undermine both the compensatory and the deterrence functions of tort law.3 One proposed solution is to give plaintiffs an independent cause of action, at the time of exposure, for recovery of reasonable medical monitoring costs.4 It is an established practice for courts to award such expenses to exposure victims, along with ordinary damages, at the time they actually develop symptoms;5 the question of interest here is whether an independent claim for such costs should exist. The medical benefit of awarding medical monitoring at exposure, it is argued, is to promote early detection and improved treatment of the illness.6 It therefore serves to ‘‘mitigate’’ the damages from exposure (Shavell, 1987, pp. 158–159). An offsetting cost is the prospect of a ‘‘flood’’ of litigation, ‘‘potentially absorbing resources better left available to those more seriously harmed y.’’7 After weighing these factors, the U.S. Supreme Court, in Metro-North Commuter Railroad Co. v. Buckley,8 denied a claim for medical monitoring by a railroad worker exposed to asbestos. The Court noted that, although several state supreme courts have allowed such claims,9 they have done so subject to certain limitations in recognition of the offsetting factors.10 The Court therefore concluded that creation of a separate cause of action for medical monitoring was not justified on policy grounds and went ‘‘beyond the bounds of currently ‘evolving common law’.’’11 To date there has been no explicit economic analysis of the desirability of allowing a tort claim for medical monitoring.12 In this paper, we develop a simple model to examine whether, or under what conditions, allowing such a claim would enhance social efficiency in the sense of reducing total social costs. We first consider a world of zero litigation costs. In that context, we show that allowing a separate claim has no effect on efficiency. Victims choose the efficient amount of mitigation and injurers choose the efficient amount of care regardless of whether a separate claim is allowed, provided awards are appropriately adjusted. However, in the more realistic world where litigation is costly, allowing a separate claim for medical monitoring can affect social costs. On the one hand, creation of an independent action
Victims of Exposure to a Toxic Substance
219
for medical monitoring increases the number of suits faced by injurers. This increases litigation costs, which, ceteris paribus, increases social costs. On the other hand, injurer costs go up as well, thereby increasing injurer care. This decreases social costs because, in the presence of litigation costs, injurers are under-deterred by standard liability rules. Whether allowing a separate action for medical monitoring increases or decreases social costs therefore depends on the relative magnitude of these two effects. This conclusion reflects the general trade-off between litigation (enforcement) costs and deterrence in a costly legal system (Becker, 1968; Shavell, 1982; Hylton, 1990). While we are unaware of any previous studies that examine explicitly the efficiency of independent medical monitoring claims, some previous studies have examined closely related issues, and we build on these analyses. For example, Landes and Posner (1984) examined the ability of tort law to deal with the problem of catastrophic, or large-scale, accidents, including mass exposures to a toxic substance. After noting the above problems with the traditional tort-for-illness rule in this context, they proposed a rule that allows victims to file at exposure for expected damages. They did not, however, explicitly address the desirability of allowing recovery for medical monitoring (in addition to expected damages) at the time of recovery. Rose-Ackerman (1989) considered a seemingly unrelated setting – a case of trespass – that nevertheless raises closely related issues. Suppose a mining company builds a dam, creating the risk of a flood (if the dam bursts) that would cause damage to a downstream landowner. The owner, foreseeing the risk, builds a dike to protect her property. Should the owner be allowed to seek compensation from the mining company for the cost of the dike? Efficiency dictates that the answer is yes, for otherwise, the mining company will underestimate the cost of building the dam. Allowing recovery, however, raises the question of when the landowner can file suit. Rose-Ackerman shows that two timing rules are equally efficient: one that allows the landowner to sue at the time the dam is built for the cost of the dike plus expected flood damages, and one that only allows a suit in the event of a flood for actual damages plus a multiple of the cost of the dike. Note that the first rule, if applied to the toxic tort context, would embody a claim for medical monitoring (in addition to expected damages) because it allows the owner to seek compensation for his precautionary behavior at the time the expenditure is undertaken, i.e., prior to the occurrence of an accident. The second also allows recovery for medical monitoring expenses, but only by victims who actually suffer damages. We demonstrate the efficiency of Rose-Ackerman’s two timing rules in the context of the model of toxic torts developed here, and also consider a third that allows two separate
220
THOMAS J. MICELI AND KATHLEEN SEGERSON
actions, one for recovery of medical monitoring expenses at the time of exposure, and one for compensatory damages at the time of illness. The remainder of the paper is organized as follows: Section 2 sets up the general exposure model assuming zero litigation costs, and shows the conditions under which each of the three timing rules leads to efficient care by injurers and efficient medical monitoring (mitigation) by victims. The results imply that all the three rules induce efficient behavior by both parties, provided damages are appropriately defined. Section 3, however, shows that when litigation costs are taken into account, this conclusion is no longer true. In particular, whereas allowing a separate tort for medical monitoring at the time of exposure improves incentives for injurer care, it results in higher costs of litigation. In light of this trade-off, the desirability of allowing such a claim depends on a balancing of these two effects. Finally, Section 4 concludes.
2. THE MODEL WITH ZERO LITIGATION COSTS We first consider a world in which litigation costs are zero. This allows us to focus on the incentive effects of the various rules regarding recovery of medical monitoring expenses. The notation of the model is as follows: p(x): probability of exposure to a toxic substance; x: injurer care, p0 o0 and p00 >0; q: probability of illness given exposure; D(M): damages in the event of illness; M: expenditure on medical monitoring by the exposure victim, D0 o0 and D00 >0. The model is one of ‘‘sequential care’’ along the lines of that in Wittman (1981) and Shavell (1983).13 That is, the injurer moves first and chooses a level of care that determines the risk of exposure. Then, if an exposure occurs, the victim chooses a level of medical monitoring aimed at mitigating the expected loss.14
2.1. Social Optimum and Filing Rules Since the parties act sequentially, we derive the social optimum by considering their optimal care choices in reverse order. Thus, in the event of an
Victims of Exposure to a Toxic Substance
221
exposure, the victim’s problem is to choose the level of medical monitoring, M, to minimize her expected costs: M þ qDðMÞ
(1)
which yields the first-order condition: 1 þ qD0 ¼ 0
(2)
Denote the solution to Eq. (2) by M. Given optimal monitoring by the victim, the injurer’s problem is to choose care, x, to minimize x þ pðxÞ½M þ qDðM Þ
(3)
which yields the first-order condition: 1 þ p0 ½M þ qDðM Þ ¼ 0
(4)
Let x be the solution to Eq. (4). Shavell (1983) and Wittman (1981) have characterized the ability of various liability rules to achieve the efficient outcome in sequential care accidents under the condition that the victim can only file at illness.15 Here, the focus is on the timing of lawsuits. We therefore restrict attention to a rule of strict liability, and consider three alternative rules that differ in the timing of recovery and whether an independent action for recovery of medical monitoring is allowed. The three rules we consider are as follows: Rule 1: Tort for Illness: This is the traditional common law rule under which the victim can only file suit at illness. At that time, she can seek both compensatory damages and reasonable medical monitoring expenses. Rule 2: Tort for Exposure: Under this rule, the victim can only file at exposure but can seek both medical monitoring costs and expected damages. Rule 3: Separate Action for Medical Monitoring: Under this rule, the victim can file a separate suit at exposure for medical monitoring costs and can then file again at illness for damages. Note that Rules 1 and 2 are the two proposed by Rose-Ackerman (1989), while Rule 3 ‘‘decouples’’ compensatory damages and medical monitoring. In the next section we examine the conditions (if any) under which these rules can achieve the efficient outcome as derived above.
222
THOMAS J. MICELI AND KATHLEEN SEGERSON
2.2. Injurer and Victim Care As above, we proceed in reverse sequence. Thus, we first consider the victim’s incentives for medical monitoring following an exposure. We begin by defining the general compensation rule C(M)Z0, which represents the present value of any payments that the victim expects to receive under the relevant timing rule as of the time she chooses M.16 This general function can be defined to embody compensation under any of the three rules that we consider. Note that, when compensation is awarded only if the victim becomes ill, C(M) embodies the probability q of receiving compensation. Given this rule, the victim’s problem can be written generally as Minimize : M þ qDðMÞ
CðMÞ
(5)
which yields the first-order condition: 1 þ qD0
@C ¼0 @M
(6)
Comparison of Eqs. (2) and (6) immediately implies that qC/qM0 in order for the victim to choose M. That is, under any of the timing rules, expected compensation to victims must be lump sum, or independent of their actual choices of M. (It follows that compensation must also be independent of their actual damages, D(M).)17 Thus, as long as compensation is lump sum, the victim will be induced to choose the efficient amount of medical monitoring regardless of whether medical monitoring expenses can be recovered at the time of exposure or only if and when illness occurs. This suggests that, in the context of zero litigation costs, allowing recovery by all victims at the time of exposure rather than only by illness victims at the time of illness will not lead to additional investment in medical monitoring. In other words, allowing a separate recovery at the time of exposure is not necessary to promote expenditures on early detection and mitigation.18 Given lump sum compensation of C(M), we can characterize the injurer’s choice of care under any of the three rules as the solution to Minimize : x þ pðxÞCðM Þ
(7)
Comparison of Eqs. (3) and (7) implies that the injurer will choose efficient care if and only if CðM Þ ¼ M þ qDðM Þ
(8)
Victims of Exposure to a Toxic Substance
223
Eq. (8) implies the following efficient versions of the above filing rules: Rule 1: The injurer is liable at illness for M/q+D(M).19 Rule 2: The injurer is liable at exposure for M+qD(M). Rule 3: The injurer is liable at exposure for M and at illness for D(M). Under Rule 1, the traditional rule, victims are precluded from filing at exposure. Thus, in addition to paying reasonable compensatory damages, injurers must also pay a multiple of the victim’s reasonable monitoring expenses, where the multiplier is the inverse of the probability of developing an illness given exposure (i.e., 1/q). This multiplier reflects the fact that victims (efficiently) invest in precaution at the time of exposure, but, under Rule 1, they can only sue for the resulting cost with probability q. Thus, in the absence of the multiplier, the injurer will avoid paying M a fraction 1 q of the time, which will cause him to underestimate the social cost of exposure. For example, if courts award M+D(M) under Rule 1 (in effect reimbursing victims for their actual monitoring costs), injurers will choose too little care. A further drawback of the multiplier is that it only compensates victims on average. Specifically, exposure victims who become ill will receive compensation that exceeds their efficient investment in medical monitoring (M/q>M), whereas exposure victims who never contract the illness will receive no compensation for their monitoring expenses. These concerns provide a rationale (apart from consideration of litigation costs) for allowing recovery of medical monitoring costs at the time of exposure (as is provided by Rules 2 and 3). Rules 2 and 3 both allow a cause of action even if no actual symptoms of illness have arisen. Under Rule 2, victims can only sue at exposure (i.e., a suit at exposure precludes any further suits). Thus, in order to achieve the efficient outcome victims must be able to collect both ‘‘reasonable’’ medical monitoring expenditures plus the damages that they expect to incur if these precautions are taken. The damages are thus discounted by the probability of illness, q.20 In contrast, under Rule 3, victims can sue at exposure for reasonable medical monitoring, and then again at illness (should it occur) for reasonable damages. In this case, efficiency is ensured if they receive reasonable medical monitoring expenses at the time of exposure and then the associated damages if and when an illness occurs. The preceding shows that victims will be induced to invest efficiently in monitoring and injurers will choose efficient care regardless of when recovery for medical monitoring is allowed, given the proper definition of damages. In other words, allowing a separate action for medical monitoring will not affect the incentives of either victims or injurers. This conclusion,
224
THOMAS J. MICELI AND KATHLEEN SEGERSON
however, has not accounted for litigation costs, which, it has been argued, may make suits for exposure too costly from a social perspective.21 We address this issue in the next section.
