3_2019.3_ARON (DO NOT DELETE) 12/25/2019 11:20 PM AN ECONOMIC PERSPECTIVE ON BALANCING UNQUANTIFIED HARMS AND BENEFITS UNDER THE CONSUMER WELFARE STANDARD Debra Aron and Steven Tenn* I. Introduction ............................................................... 849 II. The Role of Qualitative Evidence .............................. 851 A. Challenges in Assessing Qualitative Evidence ... 852 B. Balancing Pro- and Anticompetitive Effects ....... 853 III. Price and Non-Price Effects of Alleged Anticompetitive Conduct ...................................................................... 854 A. Price Effects ......................................................... 855 B. Non-Price Effects Related to Consumer Demand for the Products at Issue ..................................... 855 C. Effects Related to Consumer Demand for Products Other Than Those Directly at Issue .................... 862 IV. Conclusion .................................................................. 866 I. INTRODUCTION Judge Ginsburg’s address on how to balance unquantified benefits and harms under the consumer welfare standard highlights two distinct roles for qualitative assessment. The first is the role of qualitative evidence in antitrust analyses. The second is the consideration of both price and non-price effects. While these two topics are related—e.g., analysis of * Vice Presidents, Charles River Associates. The views and opinions expressed herein are those of the authors and do not reflect or represent the views of Charles River Associates, their clients, or any organizations with which the authors are affiliated. A version of this paper was delivered as the William Howard Taft Lecture, September 27, 2019, to the New York State Bar Association, Antitrust Law Section. 3_2019.3_ARON (DO NOT DELTE) 12/25/2019 11:20 PM 850 COLUMBIA BUSINESS LAW REVIEW [Vol. 2019 non-price competitive effects often relies on qualitative evidence—they raise distinct issues. We provide an economic perspective on these topics as a complement to Judge Ginsburg’s learned legal commentary. Specifically, we explain the central role that sound qualitative evidence traditionally and properly plays in the economic analysis of both price and non-price effects of mergers and accused conduct. We also discuss the state of economic analysis when a broad economic interpretation of the consumer welfare standard would require balancing the welfare of one group against another. We see Judge Ginsburg’s thesis as captured in the following passage of his remarks: Meeting the challenge of balancing qualitative harms or benefits in an antitrust case is not a reason to depart from the consumer welfare standard. On the contrary, this balancing exercise requires an evidence- based approach to incorporating into the consumer welfare standard those qualitative effects—both anticompetitive harms and procompetitive benefits— of a restraint of trade.1 Judge Ginsburg further elaborates on the evidentiary demands of this approach as follows: For a claimed competitive harm or benefit that remains unquantified, the decisionmaker should demand both a convincing showing that the magnitude of the effect cannot be quantified and a sound theoretical basis for nonetheless believing the effect will be real and substantial.2 We agree entirely with these principles and will offer some comments on how economists operationalize the challenges of assessing competitive effects using qualitative as well as, if available, quantitative evidence. However, there is one point on which we will argue that Judge Ginsburg perhaps unduly limits the role of economic 1 Douglas H. Ginsburg, Balancing Unquantified Harms and Benefits in Antitrust Cases Under the Consumer Welfare Standard, 2019 COLUM. BUS. L. REV. 824, 845 (2019). 2 Id. 3_2019.3_ARON (DO NOT DELETE) 12/25/2019 11:20 PM No. 3:849] AN ECONOMIC PERSPECTIVE ON BALANCING 851 analysis and the breadth of the consumer welfare standard. Judge Ginsburg offers: There have been calls for antitrust enforcers and courts to consider all manner of other criteria in addition to the welfare of consumers. Doing so would require making complex tradeoffs among incommensurable goals, without any principled way to do so. The effect would be to place unbridled and unreviewable discretion in the hands of enforcers and courts, inviting arbitrary decisions and creating uncertainty for firms. Accepting nebulous theories of harm (such as the “accumulation of political power”), and by parity of reasoning, vague “public interest” defenses (such as a claim that an agreement among rivals would reduce economic inequality) would lead to inconsistent results and evade the rule of law.3 As we will discuss in these remarks, positing that public interest considerations are separate or excluded from consumer welfare effects of a merger