Who Do We Trust for Antitrust? Deconstructing Structural IO

World Applied Sciences Journal 22 (9): 1367-1372, 2013 ISSN 1818-4952 © IDOSI Publications, 2013 DOI: 10.5829/idosi.wasj.2013.22.09.492 Who Do We Trust for Antitrust? Deconstructing Structural IO Anand N. Asthana Centrum Católica Graduate Business School, Pontificia Universidad Católica Del Perú Abstract: The annual volume of corporate mergers and acquisitions in the world exceeds two trillion dollars. The high level of merger and acquisition activity over the past quarter of a century has revitalised the field of Industrial Organisation (IO). For policy formulation, the antitrust authorities are increasingly relying on research in this field to understand the determinants of firm and market organisation and behaviour. Yet there has been little evaluation of whether or not the mergers that have been permitted are anti-competitive. The use of ever more complex models has made IO a high-tech highbrow area of research, but the theoretical adequacy, empirical validity and policy effectiveness are yet to be established. This paper looks at the research models and methodologies used in contemporary IO research and concludes that to lend credibility to IO studies a more robust analytical framework is needed. Further, focusing on the elegance of the solution is leading researchers towards less-important questions; while there is an unfulfilled need to look at areas which are probably more relevant from the point of view of likelihood of collusive behaviour. Key words: Antitrust mergers Acquisitions Industrial organisation INTRODUCTION What is the raison d'être of a firm and what determines its scope? These remain some of the central questions for business scholars, executives and corporate strategists. The annual worldwide volume of corporate mergers and acquisitions exceeds two trillion dollars. Undoubtedly, some of it was on account of hubris displayed by the ‘leaders’ of business. Asymmetric managerial incentives motivate some mergers that ultimately diminish shareholder value [1]. There is also evidence that many CEO’s pursue acquisitions to enhance their prestige and status [2]. But it is unlikely that so much time, effort and investment bankers’ fees would be spent on adjusting firm boundaries unless there was some underlying economic gain. These gains could be in terms of operational and managerial efficiency which in turn may increase consumer surplus. However, analysis of many high-profile mergers like the acquisition of Time Warner by America Online during the merger wave of late 1990’s reveals the role of ‘spin’ and the stories about merger synergies turned out to be fiction made up by the business leaders and spin doctors to provide rationale for mergers that were actually driven by market overvaluations [3]. Some mergers are aimed at increasing Corresponding Author: market power which could result in monopoly or oligopoly. In most countries mergers are allowed to proceed only with the permission of an antitrust authority which is expected to allow mergers that do not result in creation of excessive market power. High level of merger and acquisition activity over the past quarter of century has revitalised the field of Industrial Organisation (IO) which is concerned with determinants of firm and market organisation and behaviour. In the seventies, this field was preoccupied with analysis across industries. Lack of new theoretical insights and inability to find data to solve pressing problems of the day hampered its progress and it was becoming apparent that the field was losing its way [4: xv]. Pre-1980 IO literature is read these days mainly as a historical curiosity. This literature had been “so nontheoretical, or even antitheoretical, that few economic theorists were attracted to it.” [5: 1]. In the eighties its research agenda moved toward analysing individual industries and boundaries of the firm. “Market structure” became an unfashionable term in IO and the general Structure-Conduct-Performance (SCP) paradigm that made links between structure and performance was buried. Questions about the overall organisation of production in the economy were ceded to other fields of economics like Anand N. Asthana, Centrum Católica Graduate Business School, Pontificia Universidad Católica Del Perú. 1367 World Appl. Sci. J., 22 (9): 1367-1372, 2013 trade and macroeconomics. Application of game theory and better availability and use of data gave respectability to IO. Oliver Williamson proclaimed IO as “the queen of microeconomics” and insisted, “Antitrust enforcement has been and will continue to be its main beneficiary.” [6: 306]. MATERIALS AND METHODS By and large, the law makers and antitrust authorities across the world seem to be satisfied by the work they are doing. For instance, the latest report of the US Antitrust Modernization Commission while recommending more retrospective analysis of government merger enforcement, judges the state of the U.S. antitrust laws as ‘sound’ and opines that U.S. antitrust enforcement has achieved an appropriate focus on (1) fostering innovation; (2) promoting competition and