3. THE MODEL WITH LITIGATION COSTS To incorporate the impact of litigation costs, let ci be the injurer’s litigation costs, and cv be the victim’s litigation costs. In general, litigation costs can affect the victim’s decision of whether or not to file suit, her expenditure on medical monitoring, and the injurer’s choice of care. Additionally, they affect the definition of social costs because the number of suits becomes endogenous depending on the legal rule. We first consider the victim’s filing decision and investment in medical monitoring.
3.1. Victim’s Filing Decision and Care Choice The victim’s filing decision depends on the specific timing rule because the rules differ both in the point at which the victim can file suit and in the amount that she can recover. For example, under Rule 3, the victim can file at exposure for M and again at illness for D(M). We assume that suits at each point in time cost the same amount, cv. Thus, under this rule, at the time of exposure the victim will file a separate action for recovery of medical monitoring expenses if M cv
(9)
She will then file again at illness (if it occurs) if DðM Þ cv
(10)
We assume that both conditions hold. This is sufficient to assure that all possible suits will occur not only under Rule 3 but under the other two rules as well. This assumption allows us to most fully account for the impact of litigation costs on the comparison of the various legal rules.22 We now re-consider the victim’s investment in medical monitoring under each of the filing rules. Under Rule 1, the victim files at illness for
225
Victims of Exposure to a Toxic Substance
M/q+D(M). Thus, her problem at the time of exposure is to choose M to M þ DðM Þ cv (11) Minimize : M þ qDðMÞ q q Since the term in brackets is independent of M, the victim chooses M, the socially efficient level of monitoring. Intuitively, the victim’s choice of M has no effect on the amount of her compensation (because it is lump sum) or on her expected litigation costs (because q, the probability of illness, is constant).23 Thus, the victim has no incentive to distort her choice of M and chooses the efficient level. Under Rule 2, the victim files at exposure for an amount M+qD(M). Her optimization problem is therefore to Minimize : M þ qDðMÞ
½M þ qDðM Þ
(12)
cv
which, for the reasons noted above, yields the efficient level of medical monitoring. Finally, under Rule 3, the victim files at exposure for M, and at illness (should it occur) for D(M). Her optimization problem in this case is to Minimize : M þ qDðMÞ ðM
cv Þ
q½DðM Þ
cv
(13)
which again yields M. The analysis in this section has shown that the addition of litigation costs does not alter the victim’s choice of medical monitoring, which continues to be efficient under all the three rules. The reason is that the victim views the probability of a suit, and hence her expected litigation costs, as independent of her choice of M. Thus, even when litigation costs are positive, allowing a separate suit for recovery of medical monitoring at the time of exposure is sufficient but not necessary to induce the victim to invest in the efficient amount of monitoring. That investment will be efficient even if recovery is only allowed if and when an illness occurs, provided compensation is lump sum and expenditure on monitoring affects only the damages that would result if an illness occurs (and not the probability of contracting the illness).24
3.2. The Number of Suits and Social Costs The preceding analysis of victim filing decisions can now be used to determine the number of suits under the various filing rules. Let nj be the expected number of suits (or, more precisely, the probability of a suit) conditional on
226
THOMAS J. MICELI AND KATHLEEN SEGERSON
an exposure, under Rule j. Given our assumption that all possible suits are filed, we have n1 ¼ q;
n2 ¼ 1;
n3 ¼ 1 þ q
(14)
Under Rule 1, only illness victims file suit. The expected number of suits is thus given by the probability that an exposure victim will contract the illness (and hence file suit), which is simply q. Under Rule 2, all exposure victims file a single suit with certainty. Thus, the expected number of suits under this rule is simply 1. Finally, if a separate action is allowed for recovery of medical monitoring expenses at the time of exposure (Rule 3), all exposure victims will file a medical monitoring suit with certainty, and a fraction q of these will file an additional suit at the time of illness. The expected number of suits under this rule is thus 1+q. It follows from Eq. (14) that the rules can be ranked in terms of the expected number of suits as follows: n3 4n2 4n1
(15)
This provides the basis for the claim that allowing a separate claim for medical monitoring expenses will produce a ‘‘flood’’ of litigation compared to the traditional rule (especially if q is small). Since suits are costly, the definition of social costs in this section will depend on nj. As shown above, all the three rules result in efficient medical monitoring by victims. Thus, we can write social costs under Rule j (SCj) as follows: SCj ¼ xj þ pðxj Þ½M þ qDðM Þ þ nj ðci þ cv Þ
(16)
Note that this differs from Eq. (1) only by the addition of the litigation cost term in square brackets. It follows from Eq. (16) that if x, like M, were invariant to the legal rule, we could use Eq. (15) to rank the rules in terms of social costs since they would differ only in the amount of the expected litigation costs. However, unlike the victim’s choice of medical monitoring, the injurer’s choice of care is not invariant to the legal rule, as we show in the next section.
3.3. Injurer Care Given the care and filing decisions of victims, we can write the injurer’s costs under Rule j (ICj) as ICj ¼ xj þ pðxj Þ½M þ qDðM Þ þ nj ci
(17)
Victims of Exposure to a Toxic Substance
227
Note that this differs from social costs by the absence of the victim’s expected litigation costs, njcv.25 The injurer’s optimal care choice under Rule j solves the first-order condition: 1 þ p0 ½M þ qDðM Þ þ nj ci ¼ 0
(18)
Denote the solution to this by x(nj). Comparison of Eqs. (18) and (4) shows that the injurer takes more care than when litigation is costless,26 but he takes less care than is socially optimal in a world of costly litigation because he ignores the victim’s cost of litigation.27 Further, differentiation of Eq. (18) shows that qx(nj)/qnj>0. The injurer takes more care as the number of suits increases because by doing so, he reduces his expected litigation costs, given by p(x)njci. The fact that injurer care is increasing in n creates an ambiguity in the comparison of the three rules. To see this formally, differentiate the minimized value of social costs in Eq. (16) to obtain @SC j @xj @xj ¼ þ p0 ½M þ qDðM Þ þ nj ðci þ cv Þ þ pðci þ cv Þ @nj @nj @nj @xj ¼ p0 nj cv þ pðci þ cv Þ‘0 @nj
ð19Þ
where the second line is obtained by using Eq. (18). The first term in Eq. (19), which is negative, reflects the reduction in social costs resulting from greater deterrence as the number of suits increases. This is a social benefit because, as noted, the injurer ignores the victims’ litigation costs and hence takes too little care from a social perspective. The second term, which is positive, captures the offsetting increase in social costs as total litigation costs rise. This reflects the concern about the volume of litigation under a tort for exposure. The implication is that, when litigation costs are taken into account, the three filing rules cannot be ranked unambiguously in terms of social costs. Choosing the optimal rule therefore becomes an empirical question based on a balance of deterrence versus litigation costs. In particular, when a separate claim for medical monitoring is allowed, there is no effect on the victim’s monitoring choice but the number of suits is increased. This increase in the number of suits has an ambiguous effect on social costs. On the one hand, it leads to increased care by injurers, which ceteris paribus reduces social costs, but on the other hand, it leads to higher litigation costs, which increase social costs. The net effect on social costs depends on which of these effects is larger. Thus, with costly litigation,
228
THOMAS J. MICELI AND KATHLEEN SEGERSON
allowing a separate claim for medical monitoring will not necessarily increase social welfare. This conclusion is consistent with the general trade-off, well known in the law-and-economics literature, between litigation costs and deterrence. Specifically, suits are socially desirable only if the reduced accident costs owing to greater injurer care offset the costs of litigation (Shavell, 1982; Hylton, 1990).28 How do these results relate to actual legal practice? Generally, courts have been reluctant to recognize a tort for exposure (as embodied in Rules 2 and 3), even in the absence of medical monitoring. This may reflect their judgment that the potential deterrence benefits are outweighed by the additional litigation costs. One might be tempted to argue that the mitigation benefits of medical monitoring, when available, would tip the balance in favor of allowing exposure suits. Our results show, however, that such suits are not in fact necessary to induce exposure victims to invest in efficient medical monitoring – indeed all the three rules, if properly designed, yield the correct incentives for victims. This conclusion is consistent with the Supreme Court’s decision to deny an independent action for medical monitoring in Metro-North v. Buckley, and the fact that, to date, only six states have adopted such an action (Joyce, 2001).
4. CONCLUSION This paper has examined the question of whether victims of accidental exposure to a toxic substance should be allowed to seek compensation for medical monitoring costs prior to the onset of symptoms. Traditionally, the law has not recognized a separate cause of action for medical monitoring, a view that was recently affirmed by the U.S. Supreme Court. However, some state courts have begun to permit such claims under limited conditions. The question we address is how a tort for medical monitoring will affect the social costs associated with accidental exposures. The specific conclusions of the analysis are as follows: 1. One argument in support of a claim for medical monitoring is that it will encourage exposure victims to seek early detection and treatment of the threatened illness, thereby efficiently mitigating the damages. We showed that this is not necessarily the case. In particular, if victims receive lump sum compensation for reasonable medical monitoring expenses, then they will invest in efficient monitoring regardless of whether the compensation is awarded before or after the onset of the illness.