or accused conduct is difficult to defend from an economic perspective. Specifically, public interest considerations and other potentially nebulous factors also affect consumer welfare broadly conceived. Therefore, they cannot, as a matter of principle, be considered separate from economists’ general notion of consumer welfare. II. THE ROLE OF QUALITATIVE EVIDENCE The role of qualitative evidence in antitrust analysis may seem like an odd topic for economists. We are more often associated with econometric analyses and complex theoretical models than with the assessment of qualitative evidence. While antitrust economists do frequently rely on highly technical analyses, these are a means towards a more fundamental purpose: to understand the incentives faced by firms, their customers, and consumers, given the costs and benefits of their choices. To achieve that end, economic analysis does not exclusively, or even primarily, rely on quantitative evidence. 3 Id. at 826–27 (footnotes omitted). 3_2019.3_ARON (DO NOT DELTE) 12/25/2019 11:20 PM 852 COLUMBIA BUSINESS LAW REVIEW [Vol. 2019 Whether an analysis in a particular case best relies on qualitative evidence or a combination of qualitative and quantitative evidence depends on the facts of the case. Indeed, while one could, in principle, undertake a robust antitrust analysis based purely on qualitative evidence, it is almost inconceivable that one could do so based purely on quantitative evidence. The reason is simple: economic data analyses are built on models that try to capture the salient features of the factual circumstances, but with sufficient simplifications to make the analysis tractable. As a result, qualitative evidence is not only a complement to quantitative analysis, it is also the foundation upon which empirical analyses are built. One cannot begin to identify and formulate the proper quantitative analysis using economic methods without first performing a qualitative analysis of the available evidence regarding which products compete with each other, how prices are determined in the market, who the suppliers are, and other relevant facts governing the marketplace. A. Challenges in Assessing Qualitative Evidence Assessments of qualitative evidence raise particular challenges that do not arise in quantitative analyses. For example, the economics literature has developed a variety of techniques to estimate values of interest, and to quantify the precision of those empirical estimates through concepts such as standard errors and confidence intervals.4 For qualitative evidence, there is no direct equivalent for assessing precision. Instead, one must rely on a range of techniques to infer the informative value of qualitative evidence. When assessing qualitative evidence, it is important to consider how, why, and when that piece of evidence was originally created. For example, is it likely that the author of a given document was in a position, and had the requisite experience, to reliably assess the topic at hand? Often, a key issue is whether the author of a document had an incentive to describe the situation accurately. Moreover, even if an 4 See generally WILLIAM H. GREENE, ECONOMETRIC ANALYSIS 51–56 (6th ed. 2008) (explaining standard errors and confidence intervals). 3_2019.3_ARON (DO NOT DELETE) 12/25/2019 11:20 PM No. 3:849] AN ECONOMIC PERSPECTIVE ON BALANCING 853 assessment was accurate when made, it is important to consider whether the competitive situation has since changed or is likely to change such that the original assessment may not apply in the future. Especially in rapidly changing industries, the relevance of particular competitive factors may change over time. In many cases, some qualitative evidence will point in one direction on a given issue, while other evidence will suggest otherwise. This does not mean that qualitative evidence is not useful or should not be relied upon. Instead, one should try to assess a range of evidence to determine what is most likely to be true. By reviewing a wide range of qualitative evidence, individually imprecise elements can be aggregated, albeit informally rather than numerically, resulting in an informed opinion on a given issue. This is one reason why it is valuable for the economic expert to contribute early to the discovery process. Although rigorous and relevant quantitative analysis is the gold standard in some circumstances where it is feasible to perform, there are circumstances in which an insistence on quantitative evidence is counterproductive. Quantitative analysis based on inadequate or inappropriate data may have little informative value. Nevertheless, it may create the impression of precision that it does not deserve, and attract weight that its merit does not bear. An absolute requirement that all effects be quantified would