consumer welfare; and (3) aggressively punishing criminal cartel activity [7: 4]. The economists are less sanguine. Significant public and private resources are devoted to the review of the potential anticompetitive effects of mergers before they are approved. Yet there has been little evaluation of whether or not the mergers that have been permitted are anticompetitive. Without this information analysis of government policies is hardly possible [8]. Crandall and Winston [9], for example, argue that antitrust policy has not been favourable to the consumers, while in the same issue of Journal of Economic Perspectives Baker [10] argues to the contrary. If IO is to inform antitrust policy and practice, it should have its main focus on the effect of past corporate mergers on prices. This does not seem to be the case. While in the field of labour economics, one can find hundreds of empirical studies on how wages are affected by unionisation, minimum wage laws etc., there are only a few dozen empirical studies in IO with direct evaluation of the price effects of consummated mergers. Research on the aggregate effects of merger policy is even more limited [11]. The basic approach of the econometric industry studies has been christened as ‘new empirical industrial organisation’ (NEIO). The methodology of initial studies under this approach lacked sophistication. Behavioural interpretations were assigned to ‘conjectural variations’ which in turn was used as a measure of market power [13]. To avoid estimation of numerous cross-elasticities in these studies, strong restrictions on demand function were imposed. Endogeneity of prices and quantities and other identification problems were not taken care of. During the late 1990’s better techniques have developed under the brand of ‘structural IO’ [14; 15]. Demand system is usually estimated using discreet choice models of product differentiation [16]. Nested demand structures that impose restrictions on substitution effects between brands in different segments have been developed and demand modelling has centred on the trade-off between allowing flexible substitution patterns and the lack of variation in typical data that allows such substitution patterns to be flexibly identified. Demand elasticities are identified using instrumental variables like prices in other markets. Next, a model of market conduct is formulated using the substitution matrix. This enables simulation of industry behaviour with and without merger. These methods have finally removed low-brow low-tech stigma from IO but their credibility is still to be demonstrated. Mergers in the ready-to-eat-cereal industry is a case in point as it could affect the price of a product consumed by millions of households as a breakfast item for the last hundred years. Cereal is one of the most recession-resistant meals because it is fairly cheap and easy to prepare. The cereal industry has been one of the most prodigious in new brand introduction. The products are differentiated or closely related but not identical [17]. Moreover there is differing amounts of similarity across cereal brands. For example, Cheerios is closer to some brands of multigrain oats than to, say, brands corn flakes. One strategy is to divide products into segments and estimate a model that restricts substitution patterns across segments but allows flexibility within segments. As per new methodologies developed under NEIO and structural IO [18, 19], ‘front-end’ estimation of the structural parameters calculates demand functions and supply relations and these estimates are used to simulate post-merger equilibrium in the ‘back-end’ analysis. Aviv Nevo has tried to measure market power and implications of mergers in ready-to-eat cereal industry in his Ph.D. dissertation [20] and publications in esteemed journals including Rand Journal of Economics [21] and Econometrica [22] which show a deep knowledge of the industry and painstaking empirical work. Assumptions made are of some concern. The demand system formulated imposes restrictions on substitution patterns which are unconvincing. Instrumental variables are almost always difficult to find and in this case, prices in other markets used as instrumental variables could be arbitrary as it is based on independence assumption across markets. It has been assumed that the mergers affect prices through a single channel, i.e., the reduction in the number of competitors. This is difficult to believe because 1368 World Appl. Sci. J., 22 (9): 1367-1372, 2013 other factors like cost reductions can also affect prices. Similar problems beset structural models of mergers in airline industry. Supply-side effects, such as changes in marginal costs or deviations from the assumed model of firm conduct are difficult to incorporate in the model designed to measure the effect of the change in ownership on unilateral pricing incentives. Craig Peters looks at the airline industry consolidation of the 1980s [23]. He finds that structural analyses of these mergers yield poor predictions of the post-merger ticket prices and concludes that future research should aim at incorporating more