Victims of Exposure to a Toxic Substance
229
2. In the absence of litigation costs, allowing a separate claim for medical monitoring has no effect on injurer care or social costs. Given the appropriate definition of damages, it is possible to achieve the first-best outcome for both victim and injurer care, regardless of whether a separate claim is allowed or not. 3. An important factor in discussions of medical monitoring is the potential for a flood of litigation. We showed that when litigation costs are taken into account, the desirability of allowing a claim for medical monitoring hinges on a trade-off between the increased number of suits, which raises social costs, and increased injurer care, which lowers social costs by reducing the amount of under-deterrence. Efforts to quantify these costs would therefore seem to be a necessary component of the debate about whether or not to allow claims for medical monitoring.
NOTES 1. Keeton, Dobbs, Keeton, and Owen (1984, p. 165). To assert a tort claim, a victim must prove harm, causation, and, in the case of negligence, breach of duty. 2. On statutes of limitations and statutes of repose, see Miceli (2000). On the judgment-proof problem, see Shavell (1986) and Beard (1990). 3. See, generally, Landes and Posner (1984), Robinson (1985), and Slagel (1988). 4. See, e.g., Slagel (1988) and Hamrick (1998). 5. For example, in Metro-North Commuter Railroad Co. v. Buckley, 521 U.S. 424, 438 (1997), the Court stated that ‘‘y an exposed plaintiff can recover reasonable medical costs if and when he develops symptoms.’’ 6. See Slagel (1988). In Metro-North v. Buckley (1997, p. 443), the Court stated that ‘‘y providing preventive care to individuals who would otherwise go without can help to mitigate potentially serious future health effects of diseases by detecting them in early states y.’’ 7. Metro-North v. Buckley (1997, p. 442). Also see Hamrick (1998). 8. 521 U.S. 424 (1997). 9. See the survey of state treatments of medical monitoring claims in Joyce (2001). 10. In particular, the courts have noted the inappropriateness of lump-sum damages, suggesting instead the creation of court-supervised funds for medical monitoring or the use of insurance mechanisms (see Metro-North v. Buckley, 1997, pp. 440–441). 11. Metro-North v. Buckley (1997, p. 440) (quoting Consolidated Rail Corp. v. Gottshall, 512 U.S. 532, 558 (1994)). 12. But see Bathgate (2001). 13. Also see Shavell (1987, pp. 158–159), who discusses the closely related issue of mitigation of damages by victims.
230
THOMAS J. MICELI AND KATHLEEN SEGERSON
14. Note that the victim’s damage in the event of an illness, D(M), is written as a function of the latter’s investment in medical monitoring, while the probability of illness conditional on exposure, q, is not. This reflects the idea, noted above, that the primary benefit of monitoring is to detect and mitigate the illness rather than to prevent it. Ordinarily, it is not necessary to be explicit about the components of expected damages in accident models; all that matters is the product, qD(M). However, given that we are interested in the timing of lawsuits, the probability of illness, q, will be important for calculating the proper measure of damages at exposure versus at illness, and, when litigation is costly, for determining the expected number of suits under the various filing rules. 15. Also see Miceli (1997), Chapter 3. 16. For simplicity, we ignore discounting in the formal analysis. 17. Landes and Posner (1984) and Rose-Ackerman (1989) also make this point. 18. Note that this ignores any potential wealth constraints a victim might face, which could limit the ability of victims to invest efficiently. 19. To see that this rule is efficient, note that the injurer’s problem is to minimize x+p(x)q[M/q+D(M)], which is clearly equivalent to Eq. (3). 20. See Landes and Posner (1984), Robinson (1985), and Rose-Ackerman (1989). 21. An offsetting benefit of suits for exposure, which we ignore here, arises from the fact that injurers may be judgment-proof at the time an illness occurs, especially if the latency period is long. See, e.g., Slagel (1988). 22. More generally, the cost of litigation will deter some victims from filing suit (see, e.g., Hylton, 1990). Allowing this would significantly complicate the analysis but would not alter our qualitative results. 23. Suppose instead that q were decreasing in M. The first-order condition arising from Eq. (11) would then include an extra term, q0 (D(M) cv), which is positive given Eq. (10). Thus, the victim would choose MoM, or too little monitoring. It will be seen that a similar moral hazard problem would exist under the other two rules. 24. If q is a function of M, then expected litigation costs will no longer be independent of M under Rules 1 and 3. 25. If some victims do not file suit because of the cost of litigation, Eq. (17) would also exclude their damages (Hylton, 1990). 26. Note in particular that x from the previous section corresponds to x(0) in the current notation. 27. If some victims do not file suit, the injurer’s care would fall further below the social optimum and may even fall below the zero-litigation-cost optimum. 28. This same trade-off arises in any context where enforcement of the law is costly. Becker (1968) was one of the first to point this out when he showed that optimal deterrence of crime balances the enforcement costs (including apprehension, and, when relevant, imprisonment costs) against the benefits of deterrence.
ACKNOWLEDGMENT We acknowledge the comments of an anonymous reviewer.
Victims of Exposure to a Toxic Substance
231
REFERENCES Bathgate, J. (2001). The influence of a tort for risk on the incentive to file suit in a costly legal system. PhD dissertation, Department of Economics, University of Connecticut. Beard, T. R. (1990). Bankruptcy and care choice. Rand Journal of Economics, 21, 626–634. Becker, G. (1968). Crime and punishment: An economic approach. Journal of Political Economy, 76, 169–217. Hamrick, M. (1998). Comment: Theories of injury and recovery for post-exposure, presymptom plaintiffs: The Supreme Court takes a critical look. Cumberland Law Review, 29, 461–488. Hylton, K. (1990). The influence of litigation costs on deterrence under strict liability and under negligence. International Review of Law and Economics, 10, 161–171. Joyce, S. (2001). Medical monitoring: Are some states walking into a legal thicket? ALEC Policy Forum, 3, 20–25. Keeton, W., Dobbs, D., Keeton, R., & Owen, D. (1984). Prosser and Keeton on torts (5th ed.). St. Paul, MN: West Publishing Co. Landes, W., & Posner, R. (1984). Tort law as a regulatory regime for catastrophic personal injuries. Journal of Legal Studies, 13, 417–434. Miceli, T. (1997). Economics of the law. New York: Oxford University Press. Miceli, T. (2000). Deterrence, litigation costs, and the statute of limitations for tort suits. International Review of Law and Economics, 20, 383–394. Robinson, G. (1985). Probabilistic causation and compensation for tortious risk. Journal of Legal Studies, 14, 779–798. Rose-Ackerman, S. (1989). Dikes, dams, and vicious hogs: Entitlements and efficiency in tort law. Journal of Legal Studies, 18, 25–50. Shavell, S. (1982). The social versus private incentive to bring suit in a costly legal system. Journal of Legal Studies, 11, 333–339. Shavell, S. (1983). Torts in which victim and injurer act sequentially. Journal of Law and Economics, 26, 589–612. Shavell, S. (1986). The judgment proof problem. International Review of Law and Economics, 6, 45–58. Shavell, S. (1987). Economic analysis of accident law. Cambridge, MA: Harvard University Press. Slagel, A. (1988). Medical surveillance damages: A solution to the inadequate compensation of toxic tort victims. Indiana Law Journal, 63, 849–876. Wittman, D. (1981). Optimal pricing of sequential inputs: Last clear chance, mitigation of damages, and related doctrines in the law. Journal of Legal Studies, 10, 65–91.
This page intentionally left blank
MARKET CONCENTRATION, MULTI-MARKET PARTICIPATION AND ANTITRUST Dennis L. Weisman ABSTRACT This article explores the trade-offs between market concentration and multi-market participation in evaluating proposed mergers. For complementary demands, the price-decreasing effect of multi-market participation provides a countervailing influence on the price-increasing effect of higher concentration. The larger the footprint of the multi-market provider, the greater the likelihood the price-decreasing effect dominates. Higher concentration may be consistent with non-increasing prices despite the absence of merger economies. In the case of substitutes, multi-market participation compounds the price-increasing effect of higher concentration. Merger guidelines that place undue emphasis on market concentration can lead policymakers to block mergers that enhance consumer welfare and vice versa.
1. INTRODUCTION The horizontal merger guidelines (HMG) of the Department of Justice (DOJ) and the Federal Trade Commission (FTC) continue to place
Research in Law and Economics, Volume 22, 233–257 Copyright r 2007 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0193-5895/doi:10.1016/S0193-5895(06)22008-6
233
234
DENNIS L. WEISMAN
considerable weight on market concentration and do not recognize demand complementarities as a prospective countervailing influence. An emphasis on market concentration may be appropriate for evaluating mergers that do not involve multi-market participation. By contrast, for mergers that transform single-market providers (SMPs) into multi-market providers (MMPs), such an emphasis can lead policymakers to block mergers that actually enhance consumer welfare and vice versa. The potential for error is likely greatest in network industries, including telecommunications and transportation.1 The defining characteristic of these industries is that of demand complementarities. That is, increased traffic flows in one direction on the network generate increased traffic flows in the reverse direction and also between other nodes on the network as illustrated in Fig. 1. The fundamental question that we examine in this analysis concerns the reliability of market concentration (respectively, changes in market concentration) as an indicator of market power (respectively, changes in market power). We show that mergers that increase the market share and the ‘‘footprint’’ of MMPs can combine with demand complementarities to exert greater pricing discipline despite higher levels of market concentration.2 It is well known, of course, that higher concentration may benefit consumers if it results in merger economies, a supply-side effect. It is also well known that a
C
B
A
D
Fig. 1.
Network Traffic Flows.