likely result in bad analysis trumping no analysis, or in dueling bad analyses, in both cases to the detriment of sound decisionmaking. When reliable quantitative analysis is not feasible, sound decisionmaking can be based on qualitative evidence about the existence and likely magnitude of qualitative effects if that evidence is sufficiently rigorous, robust, and relevant under the theory being assessed. B. Balancing Pro- and Anticompetitive Effects Weighing whether the magnitude of procompetitive effects is likely to exceed the magnitude of anticompetitive effects can be challenging even when such effects are quantifiable. Of 3_2019.3_ARON (DO NOT DELTE) 12/25/2019 11:20 PM 854 COLUMBIA BUSINESS LAW REVIEW [Vol. 2019 course, this comparison is all the more difficult when pro- and anticompetitive effects cannot be quantified. One potential approach in this situation is to assess whether pro- or anticompetitive effects are likely to be significant. Even in instances where the magnitudes of such effects are not measurable, one may still be able to infer whether they are likely significant based on the extent to which such issues are discussed in the ordinary course of business, or from strategic actions taken in anticipation of such effects, for example. In some cases, only anticompetitive effects are likely significant, with procompetitive effects not likely significant, or vice versa. In this situation, one can reasonably conclude that either the pro- or anticompetitive effects of a given conduct are likely to dominate even if those effects cannot readily be quantified. The situation is more complicated, of course, when both pro- and anticompetitive effects are likely significant. In such circumstances, economists often rely on indirect methods for determining which effect is likely to have the predominant effect on consumer welfare. One option is to consider evidence gathered from prior similar examples of alleged anticompetitive conduct. For example, in the merger context one might consider the effect of prior acquisitions made by the same firms or by other firms in the same industry.5 While differences between acquisitions can mitigate the validity of such comparisons, and one must take care to consider key factors that may lead to significantly different effects for different transactions, this approach can still be informative. III. PRICE AND NON-PRICE EFFECTS OF ALLEGED ANTICOMPETITIVE CONDUCT Alleged anticompetitive conduct may result in three types of consequences. The first consists of price effects for the set 5 For an example of this approach, see Deborah Haas-Wilson & Christopher Garmon, Hospital Mergers and Competitive Effects: Two Retrospective Analyses, 18 INT’L J. ECON. BUS. 17 (2011) (discussing two retrospective studies of hospital mergers). 3_2019.3_ARON (DO NOT DELETE) 12/25/2019 11:20 PM No. 3:849] AN ECONOMIC PERSPECTIVE ON BALANCING 855 of products at issue. The second consists of non-price effects, such as product quality, that impact consumer demand for the set of products at issue. The third type of effect consists of price and non-price effects related to consumer demand outside of the set of products at issue. A comparison of these three types of effects illustrates how the economics literature has dealt with some of the issues highlighted in Judge Ginsburg’s address. A. Price Effects Price and volume of sales are often the primary focus in antitrust analyses. It is natural to focus on such financial effects because their implications for consumer welfare are clear. That is, generally speaking, consumer welfare is reduced by higher prices or reductions in the quantity purchased of the desired products.6 B. Non-Price Effects Related to Consumer Demand for the Products at Issue The cases discussed by Judge Ginsburg implicate non-price as well as price and quantity effects. A standard approach in the economics literature is to model products as collections of characteristics.7 One characteristic is price, while other product characteristics capture what economists think of as dimensions of product quality, broadly defined.8 Economic models of consumer demand often combine price and non-price competition based on the following paradigm. Consumers have preferences over both price and non-price product characteristics, and select from the set of available products the one that provides the highest net consumer 6 See generally HAL R. VARIAN, MICROECONOMIC ANALYSIS 94–112 (3d ed. 1992) (explaining conditions under which consumer utility is non- increasing in price, holding other factors constant). 7 See generally Kevin J. Lancaster, A New Approach to Consumer Theory, 74 J. POL. ECON. 132, 133–37 (1966). 