flexible models of firm conduct into the methodology. An alternative could be use of difference-indifference (DD) approach by considering the counterfactual, i.e., a situation if merger had not occurred. This approach is being increasingly used in the economics profession where researchers attempt to find a naturally occurring comparison group that can mimic the properties of a control group in a physical experiment. Obviously, the assignment between the treatment group and the comparison group cannot be randomised; but it can be assumed to be ‘as if random’ [24]. Using this approach Ashenfelter and Hosken analyse merger in cereal industry and state: “It is unclear why Nevo's predictions are so different from our estimates” [8, 450]. This approach has been applied by Hastings on a panel of station specific prices to examine the price effects of acquisition of a gasoline retailer, Thrifty by ARCO (Atlantic Richfield Company) [25]. The research design includes station-level fixed effects as well as city-time effects. Whereas Nevo’s framework is a complex set of equations wherein it is difficult to see what is driving the result, DD results come from a straightforward equation revealing the average change caused by the treatment. Hasting’s equation (slightly simplified) to determine the variable of interest price p at station i at time t is: pit = µ + i + t + zit + it (1) where µ is constant and i is time-invariant stationspecific deviation from it. is city dummy. zit is an indicator of competition with independent station. It could either be a dummy variable indicating whether or not the station competes with independent stations or it could be an integer indicating the number of independent station with whom the station i competes. it is the error term. The coefficient of interest is which will indicate whether presence or absence of an independent competitor has a significant effect on the local retail price. One weakness of this analysis is that it captures the effects of a merger on Thrifty’s competitors, but not on the former Thrifty stations. Even so the randomistas (a growing tribe among econometricians) like Angrist and Pischke have showered praise on this type of analysis as a fruitful change in direction [11]. However, such endorsement could be premature. The size of the estimated effect was five cents per gallon which means that retail margins went up by as much as 50 percent. This would be an eye-opener for antitrust authorities. However, six years after the publication of Hasting’s paper in the American Economic Review, came another paper in the same journal wherein the authors revisited Hasting’s analysis. The authors Taylor, Kreisle and Zimmerman [26] used almost the same dataset (of higher frequency and for a longer time) and present their results Table 2 presents the result of estimating the following regression: pit = µ + i + Conversionit + j k jk i t + it (2) where pi,t, µ, i and it have the same meanings as in equation (1). The dummy variable Conversionit takes a value of one if station i is located within a mile of a Thrifty station during period t (i.e., “competed” with an independent Thrifty outlet prior to its conversion). Thus, a negative estimate of the coefficient implies that the transaction (the loss of an independent competitor) is correlated with an increase in the average price at these competing stations. The city-time fixed effects are captured by the interaction of city dummies, i and time dummies t. Equation (2) may appear to be outwardly different, but is virtually identical to equation (1). The coefficient estimates of control variables (e.g., city-time interactions) are quite similar to those obtained by Hastings. But the coefficient estimates of the variable of main interest, i.e., Conversion are vastly different. In summary, the price increase was found to be of little economic significance - one fiftieth of that found by Hastings. This finding is robust to using various sub-samples and the authors doubt whether ARCO’s acquisition Thrifty led to higher prices. Moreover, while Hastings finds support for the underlying model of consumer preference - differentiated products with consumer brand loyalty – Taylor et al. doubt whether this model depicts consumer behaviour dispute the underlying model of consumer preferences. With time DD methodology is becoming more and more sophisticated; but is unable to shrug the criticism that it is atheoretical and highly sensitive to assumptions. 