Market Concentration, Multi-Market Participation and Antitrust
235
merger between two firms that produce complementary products can result in lower prices, a demand-side effect. The purpose of this article is to explicitly recognize the trade-off between market concentration and multimarket participation. In other words, at what rate should antitrust authorities trade-off increased market concentration for increased multi-market participation such that prices are non-increasing, post-merger? In the simple Cournot model of oligopoly comprised exclusively of SMPs, an increase in market concentration leads to an unambiguous increase in market power, ceteris paribus. This price-increasing effect of higher market concentration is also present when the market includes MMPs, although in this case there is a countervailing influence that must be taken into account. This countervailing influence is a price-decreasing effect that derives from the MMP’s participation in complementary markets. The MMP takes into account that a price increase in market i reduces demand in market h and vice versa, whereas the SMP does not.3 Under conditions to be described, the price-decreasing effect of multi-market participation can dominate the price-increasing effect of higher market concentration.4 In particular, welfare can rise with market concentration in this analysis if demand is redistributed from SMPs to MMPs.5 This occurs because the effective ‘‘super elasticity’’ confronting the MMPs in the case of complementary demands is higher than the own price elasticity confronting the SMPs. These issues figure prominently in the recent trend toward greater consolidation in the market for wireless telecommunications (as discussed in Section 4). A simple, stylized example should serve to illustrate one type of merger contemplated by this analysis. Suppose that United Airlines and Northwest airlines serve the route from City A to City B and that American Airlines and Southwest Airlines serve the route from City B to City C. Pre-merger, United and Northwest do not take into account the effect of their pricing on the demand for air travel from City B to City C. Similarly, American and Southwest do not take into account the effect of their pricing on the demand for air travel from City A to City B. Suppose now that the four airlines merge into one. There are two distinct and countervailing price effects associated with this merger. The price-increasing effect derives from the increased concentration on both the City A to City B route and the City B to City C route. The price-decreasing effect derives from the increase in multimarket participation and the internalization of demand externalities (i.e., demand complementarities) that the individual airlines previously had no incentive to take into account.6 Fig. 2 illustrates a merger of this type in which a duopoly of SMPs in two distinct markets merge to form a monopoly MMP that serves both markets.
236
DENNIS L. WEISMAN Market i
Market h
SMP1
SMP1
MMP
MMP
SMP2
SMP2
Fig. 2.
Merger of Duopoly SMPs into Monopoly MMP.
In Section 4, we discuss other types of mergers between firms that operate in separate, yet complementary markets. These mergers do not entail a vertical component per se, but nonetheless serve to lower prices for the final goods in both markets. The corresponding welfare gains are qualitatively similar to those associated with the elimination of double marginalization.7 The primary findings of this analysis are three. First, mergers that increase both market concentration and multi-market participation may be consistent with non-increasing, equilibrium prices, even in the absence of merger economies. Second, traditional measures of market concentration can cause policymakers to block mergers that actually enhance consumer welfare and vice versa.8 Third, there is a measurable trade-off between merger economies and demand complementarities. These findings may have important implications for recent consolidation trends,9 particularly in network industries, and should serve to inform the design of efficient antitrust policy in the ‘‘new economy’’ (Posner, 2001).
Market Concentration, Multi-Market Participation and Antitrust
237
The remainder of this article is organized as follows: The traditional Cournot analysis and the significance of market concentration and merger efficiencies are summarized in Section 2. A generalized pricing rule that accounts for multi-market participation and demand interdependence is discussed in Section 3. Section 4 discusses the policy implications of these findings. Section 5 summarizes the main findings and concludes.
2. TRADITIONAL COURNOT ANALYSIS In the simple Cournot model of oligopoly, there is assumed to be a single market in which each firm chooses an output level with the belief that its choice of output has no influence on the output choice of its rivals.10 Suppose that inverse market demand is given by P(Q), where P is price, Q is quantity with Q ¼ q1 þ q2 þ þ qn and qs is the output of firm s, s ¼ 1, y n. The cost function for firm s is given by Cs(qs) ¼ csqs. Each firm s chooses a level of output, qs, to maximize its profit, Ps ; or max Ps ¼ qs ½PðQÞ fqs g
cs ;
s ¼ 1; :::; n
(1)
It is straightforward to show that in the Cournot–Nash equilibrium,11 the mark-up of price over marginal cost, a measure of market power, is given by P
cs P
¼
ss
(2)
where ss ¼ qs/Q is the market share of firm s and ¼ ðdQ=dPÞðP=QÞ is the own price elasticity of demand (Martin, 1993, p. 21). The left-hand side of Eq. (2) is the familiar Lerner index of market power (Lerner, 1934; Carlton & Perloff, 2005, p. 283). Eq. (2) indicates that the mark-up of price over marginal cost for firm s is increasing with its market share, ceteris paribus. This is the basis for the claim that ‘‘market share is synonymous with market power.’’ The relationship in Eq. (2) must hold for each of the n firms in the market. Multiplying both sides of the expression in Eq. (2) by ss and summing over all n firms in the market yields: P Pn Pn Pn cs Þ P ns¼1 ss P Sn ðss Þ2 s¼1 ss cs s¼1 ss cs s¼1 ss ðP ¼ ¼ ¼ s¼1 (3) P P P
238
DENNIS L. WEISMAN
P since ns¼1 ss ¼ 1: Appealing to the definition of the Herfindahl–Hirschman index, we obtain: P
c¯ P
¼
H
(4)
P where c¯ ¼ ns¼1 ss cs is the weighted average industry marginal cost and H is the Herfindahl–Hirschman index. Eq. (4) indicates that the mark-up of price over average industry marginal cost is increasing with market concentration, ceteris paribus (Schmalensee, 1988, p. 660). This is a primary cause for concern with increasing market concentration. Suppose now that all firms in the market have the same marginal cost, c, and that the price elasticity of demand, e, is a constant. Rearranging the terms in Eq. (4) and solving for the market price yields: h i P¼ c (5) H Eq. (5) implies that an increase in market concentration (DH>0) must induce greater efficiencies (Dco0) if market price is to be non-increasing, post-merger. Notably, as discussed in greater detail below, the HMG explicitly allow merger efficiencies to be used as a defense for a proposed merger. The final question that we address in this section concerns the precise nature of the trade-off between market concentration and merger efficiencies necessary for the non-increasing price condition (DPr0) to be satisfied, post-merger. Taking the total differential of Eq. (5), setting the resulting expression to be less than or equal to zero, and simplifying yields: dc c dc H c H %Dc H ) ) (6) dH dP0 H c dH H c %DH H
or12
%Dc
H %DH H
(7)
Eq. (7) indicates that the non-increasing price condition is satisfied when each 1% increase in H is accompanied by a reduction in c of at least H/ (e H) percent. The following is an example. Example 1. Let c ¼ 8, e ¼ 2 and H ¼ 0.4. This yields an equilibrium market price of 10 upon applying Eq. (5). Suppose that a merger is proposed that would increase market concentration by 10% to H ¼ 0.44. Absent any change in marginal cost, price would rise to approximately
Market Concentration, Multi-Market Participation and Antitrust
239
10.26. Conversely, if costs decrease to 7.8, a reduction of 2.5%, following the increase in market concentration, the market price remains unchanged as may be confirmed by Eq. (5). It follows from Eq. (7) that costs must fall by at least 2.5% in order for a 10% increase in H not to generate an increase in market price.
3. A GENERALIZED PRICING RULE The traditional analysis discussed in the previous section is based on the strong assumption that there is only a single market and hence no scope for multi-market provisioning or demand interdependence. In many cases, a more realistic assumption may well be that there are multiple markets and there is demand interdependence across markets. The primary purpose of this section is to present just such a generalization.