8 Id. 3_2019.3_ARON (DO NOT DELTE) 12/25/2019 11:20 PM 856 COLUMBIA BUSINESS LAW REVIEW [Vol. 2019 surplus after accounting for the price of the product.9 A key implication of this paradigm is that, for a given consumer, substitution occurs in response to non-price as well as price factors. This is because the consumer does not choose products based solely on price, or solely on non-price factors, but rather on the overall attractiveness of the bundle of price and non- price attributes.10 The effects of competition on price and non-price attributes are, therefore, intrinsically linked, but they do not always pull in the same direction. Indeed, non-price effects may have the opposite impact on consumer welfare of the price effects (or even other non-price effects); for example, if price is expected to rise but so is quality, or if product durability is expected to rise but convenience of purchase is expected to fall. In these cases, the decisionmaker is confronted with the necessity of balancing countervailing effects on consumer welfare against each other. While clearly more realistic than a price-only framework in which all other product characteristics are held fixed, this generalized framework raises a number of conceptual and practical challenges. First, allegedly anticompetitive conduct could, in principle, affect several price and non-price characteristics. Consequently, a general analysis of such conduct may present a high-dimensionality problem, given that even relatively simple products may have many characteristics. Second, it is important to recognize that price and non- price characteristics are not independently determined but rather are mutually determined. The optimal (profit maximizing) set of product characteristics will typically depend on the price the firm can charge for the product with those characteristics, and the optimal price will depend on the product’s characteristics. Often, the firm’s profit maximizing price will be higher for products with characteristics that 9 See, e.g., Steven Berry et al., Automobile Prices in Market Equilibrium, 63 ECONOMETRICA 841, 844–51 (1995). See also Aviv Nevo, A Practitioner’s Guide to Estimation of Random-Coefficients Logit Models of Demand, 9 J. ECON. MGMT. STRATEGY 513, 516–21 (2000). 10 Id. 3_2019.3_ARON (DO NOT DELETE) 12/25/2019 11:20 PM No. 3:849] AN ECONOMIC PERSPECTIVE ON BALANCING 857 consumers demand more highly (all else being equal). Consequently, one cannot assess the procompetitive impact of quality changes in isolation of price changes. While a potential defense of allegedly anticompetitive merger effects, for example, is that the merger will result in a higher quality product (or other non-price consumer benefits), the value of any such benefits must be weighed against the price effects. A higher price is bad for consumers, but higher quality is good, at least for some consumers. Whether consumers are better off on a net basis depends on whether the quality-adjusted price has gone up or down. The hospital merger case discussed by Judge Ginsburg, Saint Alphonsus Medical Center-Nampa Inc. v. St. Luke’s Health System, Ltd., raised this specific issue. While the district court disallowed the merger on the grounds that the quality-related efficiencies were not merger specific,11 the Ninth Circuit concluded that, even if the quality effects were unavailable without the merger, showing an increase in quality would not be sufficient to justify it.12 Specifically, the court found: But even if we assume that the claimed efficiencies were merger-specific, the defense would nonetheless fail. At most, the district court concluded that St. Luke’s might provide better service to patients after the merger. That is a laudable goal, but the Clayton Act does not excuse mergers that lessen competition or create monopolies simply because the merged entity can improve its operations. . . . The district court did not clearly err in concluding that whatever else St. Luke’s proved, it did not demonstrate that efficiencies resulting from the merger would have a positive effect on competition.13 11 Saint Alphonsus Med. Ctr.-Nampa, Inc. v. St. Luke’s Health Sys., Ltd., Nos. 1:12-CV-00560-BLW, 1:13-CV-00116-BLW, 2014 WL 407446, at *12 (D. Idaho Jan. 24, 2014). 12 Saint Alphonsus Med. Ctr.-Nampa Inc. v. St. Luke’s Health Sys., Ltd., 778 F.3d 775, 791–92 (9th Cir. 2015). 13 Id. (citation omitted). 