1369 World Appl. Sci. J., 22 (9): 1367-1372, 2013 Question of Trust: If we can trust neither the structurally derived estimates nor direct DD estimates, where do we go from here? Structural models are likely to become more sophisticated with time. Whether they will be able to rest on fewer and plausible assumptions remains an open question. Randomistas believe that a few structural models should be tried out and whichever fits the direct estimates better should be declared the winner. In this vein Hausman and Leonard [27] use three structural models in their study relating to new brands of toilet paper and find that the Nash-Bertrand model which is frequently employed in studies of the competitive effects of mergers yield indirect estimates reasonably similar to the direct estimates and superior to the indirect estimates produced by the two alternative models they tried. This begs the question: are the direct estimates the gold standard? As seen in the case study of acquisition of Thrifty gas stations by ARCO, DD estimates cannot be relied upon. On theoretical grounds many eminent scholars have decried emphasis on experimental or quasi-experimental results. For example Nobel Laureate James Heckman points out that the retreat to statistics has abandoned economic choice theory. Important distinctions about ex ante and ex post outcomes and subjective and objective evaluations that are central to structural econometrics are forgotten [28]. Finally, even if we are able to draw some plausible conclusions from past mergers, how relevant are these estimates to future mergers? Liran Einav and Jonathan Levin ask whether the votaries of direct measurement seriously think that while reviewing a merger of Microsoft and Yahoo! the Department of Justice should rely on the price effect of past airline or office supply company mergers or the subsets obtained from chance meetings of CEO’s or from lunar eclipses [29: 159]. Merger analysis presents peculiar problems in case of multinationals. While it is not difficult to take stock prices and balance sheets into account, the cultural context is difficult to grasp. For example, there are significant differences between human resource management practices of western multinationals and Asian multinationals operating outside the countries of their national origin [30]. As far as between-industry differences are concerned, the US Antitrust Modernization Commission reported in 2007 that it does not believe that new or different rules are needed to address so-called “new economy” and insisted that the antitrust laws remain relevant in today’s environment and tomorrow’s as well [7: 4].. Further the Commission submitted that differential treatment to different industries. In the current state of IO research there is insufficient emphasis on applied work on measurement based data that that continues by framing the empirical exercise in terms of a coherent economic model. CONCLUSION Economic theory has not had much to say about exactly how organisation should matter. On the other hand, non-economists in business schools generally think that organization matters and that firms are not, regardless of what economic theory may suppose, undifferentiated profit maximizing agencies which react to given market situations in ways which are independent of their organisation. The most important area of public action related to market structure that IO economists have sought to inform has been the merger policy. U.S. policy toward horizontal mergers is enormously more sophisticated now than it was when the first Guidelines were issued in 1968 [31]. This is a result of the influence of IO scholars as compared to that of lawyers and jurists [32]. Market shares are no longer the main structural indicator considered. Unilateral effects, the performance impacts of changing structure assuming no change in the nature or intensity of competition, are considered more important. The Guidelines issued in 2010 outline a much more complex analytical framework based on advances in theory and enforcement experience but not on empirical findings. For reasons not made clear, coordinated effects - adverse changes in (expected) market performance that occur because changes in market structure make collusive behaviour more likely – have taken a back seat. According to Richard Schmalensee it could be so because the tools available to analyse unilateral effects have become much more powerful [33]. Merger simulation models formulated by Budzinski and Ruhmer [34] can be employed to integrate information from a variety of sources and the newly introduced Upward Pricing Pressure (UPP) test is an improvement over the traditional market definition approach in case of differentiated products. But these new tools shed no light whatever on coordinated effects. Merger simulation models usually assume single-period Bertrand competition and the UPP test assumes that the demand curves facing the merging firms do not change as a consequence of their merger or their post-merger price changes [35]. Antitrust policy is incorporating the areas where heavy artillery of econometrics has cleared the way. The areas which are probably more relevant from the point of view of likelihood of collusive behaviour as revealed through coordinated effects lie largely unexplored. 