3.1. Economic Analysis Suppose that there are z distinct markets, where z>1 is a positive integer, with inverse demand functions, Pi(Q1, y, Qz), where Qi is the output in market i, i ¼ 1, y, z. There are nis identical SMPs and nim identical MMPs, where nis nim :13 The output of each SMP and each MMP in market i is qis and qim ; respectively. The cost functions for the representative SMP and MMP are given, respectively, by C s ðqis Þ ¼ cs qis ¼ cqis and C m ðqim Þ ¼ cm qim ¼ ð1 sÞcqim ; where so1. The generalized mark-up rule, the multi-market counterpart to Eq. (4),14 is given by15 ! nim z X Pi c¯ H i X shm Rh i s ¼ ; i; h ¼ 1; . . . ; z (8) ii m¼1 m hai ih Ri Pi where c¯ is the weighted-average industry marginal cost. The left-hand side of Eq. (8) is once again the familiar Lerner index. The first term on the righthand side of Eq. (8) is identical to that in Eq. (4), where Hi is the Herfindahl–Hirschman index in market i and eii is the own price elasticity of demand in market i as previously defined. The second term on the righthand side of Eq. (8) is an adjustment to the simple mark-up rule to account for multi-market participation and demand interdependence across markets. The term sim is the market share of the representative MMP in market i and
240
DENNIS L. WEISMAN
shm is the market share of the representative MMP in market h, where hai: The term ih ¼ ð@Ph =@Qi Þ 1 ðPh =Qi Þ in Eq. (8) is the cross-demand elasticity.16 In the case of complements, eih>0, and in the case of substitutes, eiho0.17 Rh and Ri denote the revenues in markets h and i, respectively. Recognize that when there is no multi-market participation, sim ¼ 0 for all markets i and the generalized mark-up rule in Eq. (8) reduces to the simple mark-up rule in Eq. (4). In addition, when the products in markets i and h are independent,18 the generalized mark-up rule in Eq. (8) again reduces to the simple mark-up rule in Eq. (4). A careful examination of the right-hand side of Eq. (8) reveals that the first term is positive and the second term is negative in the case of complements, eih>0. Hence, the larger the footprint of the MMPs in complementary (substitute) markets, as measured by the term in parenthesis in Eq. (8), the lower (higher) the equilibrium price in market i, ceteris paribus. Furthermore, a necessary condition for an increase in market concentration to result in a decrease in the equilibrium price when demands are complementary is that the collective market share of the MMPs increases, post-merger. This suggests a trade-off between market concentration and multi-market participation. We turn now to a careful examination of this trade-off. Solving Eq. (8) for Pi yields the generalized pricing rule, or 2 !3 1 nim z i h h X X H s R m 5 c¯ ; i; h ¼ 1; . . . ; z (9) þ si Pi ðnis ; nim Þ ¼ 41 ii m¼1 m hai ih Ri
where Pi ðnis ; nim Þ denotes the equilibrium market price when there are nis SMPs and nim MMPs serving market i. 3.2. Symmetric Costs
In this section, we assume that costs are symmetric so that s ¼ 0 and hence there are neither economies nor diseconomies associated with multi-market provisioning. To facilitate the exposition without significant loss of generality, we make two simplifying assumptions. First, we assume that the own and cross-price elasticities, eii and eih, are constants. Second, we assume that all markets i are identical. We are primarily interested in the conditions that must prevail in order for the non-increasing price condition to be satisfied, or Pi ðnis ; 0Þ Pi ð0; nim Þ
(10)
Market Concentration, Multi-Market Participation and Antitrust
241
In other words, under what conditions will a market that is served exclusively by nim MMPs result in an equilibrium price that is no higher than a market served exclusively by nis SMPs?19 Satisfaction of the condition in Eq. (10) implies from Eq. (9) that 2 !3 1 i nim z h h X X H i ii sm R 5 nm þ sim c 41 c (11) ii Ri ii H inis m¼1 hai ih
where H inis and H inim correspond to the Herfindahl–Hirschman index when the market is served exclusively by nis SMPs and when the market is served exclusively nim MMPs, respectively. Simplifying Eq. (11) upon recognizPnby i m i ing that m sm ¼ 1 and Rhm ¼ Rim since the markets are identical yields: 1 1 nis ii nim ii
z 1 nim ih
(12)
since H ini ¼ 1=nis and H ini ¼ 1=nim when all SMPs are identical and all s m MMPs are identical, respectively. Solving Eq. (12) for eih>0 yields, after some algebraic manipulation, ih
nis ðz nis
1Þii nim
(13)
Eq. (13) provides an upper bound on the cross-demand elasticity sufficient for the non-increasing price condition to be satisfied. A number of other useful results follow directly from Eq. (13). First, a market served exclusively by nim MMPs yields a lower equilibrium price than a market served exclusively by nis ¼ nim SMPs when demands are complementary.20 To see this, recognize that when nis ¼ nim ; the right-hand side of Eq. (13) approaches infinity. Hence, the condition in Eq. (13) is always satisfied. This result derives from the fact that the complementary nature of demand disciplines the pricing behavior of the MMPs, but exerts no corresponding discipline on the behavior of the SMPs. In other words, the MMPs have an incentive to internalize demand externalities through lower prices, while there is no comparable incentive affecting the behavior of the SMPs. The following is an example. Example 2. Let c ¼ 10, eii ¼ 1.5, eih ¼ 3 and z ¼ 2. Suppose that market i is initially served by 4 SMPs and 0 MMPs so that nis ¼ 4 and nim ¼ 0: This implies that Hi ¼ 0.25. A merger is proposed that would result in the market being served exclusively by MMPs. It follows from Eq. (9) that Pi ð0; nim Þ Pi ð4; 0Þ ¼ 12 for all nim 2:21 Moreover, when nim ¼ 2;
242
DENNIS L. WEISMAN
Pi ð0; 2Þ ¼ Pi ð4; 0Þ ¼ 12 and Hi ¼ 0.5. Hence, the Herfindahl–Hirschman index doubles with the proposed merger while the market price is unchanged. In contrast, traditional merger analysis applied to these data would predict a post-merger market price of 15, or a price increase of 25%. Hence, in the case of complementary demands, traditional merger analysis overstates the upward pricing pressures resulting from higher market concentration.22 Second, the larger the eii and the smaller the eih, the more likely Eq. (13) is to be satisfied, ceteris paribus. The economic intuition is as follows. The larger the eii, the fewer the absolute number of market providers required to sustain any given level of pricing discipline. The smaller the eih, the greater the demand complementarities and the stronger the pricing discipline exerted by the MMPs’ participation in complementary markets.23 Third, if z is ‘‘sufficiently large’’ and eih is ‘‘sufficiently small,’’ the equilibrium price will be lower in a market served by a monopoly MMP than in a relatively unconcentrated market served exclusively by SMPs. The following is an example. Example 3. Let c ¼ 10, eii ¼ 1.5 and eih ¼ 3. Suppose that market i is initially served by 4 SMPs and 0 MMPs so that ns ¼ 4 and nm ¼ 0. This implies that Hi ¼ 0.25 and Pi ¼ 12. A proposed merger would result in the market being served exclusively by a monopoly MMP. It is straightforward to show from Eq. (9) that Pi ð0; 1ÞoPi ð4; 0Þ ¼ 12 for all z 3:24 Observe further that since this relation holds for symmetric costs, it must also hold for multi-market diseconomies that are ‘‘sufficiently small.’’25 3.3. Asymmetric Costs In this subsection, we relax the assumption of symmetric costs and allow for the possibility that there are economies (s>0) or diseconomies (so0) associated with multi-market provisioning. Following Eq. (9), Pi ðnis ; 0Þ Pi ð0; nim Þ whenever ii 1 z 1 1 ð1 sÞc (14) þ c 1 nim ii nim ih ii 1=nis
since H inis ¼ 1=nim and H inm ¼ 1=nim : Solving Eq. (14) for eih>0 yields, after some algebraic manipulation, i ns ðz 1Þii nis nim ih (15) nis nim ðnis nim Þ snim ðnis ii 1Þ
Market Concentration, Multi-Market Participation and Antitrust
243
when s>0.26 Eq. (15) provides an upper bound on the cross-demand elasticity sufficient for the non-increasing price condition to be satisfied when there are multi-market economies.27 Comparing Eqs. (13) and (15) reveals that i ni ðz 1Þii ns ðz 1Þii nis nim o ih s i (16) ns nim nis nim ðnis nim Þ snim ðnis ii 1Þ since ðnis nim Þ=½ðnis nim Þ snim ðnis ii 1Þ41 for s40 ‘‘sufficiently small.’’ Eq. (16) confirms that when there are economies associated with multi-market provisioning, the demand complementarities required to satisfy the nonincreasing price condition are smaller than those indicated by Eq. (13). That is, the upper bound on eih is correspondingly higher.28 This implies that there is a trade-off between multi-market economies and demand complementarities. The following is an example. Example 4. Assume the same data set as in Example 2, except s ¼ 0.1 and eih is now a variable. The marginal cost for the representative MMP is given by (1 0.1)10 ¼ 9. It follows from Eq. (9) that Pi ð0; 2Þ Pi ð4; 0Þ ¼ 12 for all eihr6. Recall from Example 2 that Pi ð0; 2Þ Pi ð4; 0Þ ¼ 12 for all eihr3. Hence, the upper bound on eih is increasing with s, ceteris paribus. In other words, the greater the multi-market economies, the weaker the demand complementarities required to satisfy the non-increasing price condition. 3.4. Summary A brief summary of the main findings of this analysis is useful for the discussion of policy implications in the next section. (1) When demands are complementary, multi-market participation provides a countervailing influence on the price-increasing effect of higher market concentration. Traditional merger analysis will tend to overstate the upward pricing pressures resulting from higher market concentration under these conditions. The implication is that higher market concentration (including possibly market monopolization) need not lead to higher prices, even in the absence of merger economies. (2) When demands are substitutable, multi-market participation compounds the price-increasing effect of higher market concentration. Traditional merger analysis will tend to understate the upward pricing pressures resulting from higher market concentration under these conditions. The implication is that higher market concentration will lead to higher prices unless accompanied by non-trivial merger economies.
244
DENNIS L. WEISMAN
(3) The non-increasing price condition is satisfied if the equilibrium price when the market is served exclusively by nm MMPs is no higher than the equilibrium price when the market is served exclusively by nsZnm SMPs. The greater the multi-market economies and the stronger the demand complementarities, the smaller the number of MMPs required to satisfy the non-increasing price condition. The implication is that there is a tradeoff between multi-market economies and demand complementarities. (4) The potential for error from the use of traditional merger analysis is likely most pronounced when there is a high degree of demand interdependence and the footprint of the MMPs is large.
4. POLICY IMPLICATIONS The findings in the previous section attest to the fact that demand complementarities provide a countervailing influence on the upward pricing pressures typically associated with increased market concentration. In the case of complementary demands, the larger footprint of the MMP forces it to account for the fact that a higher price in market i reduces demand in market h. As a result, equilibrium prices may be decreasing while market concentration is increasing, even in the absence of merger economies. These findings suggest that antitrust guidelines that place undue emphasis on market concentration can lead policymakers to block mergers that have the potential to enhance consumer welfare and vice versa – an outcome seemingly at odds with the goals of the antitrust laws.29 Moreover, these findings indicate that even consolidation to monopoly can potentially benefit consumers through lower prices. Recognize that this type of market consolidation would pose no difficulties for Judge Bork’s definition of ‘‘competition.’’30 Moreover, this argument applies whether competition is equated with economic efficiency (Bork) or whether competition is equated with the well-being of consumers in the relevant market (Lande, 1982). The above possibility, while perhaps intriguing, would still have to be reconciled with the specific wording contained in Section 7 of the Clayton Act that proscribes acquisitions ‘‘wherein any line of commerce or in any activity affecting commerce in any section of the country, the effect of such acquisition may be substantially to lessen competition, or tend to create a monopoly.’’31 It is noteworthy that this very same issue arises with respect to the ‘‘efficiencies defense.’’ The HMG resolve that issue by indicating that efficiencies might sometimes justify a merger to monopoly, though the situation would be very rare.
Market Concentration, Multi-Market Participation and Antitrust
245
Hence, it may be necessary for policymakers to go beyond calculating simple measures of market concentration and investigate the distribution of market concentration across SMPs and MMPs, post-merger.