3_2019.3_ARON (DO NOT DELTE) 12/25/2019 11:20 PM 858 COLUMBIA BUSINESS LAW REVIEW [Vol. 2019 The appropriate economic analysis consistent with the consumer welfare standard considers whether efficiencies increase consumer welfare enough to offset the negative effect of higher prices on consumer welfare. The justification for promoting competition is, as Judge Ginsburg said, to advance consumer welfare. Quality of service directly affects consumer welfare—indeed, in health care, quality of service is central to the determination. Hence, it is necessary, from an economic standpoint, to consider both price and quality effects when determining whether a merger would advance consumer welfare. While often difficult to put into practice, weighing price and quality effects centers on a straightforward question: are consumers better off under the post-merger combination of price and quality compared to the pre-merger combination of price and quality? When undertaking such an analysis it would be relevant to assess, if possible, whether, on a net basis, the total quantity of sales would increase or decrease as a result of the merger. If the net effect of the price and quality increases would be to increase the volume of sales, then that should create a rebuttable presumption that social welfare will go up with the consummation of the merger. If the net effect would be a reduction in volume of sales, the rebuttable presumption should be that consumer welfare will go down. It may not be feasible to predict reliably whether the net effect of changes in price and quality would be an increase or decrease in the volume of sales. In that case, it would be useful to consider several factors, which we refer to as magnitude, likelihood, and risk. Regarding magnitude, it is relevant to consider whether, when effects cannot be quantified, there is evidence of relative orders of magnitude. As discussed earlier, qualitative evidence may allow an inference regarding which effects are likely to be significantly larger than others even in cases where the precise size of those effects cannot be quantified. Regarding likelihood, one should consider how persuasive the evidence is on price increases, and how persuasive the evidence is on quality increases. Where the likelihood of an effect is relatively low, or the evidence is relatively weak, the 3_2019.3_ARON (DO NOT DELETE) 12/25/2019 11:20 PM No. 3:849] AN ECONOMIC PERSPECTIVE ON BALANCING 859 magnitude of the effect should be discounted when attempting to weigh countervailing effects. Regarding risk, it is relevant to assess the effects of type one versus type two errors; that is, whether the effect on consumer welfare is likely to be more damaging if the merger is erroneously approved versus whether it is erroneously blocked. For example, if a merger is approved even though the price increase would, in fact, outweigh the consumer benefits from a quality increase, the anticompetitive effects may be overcome by a timely market response. Any conclusion as to whether the merger would increase price should have also incorporated an analysis of entry barriers and supply responses. If the evidence for a post-merger supply response is relatively strong, the risk of erroneously approving the merger may be mitigated. In contrast, if the evidence is persuasive that post-merger supply responses are unlikely, one would be more reluctant to allow the merger to proceed because the likelihood that the market will discipline any unexpected price increase is relatively low and the risk of a poor outcome from an erroneously approved merger is correspondingly higher. Our discussion to this point has treated consumer preferences as homogeneous. In the context of this discussion, however, it is important to recognize that preferences over non-price characteristics may vary across consumers. For example, while some consumers may highly value an improvement in the orthopedic department of a healthcare group that will purportedly result from a merger, others may care only about the effect of the merger on the endocrinology or pediatric departments. In fact, since consumers may have very different preferences, a change in a product characteristic may be viewed as an increase in product quality by some consumers but a reduction in product quality by others. For example, the simplification of a software product that makes it more user-friendly may be of great value to some consumers, but reduce the desirability of the product for those consumers who used features that were hidden or eliminated by the simplification. These considerations do not arise in a 3_2019.3_ARON (DO NOT DELTE) 12/25/2019 11:20 PM 860 COLUMBIA BUSINESS LAW REVIEW [Vol. 2019 price-only model because, all else equal, consumers generally prefer lower prices over higher prices. Because consumers may have heterogeneous preferences over non-price factors, alleged anticompetitive conduct may hurt some consumers while benefitting others. In order to determine whether, on balance, the conduct in question is likely pro- or anticompetitive under a consumer-welfare standard, the decisionmaker must weigh the individual welfare of different consumers against each other in some manner. As a general proposition, economics provides no single “correct” way to define a social-welfare function that aggregates individual utilities or weighs them