1370 World Appl. Sci. J., 22 (9): 1367-1372, 2013 REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. Morck, R., A. Shleifer and R.W. Vishny, 1990. Do Managerial Objectives Drive Bad Acquisitions? The Journal of Finance, 45(1): 31-48. Avery, C., J.A. Chevalier and S. Schaefer, 1988. Why Do Managers Undertake Acquisitions? An Analysis of Internal and External Rewards for Acquisitiveness. Journal of Law, Economics and Organization, 14(1): 24-43. Shleifer, A. and R.W. Vishny, 2003. Stock market driven acquisitions. Journal of Financial Economics, 70: 295-311. Fuchs, V.R., 1972. Policy Issues and Research Opportunities in Industrial Organization. NBER Stigler, G.J., 1968. The Organization of Industry. Irwin. Williamson, O.E., 1996. The Mechanisms of Governance. Oxford University Press. Antitrust Modernization Commission, 2007. Report and Recommendations. AMC. Ashenfelter, O. and D. Hosken, 2010. The Effect of Mergers on Consumer Prices: Evidence from Five Mergers on the Enforcement Margin. Journal of Law and Economics, pp: 417-466. Crandall, R.W. and C. Winston, 2003. Does Antitrust Policy Improve Consumer Welfare? Assessing the Evidence. The Journal of Economics Perspectives, 17(4): 3-26. Baker, J.B., 2003. The Case for Antitrust Enforcement. Journal of Economic Perspectives, 17(4): 27-50. Angrist, J.D. and J.S. Pischke, 2010. The Credibility Revolution in Empirical Economics: How Better Research Design Is Taking the Con Out of Econometrics. Journal of Economic Perspectives, 24(2): 3-30. Bresahan, T.F., 1989. Empirical Studies of Industries with Market Power. In Schmalensee, R. & R. D. Willig (Eds.), Handbook of Industrial Organization, Volume 2 (pp: 1011-1057). North-Holland. Corts, K.S., 1999. Conduct Parameters and the Measurements of Market Power. Journal of Econometrics, 88(2): 227-250. Ackerberg, D., L. Benkard, S. Berry and A. Pakes, 2007. Econometric Tools for Analyzing Market Outcomes. In Heckman, J.J. and E.E. Leamer (Eds.), Handbook of Econometrics, Volume 6, Part A (pp: 4171-4276). Elsevier. 15. Reiss, P.C. and F.A. Wolak, 2007. Structural Econometric Modeling: Rationales and Examples from Industrial Organization. In Heckman, J.J. and E.E. Leamer (Eds.), Handbook of Econometrics, Volume 6, Part A (pp: 4227-4415). Elsevier. 16. Berry, S.T., 1994. Estimating Discrete Choice Models of Product Differentiation. Rand Journal of Economics, 25(2): 242-262. 17. Hausman, J.A., 1997. Valuation of New Goods under Perfect and Imperfect Competition. In Bresnahan, T.F. and R.J. Gordon, (Eds.), The Economics of New Goods: 209-237. Chicago: University of Chicago Press. 18. Baker, J. and T.F. Bresnahan, 1985. The Gains from Merger or Collusion in Product Differentiated Industries. Journal of Industrial Economics, 33(4): 427-444. 19. Berry, S.T. and A. Pakes, 1993. Some Applications and Limitations of recent advances in Empirical Industrial Organization: Merger Analysis. American Economic Review, 83: 247-252. 20. Nevo, A., 1997. Demand for Ready-to-Eat Cereal and its Implications for Price Competition, Merger Analysis and Valuation of New Brands. Ph.D. Dissertation. Harvard University 21. Nevo, A., 2000. Mergers with Differentiated Products: The Case of the Ready-to-Eat Cereal. Rand Journal of Economics, 31(3): 395-421. 22. Nevo, A., 2001. Mergers with Differentiated Products: The Case of the Ready-to-Eat Cereal. Econometrica, 69(2): 307-342. 23. Peters, C., 2010. Evaluating the performance of Merger Simulation: Evidence from the US Airline Industry. Journal of Law and Economics, 49(2): 627-49. 24. Asthana, A.N., S. Mohan and S. Khorana, 2012. Physics envy and natural experiments in business and economics. World Applied Sciences Journal, 20(3): 464-469. 25. Hastings Justine, S., 2004. Vertical Relationships and Competition in Retail Gasoline Markets: Empirical Evidence from Contract Changes in Southern California. American Economic Review, 94(1): 317-28. 26. Taylor, C., N. Kreisle and P. Zimmerman, 2010. Vertical Relationships and Competition in Retail Gasoline Markets: Empirical Evidence from Contract Changes in Southern California: Comment. American Economic Review, 100(3): 1269-1276. 1371 World Appl. Sci. J., 22 (9): 1367-1372, 2013 27. Hausman, J.A. and G.K. Leonard, 2002. The Competitive Effects of a New Product Introduction: A Case Study. Journal of Industrial Economics, 50(3): 237-63. 28. Heckman, J.J., 2010. Building Bridges between Structural and Program Evaluation Approaches to Evaluating Policy. Journal of Economic Literature, 48(2): 356-398. 29. Einav, L. and J. Levin, 2010. The Credibility Revolution in Empirical Economics: Empirical Industrial Organisation: A Progress Report. Journal of Economic Perspectives, 24(2): 145-162. 30. Nigam, R. and S. Zhan, 2011. Management in Emerging versus Developed Countries: A Comparative Study from an Indian Perspective. Journal of Centrum Cathedra, 4(1): 121-133. 31. Shapiro, C., 2010. The 2010 Horizontal Merger Guidelines: From Hedgehog to Fox in Forty Years. Antitrust Law Journal, 77(1): 701-759. 32. White, L.J., 2010. The Growing Influence of Economics and Economists on Antitrust: An Extended Discussion. Economics, Management and Financial Markets, 5(1): 26-63. 33. Schmalensee, R., 2012. On a Level with Dentists? Reflections on the Evolution of Industrial Organization. Review of Industrial Organizations, 41(3): 157-179. 34. Budzinski, O. and J. Ruhmer, 2010. Merger Simulation in Competition Policy: A Survey. Journal of Competition Law and Economics, 6(2): 277-319. 35. Jaffe, S. and E.G. Weyl, 2013 The First-Order Approach to Merger Analysis. American Economic Journal: Microeconomics, (forthcoming). 1372