4.1. Contemporary Applications In terms of current applications, mergers in network industries,32 including telecommunications (Lehman & Weisman, 2000),33,34 commercial airlines (Morrison & Winston, 2000),35 and railroads (Grimm & Winston, 2000; Park, Babcock, Lemke, & Weisman, 2001) would seem to be particularly noteworthy. This is the case because increased traffic flows from node A to node B on a telecommunications or transportation network generate increased traffic flows from node B to node A and also increased traffic flows between other nodes on the network as illustrated in Fig. 1. It is significant that each of these industries is characterized by demand complementarities and multi-market participation of the sort shown to be most damning to traditional merger analysis. 4.1.1. Wireless Telecommunications With respect to the telecommunications industry, and wireless telecommunications, in particular, the industry appears poised for significant consolidation as market providers seek to expand the size of their footprint.36 There are currently four wireless telecommunications providers in the U.S. with a ‘‘national footprint,’’37 following Sprint’s recent acquisition of Nextel, and a large number of non-national or regional providers.38,39 It is anticipated that any wholesale movement to consolidate would invite antitrust scrutiny as policymakers may be concerned that higher levels of concentration will lead to higher prices.40 The findings of this analysis suggest that the price-decreasing effect of multi-market participation may dominate the price-increasing effect of greater concentration. In other words, reducing the number of independent providers through consolidation will allow for the internalization of demand externalities and possibly lower prices, despite reduced competition.41 4.1.2. Wireline Telecommunications There has also been considerable consolidation among wireline telecommunications providers.42 Additional mergers have recently been consummated. Specifically, Verizon has acquired MCI and SBC (now ‘‘at&t’’) has acquired AT&T – its former parent. These mergers are controversial in some
246
DENNIS L. WEISMAN
respects and yet uncontroversial in others. It is instructive to digress briefly to comprehend fully the source of the controversy. In 1984, the government carried out the largest court-ordered breakup of a corporation in U.S. history when it forced AT&T to divest itself of its seven local operating companies.43 In essence, the government wanted to separate long-distance from local telephone service in order to foster competition in long-distance and telephone equipment markets. William Baxter, the Assistant Attorney-General for Antitrust in the Reagan administration and the architect of the divestiture accord, faced ambivalence if not downright opposition about the proposed divestiture of AT&T from his superiors at Justice and cabinet members in the Reagan administration.44 Writing on the fifth anniversary of the breakup of the Bell System, Baxter reflected on the rationale for the AT&T divestiture. The decree implicitly made a wager that the regulatory distortions of those portions of the economy which could have been workably competitive, yielded social losses in excess of the magnitude of the economies of scope that would be sacrificed by this approach. It was a wager, a guess. It would be absurd to pretend it was made on the basis of detailed econometric data. It was not; we did not have the data. Of course, all other courses from that point were also guesses. Clear proof was not about to become available anytime soon. It was a judgment call, and I guess, in some senses, I do not yet know. Maybe we will never know whether it was right or wrong.45
The pending mergers in wireline telecommunications markets may strike some observers as simply an exercise in putting the old Bell System back together again. It may also be tempting to conclude that this consolidation is merely an attempt to reclaim the economies of scope associated with long distance and local telephone service that Baxter was willing to sacrifice at the AT&T divestiture This, however, would be an overly simplistic view of the ‘‘creative destructive’’ now underway in telecommunications markets.46 This is necessarily the case for at least three reasons. First, given the emerging competition in local telephone service markets in the form of wireless and VoIP (voice over Internet protocol) and the largescale entry of the cable companies into telecommunications markets, it may no long be necessary to sacrifice economies of scope in order to foster ‘‘workable competition’’ in these markets.47 Second, MCI and AT&T have suffered financially in recent years and there is legitimate reason to question whether these carriers would have a viable business model as stand-alone firms. Should this be the case, any proposed merger could presumably avail itself of the ‘‘failing firm’’ provisions in the HMG.48,49
Market Concentration, Multi-Market Participation and Antitrust
247
Third, and most importantly for the extant analysis, these proposed mergers involve trade-offs between market concentration and multi-market participation. AT&T and MCI have a(n) (inter)national, as opposed to a regional, presence. Heretofore, the so-called Regional Bell Operating Companies, which include BellSouth, Qwest, Verizon and SBC (now ‘‘at&t’’), have been somewhat reticent to enter each other’s territories. Given that AT&T and MCI have numerous national accounts, arguably their most highly prized assets, such consolidation could serve to end this apparent selfimposed quarantine. Hence, while these mergers clearly increase concentration in selected local and long-distance telephone service markets, they also increase multi-market participation in other markets in a manner that may well be pro-competitive rather than anti-competitive. Given the cost structure of the telecommunications industry and the pentup demand for one-stop shopping for telecommunications services,50 the degree of consolidation that has occurred in the industry was perhaps foreseeable and, at least, in a certain sense a product of the AT&T divestiture itself: The AT&T divestiture may eventually be seen as an experiment in industry genetics, the long-run effect of which was to clone multiple copies of the former Bell System. Divestiture brought to an end AT&T’s control over end-to-end connectivity in order to eliminate its ability to discriminate against its rivals. Some ten years after divestiture, the industry trend is one of vertically-integrated supply, as consortia are forming once again to provide for end-to-end connectivity. The Bell System is no longer, but its progeny lives on, albeit more combative, market oriented and financially focused than its ancestry. In this sense, we have come full circle in the telecommunications industry from end-toend connectivity to partitioned markets, back to end-to-end connectivityy. We have seemingly arrived at a parallel universe in which there is not one Bell System, but several. Questions of terms and timing remain, but these are merely the rules of engagement. A Galactic battle of the titans seems inevitable.51
4.1.3. Railroads Concerns about the possible adverse effects of further consolidation among railroads recently led the Surface Transportation Board (STB) to revise its policies governing mergers and acquisitions.52 Our revised rules reflect a significant change in the way in which we will apply the statutory public interest test to any major rail merger application. Because of the small number of remaining Class I railroads, y we believe that future merger applicants should bear a heavier burden to show that a major rail combination is consistent with the public interest. Our shift in policy places greater emphasis in the public interest assessment on enhancing competition while ensuring a stable and balanced rail transportation system.53
248
DENNIS L. WEISMAN
However, we know from the last round of mergers that another merger involving two very large railroads would not likely be an isolated event, but instead would trigger responsive proposals that, if granted, could lead to a transcontinental railroad duopoly.54
The STB further noted that it ‘‘would require applicants in future merger proceedings to present proposals that enhance, not merely preserve, competition, in order to secure our approval.’’55,56 The key premise underlying the STB’s revised merger policy is apparently that reduced competition in the industry would necessarily lead to higher prices, in part, because the ‘‘efficiencies y likely to be realized from further downsizing of rail route systems are limited’’ (STB, 2001, p. 14). The findings of this analysis suggest that further consolidation among railroads – even consolidation to a ‘‘transcontinental railroad duopoly’’ – could potentially lead to lower prices even if such consolidation fails to yield merger economies.57 4.2. Implementation Issues It is commonplace today for company press releases announcing proposed mergers to make reference to prospective efficiency gains, synergies and cost savings. These references are designed, of course, to emphasize a countervailing influence on the increased concentration that the merger inevitably creates. The problems with the ‘‘efficiency defense’’ are well known. First, these gains are only speculative in nature and the merger firms are in possession of better information than the antitrust authorities. The HMG underscore the fact that: Efficiencies are difficult to verify and quantify, in part because much of the information relating to efficiencies is uniquely in the possession of the merging firms. Moreover, efficiencies projected reasonably and in good faith by the merging firms may not be realized y Efficiency claims will not be considered if they are vague or speculative or otherwise cannot be verified by reasonable means.58
Second, the merging firms carry the burden for credibly demonstrating that these efficiencies could not have been realized absent the merger. Demand complementarities are a different animal altogether and in many, though perhaps not all, cases it will be clear that the efficiency gains from internalization of demand externalities cannot be achieved absent a merger. For example, wireless licenses are limited in number due to scarcity of electromagnetic spectrum, so there may be no real prospect that a firm operating in one market could begin operating in another market absent a merger or the purchase of the spectrum rights currently in possession of another firm, which would have essentially the same effect as a merger.
Market Concentration, Multi-Market Participation and Antitrust
249
These findings would seem to suggest that evidence of demand complementarities could substitute, at least in part, for evidence of merger economies.59 In other words, a proposed merger that entails demand complementarities would presumably carry less of a burden in demonstrating merger efficiencies than a merger that entails demand substitutes or independent demands. The precise nature of the trade-off between merger efficiencies and demand complementarities would have to be worked out and may vary on a case-by-case basis depending upon the initial market structure. For example, there may be some merit in a sliding-scale approach in which mergers to monopoly or duopoly require stronger evidence of demand complementarities than mergers that reduce the number of providers in the market from say four to three.60 Another intriguing possibility suggested by these findings concerns the prospect that the government could actually oppose a divestiture on grounds that it would serve to place upward pressure on prices.61 Suppose a merger raises concentration in the relevant market but also creates an MMP, thereby lowering prices. In addition, suppose that at some future point in time, the merged firm proposes to divest all its operations in markets other than the relevant market. In order to prevent a price increase, the government could conceivably oppose such a divestiture using the same rationale that it used initially to approve the merger.62 In order to prevail, the firm would then have the burden of proof for demonstrating that market and/or production conditions had materially changed since the time the merger was approved.
5. CONCLUSION Concerns about market concentration have a long and revered history in the United States – dating back to the earliest days of the new republic and the sentiment that the concentration of economic power invariably leads to the concentration of political power. While the more modern interpretation of the proper role of the antitrust laws emphasizes economic over sociopolitical considerations, concerns over the adverse economic effects of increased market concentration have not abated. Recent mergers in network industries, including telecommunications and transportation, have triggered a re-examination of age-old questions and prompted interest in new ones. A primary objective of this analysis is to investigate the trade-off between market concentration and multi-market participation when demands are interdependent. A key finding is that
250
DENNIS L. WEISMAN
mergers that increase both market concentration and multi-market participation may be consistent with non-increasing, equilibrium prices even in the absence of merger economies. The larger footprint of the merging firms provides a countervailing influence on the upward pricing pressures typically associated with greater market concentration. [In the case of substitutable demands, this effect is reversed so that the larger footprint has a compounding influence on the upward pricing pressures typically associated with greater market concentration.] These findings may call into question, at least in certain industries, the emphasis that the HMG place on market concentration in evaluating proposed mergers. This article suggests a plausible, ‘‘pro-competitive’’ rationale for recent consolidation trends in network industries that depends not on the realization of merger economies but on the recognition of demand complementarities. The precise nature of the trade-off between merger economies and demand complementarities and its role in evaluating the merits of proposed mergers is an important topic for further research. The findings of this research should serve to inform the design of efficient antitrust policies in the ‘‘new economy.’’