against one another. Indeed, in the absence of altruism, it is individually rational for each consumer to prefer a very simple social- welfare function: give 100% weight to one’s own preferences and 0% weight to the preferences of others. Of course, those who care about friends, family, and even society at large may prefer a different social-welfare function that gives less weight to their own preferences and more to those of others. Pareto efficiency is a concept in the economics literature that can be used to determine whether a given outcome is unambiguously better than another.14 Specifically, a Pareto- efficient outcome is one in which it is impossible to make at least one consumer better off without making at least one other consumer worse off. Applied to antitrust analyses, one might conclude that a given conduct is procompetitive if no consumers are worse off compared to the counterfactual outcome that would have occurred in the absence of the conduct in question, and some consumers are better off. Similarly, one might conclude that a given conduct is anticompetitive if no consumers are better off, and some are worse off, as compared to the counterfactual outcome. In a price-only model in which all consumers face the same price, this welfare standard corresponds to whether the conduct in question led to a higher or lower price, because generally speaking, consumers prefer lower prices to higher 14 See VARIAN, supra note 6, at 225, 404–09 (explaining the concept of Pareto efficiency). 3_2019.3_ARON (DO NOT DELETE) 12/25/2019 11:20 PM No. 3:849] AN ECONOMIC PERSPECTIVE ON BALANCING 861 prices, holding product attributes constant.15 The obvious problem with relying on Pareto efficiency in a more general framework in which both price and non-price factors may change is that, due to heterogeneous consumer preferences for non-price product characteristics, it may be the rare case where (literally) all consumers are either worse or better off. If welfare effects are mixed across consumers, then the concept of Pareto efficiency offers little guidance as to whether, on balance, a given conduct is pro- or anticompetitive. A potentially more useful welfare concept from the economics literature is known as “compensating variations,” which applies the following thought experiment.16 For each consumer, compare their welfare given the conduct in question to their welfare in the counterfactual outcome that would have occurred in the absence of the given conduct. Calculate the amount of money that the consumer would need to receive, given the conduct, to make them indifferent between the two outcomes. For consumers harmed by the conduct, the money transfer would be positive, while the money transfer would be negative for consumers who benefitted from the conduct. If net money transfers across all consumers are positive, then, by the “compensating variations” theory one would conclude that the conduct is anticompetitive, while if net money transfers are negative, then one would conclude that the conduct is procompetitive. The use of compensating variations for measuring consumer-welfare effects is based on allocative efficiency.17 If there were a social planner who could engage in money transfers with consumers, then it would, in theory, be possible to make all consumers better off compared to the counterfactual outcome where the conduct in question had not occurred, when (and only when) net transfers are negative. Because monetary transfers are a standard unit of measure, 15 See VARIAN, supra note 6, at 94–112. 16 See VARIAN, supra note 6, at 160–63. 17 See VARIAN, supra note 6, at 224–27. 3_2019.3_ARON (DO NOT DELTE) 12/25/2019 11:20 PM 862 COLUMBIA BUSINESS LAW REVIEW [Vol. 2019 this theoretical framework allows welfare to be compared across consumers with heterogeneous preferences. One important limitation of compensating variations is that it is a purely theoretical construct. In the real world, there is no social planner that is engaged in monetary transfers with consumers to offset the impact of a particular conduct. This means that, even if a given conduct were welfare-improving in the aggregate according to the theory of compensating variations, a significant portion of consumers might still be significantly worse off as a result of the conduct. What to do in this situation is open to debate. It may be reasonable to allow conduct that is generally procompetitive even if it results in some (limited) harm to a small group of consumers. However, depending on one’s view of the “optimal” social-welfare function, it may be harder to defend that position as the percentage of harmed consumers grows, or as the harm suffered by that group grows larger in magnitude. As noted earlier, economics is limited in its ability to offer guidance to decisionmakers when the