NOTES 1. There has been significant merger activity in the telecommunications and transportation industries in recent years. See, for example, Dreazen, Ip, and Kulish (2002), Dreazen (2002) and Weisman (1999). 2. The term ‘‘footprint’’ in this context refers to the degree to which the MMP participates in other markets. 3. These are sometimes referred to as ‘‘network effects’’ or network externalities. See Liebowitz and Margolis (2002) for a comprehensive survey of the literature. For a discussion of network effects and their prominent role in recent high-profile court cases, see Shelanski and Sidak (2001). 4. In contrast, when demands are substitutable, multi-market participation serves to compound the price-increasing effect of higher market concentration. 5. The parallels with Farrell and Shapiro (1990) are noteworthy. Specifically, welfare can rise with market concentration in their framework if demand is redistributed from relatively inefficient to relatively efficient firms. 6. Fred Kahn has suggested an alternative interpretation in which the A to B route is treated as an input for the B to C route and vice versa. Under this interpretation, the incentive for the ‘‘vertical’’ component of this merger is the elimination of double marginalization rather than the internalization of demand externalities. The term ‘‘double marginalization’’ refers to a scenario in which a firm with market power in the input market supplies inputs to a separate firm with market power in the output market. The possession of market power at both stages of production
Market Concentration, Multi-Market Participation and Antitrust
251
implies that there are two mark-ups (double marginalization) reflected in the final output price. In this case, a vertical merger between the input supplier and the output supplier would eliminate the double marginalization, reduce price, raise profits and hence increase economic welfare (Tirole, 1988, pp. 173–181). 7. The two cases are similar in that internalization of demand externalities and the elimination of double marginalization both lead to lower final product prices. The two cases are different in that internalizing demand complementarities are concerned with two final products, whereas eliminating double marginalization is concerned with an input and a final product. See Economides (2005) for a discussion of the incentives for vertical integration. 8. See Crandall and Winston (2003) for some recent evidence that antitrust merger policy does not enhance consumer welfare. 9. White (2002) conducted a recent analysis of aggregate concentration trends in the U.S. economy. Despite significant merger activity in selected industries, he finds no evidence of a wholesale increase in aggregate market concentration. 10. The firms in this model compete in a standard strategic substitutes Cournot game with homogenous output (Bulow, Geanakoplos, & Klemperer, 1985; Vives, 1999). 11. In the Nash equilibrium of the Cournot game, each firm chooses an output level that maximizes its profit given the output choice of each of its rivals. A Nash equilibrium thus represents a simultaneously rational choice of output for each firm in the market. 12. See Williamson (1968) for an early formal analysis of the trade-off between market concentration and merger efficiencies. 13. An SMP in market i serves only market i. An MMP in market i serves market i and at least one other market h, i6¼h. 14. This is based on a slight variation of the model derived in Weisman (2003). Details are available upon request from the author. 15. Tirole (1988, p. 70) derives a mark-up rule for a multi-product monopolist with interdependent demands. When the goods are complements, the multi-product monopolist sets a lower price than a single-product monopolist operating independently in each market. The complementary nature of demand forces the multi-product monopolist, but not the single-product monopolist, to account for the fact that a higher price in market i reduces demand in market h. See also Allen (1938, pp. 59–62) for an early analysis of the behavior of complementary-demand monopolists. The logic underlying this analysis is similar except that it is cast in terms of MMPs and SMPs rather than multi-product and single-product monopolists, respectively. 16. The cross-demand elasticity measures the percentage change in quantity demanded in market i with respect to a 1% change in demand in market h. This differs from the more familiar cross-price elasticity, which measures the percentage change in quantity demanded in market i with respect to a 1% change in price in market h. 17. It is important to recognize that the degree of complementarity (substitutability) increases as the cross-demand elasticity decreases in absolute value. 18. Under these conditions, a change in quantity in market i has no effect on (inverse) demand in market h, or @Ph =@Qi ¼ 0 ) ih ! 1: 19. It is possible to examine this trade-off when SMPs and MMPs operate simultaneously in a given market, but this requires imposing additional structure on
252
DENNIS L. WEISMAN
the demand functions. For example, Weisman (2005) derives a simple expression for the marginal rate of substitution of MMPs for SMPs that depends only on the parameters of the symmetric, linear demand functions. 20. The opposite result holds when demands are substitutable. 21. The data imply that Pi ð0; nim ÞoPi ð4; 0Þ for all nim 42 when costs are symmetric (s ¼ 0). It follows from the continuity of the price function that this relationship must also hold for multi-market diseconomies ‘‘sufficiently small’’ (i.e., values of so0, but small in absolute value). For example, it follows from Eq. (9) that Pi ð0; 3ÞoPi ð4; 0Þ ¼ 12 for all values of s> 0.0667. 22. Alternatively, if eih ¼ 3 then Pi ð0; 2Þ ¼ 20415; ceteris paribus. Hence, in the case of substitutes, traditional merger analysis understates the upward pricing pressures resulting from higher market concentration. 23. Recall that, in contrast to the traditional cross-price elasticity, with the crossdemand elasticity the degree of complementarity (respectively, substitutability) increases as the elasticity measure approaches zero. See notes 16 and 17 supra. 24. The critical value of z is actually 2.5, but recall that z is constrained to take on only integer values. 25. See the related discussion in note 21 supra. 26. Note that the condition in Eq. (15) reduces to that in Eq. (13) when s ¼ 0. 27. Note that for s>0 ‘‘sufficiently large,’’ the right-hand side of Eq. (15) will be negative. In this case, a merger will satisfy the non-increasing price condition provided that demand substitutability is not too strong, or eiho0 is ‘‘sufficiently large in absolute value.’’ 28. In the case of multi-market diseconomies ðso0Þ; ðnis nim Þ=½ðnis nim Þ snim ðnis ii 1Þo1 and the upper bound on eih is correspondingly lower. 29. Senator Sherman argued that the antitrust laws should serve to prohibit arrangements that ‘‘tend to advance the cost to the consumer y.’’ (Thorelli, 1955, p. 166). 30. Bork argues that ‘‘competition’’ for purposes of antitrust analysis, must be understood as a term of art signifying any state of affairs in which consumer welfare cannot be increased by judicial decree. Bork therefore rejects the idea that ‘‘competition’’ is synonymous with ‘‘rivalry’’ (1978, p. 58). 31. 15 USCS y 18 (2003). 32. This is not to suggest that the applications discussed herein are necessarily restricted to network industries. Consider, for example, the possibility that consumption of a particular good in location (market) A increases the likelihood of consumption of that good in location (market) B, due to brand recognition, product familiarity, etc. In this respect, the increasing prevalence of ‘‘chain stores’’ likely reflects the leveraging of both demand-side and supply-side economies. 33. Empirical demand analysis in the telecommunications industry confirms the existence of demand complementarities in the form of point-to-point traffic patterns. See, for example, Taylor (1994) and Larson, Lehman, and Weisman (1990). 34. There is some evidence of revisionist thinking among policymakers as it relates to mergers in the telecommunications industry, although the rationale – financial distress and competition – is different from that suggested herein. In 1997, shortly after AT&T and SBC announced a proposed merger, Reed Hundt, then chairman of the Federal Communications Commission (FCC), remarked that a merger between
Market Concentration, Multi-Market Participation and Antitrust
253
AT&T and a Bell Company would be ‘‘unthinkable.’’ See Telecommunications Reports, June 23, 1997. More recently, Michael Powell, the immediate past FCC chairman, stated that no telecommunications merger proposal would be ‘‘deemed unthinkable’’ by his agency. In addition, Deborah Majoras, then Deputy Assistant Attorney General for Antitrust, intimated that a more lenient policy toward mergers in the telecommunications industry may be in the offing (Kaplan, 2002). See Huber, Kellogg, and Thorne (1999, Chapter 7) for a review of the specific standards governing mergers and acquisitions in the telecommunications industry. 35. These findings may have implications not only for mergers, but also for alliances between commercial airlines. For example, Brueckner and Whalen (2000) found that international alliances can reduce interline airfares without necessarily raising fares in those markets in which the alliance partners compete directly. 36. The recent merger between AT&T Wireless and Cingular created the largest cellphone provider in the U.S. See Latour, Drucker, Sidel, and Raghavan (2004), The Wall Street Journal (2004a) and FCC (2004b). 37. These are Cingular, Sprint, T-Mobile and Verizon Wireless. See FCC (2004a, 36). 38. These carriers collectively operate 3,123 wireless systems in the U.S. and serve in excess of 202 million subscribers as of January 22, 2006. See CTIA at http:// www.ctia.org/ and CTIA (2004). 39. According to the FCC, 276 million people, or 97% of the population in the U.S., live in counties in which there are three or more wireless providers. Approximately 250 million people, or 87% of the population in the U.S., live in counties in which there are five or more wireless providers. More than 216 million people, or 76% of the population in the U.S., can now choose from among six or more different wireless providers. Finally, 84 million people, or almost 30% of the population, live in counties served by seven or more different wireless providers. See FCC (2004a, 49). This increasing competition has led to a pronounced reduction in prices. For example, average revenue per minute declined from $0.47 per minute in 1994 to $0.10 at the beginning of 2003, a reduction of 79%. See FCC (2004a, 171). 40. Latour et al. (2004) caution that ‘‘industry consolidation could lead to higher prices for consumers.’’ 41. Zimmerman (2005) investigates the implications of wireline–wireless affiliated companies for intermodal competition. He finds some evidence to suggest that Cingular and Verizon Wireless may have designed their wireless offerings to mitigate cannibalization of their parents’ wireline business. In other words, their strategy is to position wireless as a complement rather than a substitute for wireline telephone to dissuade customers from ‘‘cutting the cord.’’ 42. Of the seven original Bell Operating Companies spun off as part of the AT&T divestiture, only four remain: BellSouth, Qwest (formerly US West), SBC (now ‘‘at&t’’) and Verizon. See Lehman and Weisman (2000, Chapter 2). 43. With this divestiture, the government attempted to separate the ‘‘competitive’’ from the ‘‘non-competitive’’ sectors of the telecommunications industry. See Brock (1994, Chapter 9) and Temin (1987) for a historical account of the AT&T divestiture. 44. Ibid. See also Sappington and Weisman (1996, Chapter 2) for an overview of these developments. 45. Baxter (1991, p. 30). 46. See Neuchterlein and Weiser (2005, Chapter 1).
254
DENNIS L. WEISMAN
47. The long-distance market in the U.S. was deregulated in late 1995 and there is no longer any serious debate among informed observers as to whether this market is rivalrous. 48. Department of Justice and Federal Trade Commission (1992, y 5.1). 49. Despite the fact that AT&T and MCI may no longer be viable, stand-alone, business enterprises going forward, the government still has an interest in ensuring that any subsequent transfer of assets does not serve to facilitate market dominance in the telecommunications sector. Qwest raised this argument in the course of its bidding war with Verizon for the acquisition of MCI. See, for example, Notebaert (2005). Notably, SBC (now ‘‘at&t’’) and Verizon are the two largest telephone companies in the U.S. 50. See Ware (1998) for a discussion of diversification trends in telecommunications markets. 51. Sappington and Weisman (1996, pp. 68–69). 52. As a result of consolidation, the number of Class I railroads in the U.S. declined from 40 in 1980 to 12 in 1993 (Association of American Railroads, 1981, p. 2; 1994, p. 3). [Class I railroads are defined by operating revenue thresholds that are adjusted annually for inflation. In 2002, a class I railroad was defined as any railroad with at least $272 million in annual revenues (Association of American Railroads, 2003, p. 3)] In 2003, there were only seven remaining class I railroads in the U.S. These are the Norfolk Southern, the Kansas City Southern, the Burlington Northern/Santa Fe, the Canadian National, the Soo Line (owned by the Canadian Pacific), the Union Pacific and CSX Transportation (STB, 2002, p. 3). 53. STB (2001, p. 9). 54. STB (2001, p. 43). 55. STB (2001, p. 10). 56. The STB goes on to note that whereas their previous policy statement on mergers focused on ‘‘greater economic efficiency’’ and ‘‘improved service’’ as the most likely and significant public service benefits, the new policy statement adds enhanced competition as an important public interest benefit (STB, 2001, p. 14). 57. In fact, despite significant consolidation in the railroad industry, inflation-adjusted, railroad rates have decreased by more than 45% since 1984. See STB (2001, note 11). 58. Department of Justice and Federal Trade Commission (1992/1997, y4). 59. Nonetheless, it should be recognized that demand complementarities, while perhaps somewhat less speculative in nature than merger efficiencies, are potentially subject to measurement problems of their own. These include limited data availability and econometric estimation of the underlying demand system. 60. I am grateful to an anonymous referee for this insight. 61. I am grateful to the co-editor, Jack Kirkwood, for suggesting this intriguing possibility. 62. Presumably, the government could also oppose a divestiture on grounds that loss of economies of scale or scope would lead to higher prices.
ACKNOWLEDGMENTS The author is grateful to Michael Babcock, Philip Gayle, Alfred Kahn, Andrew Kleit, Dale Lehman, Michael Redisch, David Sappington and
Market Concentration, Multi-Market Participation and Antitrust
255
Kelley Weber for helpful discussions. The author is also grateful to the coeditors Jack Kirkwood and Dick Zerbe, and an anonymous referee for providing many constructive suggestions for revision. The usual caveat applies.
REFERENCES Allen, R. (1938). Mathematical analysis for economists. New York: St. Martin’s Press. Association of American Railroads. (1981). Railroad Facts. Association of American Railroads. (1994). Railroad Facts. Association of American Railroads. (2003). Railroad Facts. Baxter, W. (1991). Questions and answers with the three major figures of divestiture. In: B. Cole (Ed.), After the break-up: Assessing the new post-AT&T divestiture era (pp. 21–49). Columbia University Press. Bork, R. (1978). The antitrust paradox. New York: Macmillan. Brock, G. (1994). Telecommunications policy for the information age. Cambridge, MA: Harvard University Press. Brueckner, J., & Whalen, W. (2000). The price effects of international airline alliances. Journal of Law and Economics, 43(October), 503–545. Bulow, J., Geanakoplos, J., & Klemperer, P. (1985). Multimarket oligopoly: Strategic substitutes and complements. The Journal of Political Economy, 93(3), 488–511. Carlton, D., & Perloff, J. (2005). Modern industrial organization (4th ed.). Boston, MA: Addison-Wesley. Cellular Telecommunications and Internet Association (CTIA). (2004). www.wow-com.com Clayton Act. 15 USCS y 18 (2003). Crandall, R., & Winston, C. (2003). Does antitrust policy improve consumer welfare? Assessing the evidence. The Journal of Economic Perspectives, 17(4), 3–26. Department of Justice and the Federal Trade Commission. (1992). Horizontal Merger Guidelines [Inclusive of April 8, 1997 Revisions]. Dreazen, Y. (2002, July 15). FCC, faced with telecom crisis, could let a Bell buy WorldCom. The Wall Street Journal, A1. Dreazen, Y., Ip, G., & Kulish, N. (2002, February 25). Oligopolies are on the rise as the urge to merge grows. The Wall Street Journal, A1. Economides, N. (2005, January). The incentive for vertical integration. Working Paper #05-01. Stern School of Business, New York University. Farrell, J., & Shapiro, C. (1990). Horizontal mergers: An equilibrium analysis. The American Economic Review, 80(1), 107–126. Federal Communications Commission. (2004a, September 28). Annual report and analysis of competitive market conditions with respect to commercial mobile services, ninth report. WT Docket No. 04-111. Federal Communications Commission. (2004b, October 26). In the matter of applications of AT&T Wireless Services, Inc. and Cingular Wireless Corporation for consent to transfer control of licenses and authorizations. WT Docket No. 04-70. Grimm, C., & Winston, C. (2000). Competition in the deregulated railroad industry: Sources, effects, and policy issues. In: S. Peltzman & C. Winston (Eds), Deregulation of network industries (pp. 357–396). Washington, DC: AEI-Brookings Joint Center For Regulatory Studies.
256
DENNIS L. WEISMAN
Huber, P., Kellogg, W., & Thorne, J. (1999). Federal telecommunications law. Frederick, MD: Aspen Law and Business. Kaplan, P. (2002, February 4). U.S. may ok more telecom mergers – official. Reuters. Available at http://biz.yahoo.com/rb/030204/telecoms_antitrust_1.html Lande, R. (1982). Wealth transfers as the original and primary concern of antitrust: The efficiency interpretation challenged. Hastings Law Journal, 34(September), 65–131. Larson, A., Lehman, D., & Weisman, D. (1990). A general theory of point-to-point longdistance demand. In: A. de Fontenay, M. Shugard & D. Sibley (Eds), Telecommunications demand modelling (pp. 299–318). Amsterdam: North-Holland Publishing Co. Latour, A., Drucker, J., Sidel, R., & Raghavan, A. (2004, February 17). Cingular deal will reshape industry. The Wall Street Journal (On-Line Edition). Lehman, D., & Weisman, D. (2000). The Telecommunications Act of 1996: The ‘‘costs’’ of managed competition. Boston, MA: Kluwer Academic Publishers. Lerner, A. (1934). The concept of monopoly and the measurement of monopoly power. The Review of Economic Studies, 1(3), 157–175. Liebowitz, S., & Margolis, S. (2002). Network effects. In: M. Cave, S. Majumdar & I. Vogelsang (Eds), Handbook of telecommunications economics (pp. 75–96). Amsterdam: North-Holland. Martin, S. (1993). Advanced industrial economics. Cambridge, MA: Blackwell Publishers. Morrison, S., & Winston, C. (2000). The remaining role for government policy in the deregulated airline industry. In: S. Peltzman & C. Winston (Eds), Deregulation of network industries (pp. 1–40). Washington, DC: AEI-Brookings Joint Center For Regulatory Studies. Neuchterlein, J., & Weiser, P. (2005). Digital Crossroads, American telecommunications policy in the Internet age. Cambridge, MA: The MIT Press. Notebaert, R. (2005, May 16). Two telecom behemoths equal one duopoly. The Wall Street Journal, A17. Park, J., Babcock, M., Lemke, K., & Weisman, D. (2001). Simulating the effects of railroad mergers. Southern Economic Journal, 67(4), 938–953. Posner, R. (2001). Antitrust in the new economy. Antitrust Law Journal, 68, 925–943. Sappington, D., & Weisman, D. (1996). Designing Incentive Regulation For The Telecommunications Industry. Washington, D.C.: American Enterprise Institute; Cambridge, MA: The MIT Press. Schmalensee, R. (1988). Industrial economics: An overview. The Economic Journal, 98(392), 643–681. Shelanski, H., & Sidak, J. (2001). Antitrust divestiture in network industries. The University of Chicago Law Review, 68(1), 1–99. Surface Transportation Board Decision. (2001, June 11). SBC Ex Parte No. 582 (Sub-No. 1), Major rail consolidation procedures. Surface Transportation Board. (2002). Office of Economics, Environmental Analysis, and Administration, Statistics of Class I Freight Railroads in the United States for the Year Ended December 31, 2002. Taylor, L. (1994). Telecommunications demand in theory and practice. Boston, MA: Kluwer. Telecommunications Reports. (1997, June 23). Hundt AT&T SBC merger unthinkable. Temin, P. (1987). The fall of the Bell System. Cambridge, MA: Cambridge University Press. The Wall Street Journal Online. (2004a, October 26). Cingular closes merger with AT&T Wireless.
Market Concentration, Multi-Market Participation and Antitrust
257
Thorelli, H. (1955). The Federal Antitrust Law. Baltimore, MD: The John Hopkins Press. Tirole, J. (1988). The theory of industrial organization. Cambridge, MA: The MIT Press. Vives, X. (1999). Oligopoly pricing. Cambridge, MA: The MIT Press. Ware, H. (1998). Competition and diversification trends in telecommunications: Regulatory, technological and market pressures. Journal of Regulatory Economics, 13(1), 59–94. Weisman, D. (1999). Footprints in cyberspace: Toward a theory of mergers in network industries. Info, 1(4), 305–308. Weisman, D. (2003). A generalized pricing rule for multi-market Cournot oligopoly. Economics Letters, 81(1), 95–99. Weisman, D. (2005). Market concentration, multi-market participation and mergers in network industries. Review of Network Economics, 4(2), 129–141. White, L. (2002). Trends in aggregate concentration in the United States. Journal of Economic Perspectives, 16(4), 137–160. Williamson, O. (1968). Economies as an antitrust defense: The welfare trade-offs. The American Economic Review, 58(March), 18–34. Zimmerman, P. R. (2005). The Cingular/AT&T Wireless merger, wireline-affiliated wireless carriers, and intermodal competition in telecommunications. Working Paper.
This page intentionally left blank