effects of potentially anticompetitive conduct benefit some consumers while harming others. C. Effects Related to Consumer Demand for Products Other Than Those Directly at Issue From an economic perspective, effects unrelated to consumer demand for the products directly at issue raise essentially the same concerns as effects related to consumer demand for the products directly at issue. The main distinction between the two is the set of impacted consumers: effects unrelated to consumer demand for the set of products at issue will often impact a set of consumers largely distinct from those who are purchasing goods in the relevant antitrust market. When distinct groups are impacted by a given conduct, it may be the case that one group benefits while a different group is harmed. The fact that the consumers who benefit are distinct from those who are harmed does not preclude, in theory, applying the consumer-welfare standard. There are two practical challenges associated with adjudicating a case 3_2019.3_ARON (DO NOT DELETE) 12/25/2019 11:20 PM No. 3:849] AN ECONOMIC PERSPECTIVE ON BALANCING 863 in which the benefiting group is distinct from the harmed group, however. The first is the problem of the “slippery slope.” Once one opens for consideration the economic repercussions of any alleged conduct beyond its effects on the directly affected products (and directly affected consumers), those repercussions may ripple to a nearly endless potential set of effects. Naturally, not all such effects can be considered, as a practical matter—however the perfect should not be the enemy of the good. Acknowledging that there may be important positive or negative effects on consumers other than those in the directly-affected group should not open the analysis to all possible avenues of effect; but rather, as in any other antitrust analysis, evidence-based judgment must be applied to limit the analysis to the most important areas of consideration or most important ripple effects of an alleged conduct. The second practical challenge associated with adjudicating a case in which the benefiting group is distinct from the harmed group is the question of how to balance the welfare of two distinct groups of consumers. Economics has little to offer here, as elaborated above. In O’Bannon v. NCAA, for example, the court concluded that the value of amateurism to sports fans outweighed the cost to the student-athletes of being deprived of income they would otherwise presumably earn.18 That judgment, whether right or wrong, was not based on economic guidance regarding how the welfare of those two groups should be balanced. This issue of which set of consumers matters when considering welfare effects is analogous to the antitrust debate over whether “out-of-market” efficiencies should be included when evaluating the net welfare effects of a given conduct. In the merger context, the antitrust agencies generally disallow out-of-market efficiencies and instead take the view that competitive impacts should be separately 18 O’Bannon v. NCAA, 802 F.3d 1049, 1070, 1076–79 (9th Cir. 2015). 3_2019.3_ARON (DO NOT DELTE) 12/25/2019 11:20 PM 864 COLUMBIA BUSINESS LAW REVIEW [Vol. 2019 evaluated in each antitrust market.19 That is, procompetitive effects in one market cannot be used to offset anticompetitive effects in another market, although the Interagency Merger Guidelines do highlight that this is subject to prosecutorial discretion by the agencies.20 Whether or not out-of-market efficiencies should be accepted is analogous to the issue discussed earlier of how to aggregate heterogeneous welfare effects across consumers. Depending on one’s view of the optimal social welfare function, it may be just as appropriate, or inappropriate, to offset pro- and anticompetitive effects across consumers purchasing in distinct antitrust markets as it is to offset pro- and anticompetitive effects across consumers who are purchasing in the same antitrust market but have heterogeneous preferences. In his address, Judge Ginsburg distinguishes consumer welfare concerns from other non-competitive public interest considerations.21 We understand that Judge Ginsburg believes that the former is the purview of antitrust, while the latter is not. From an economic perspective, this dichotomy may be difficult to defend. Considerations such as income or social inequality, ecological effects, or other factors mentioned by Judge Ginsburg,22 are not outside the scope of the consumer welfare standard from an economic perspective. On the contrary, they are squarely within it so long as consumers meaningfully care about these effects. Economics recognizes that, for example, income inequality could be detrimental to consumer welfare— mostly because economics is essentially agnostic about what preferences determine consumer welfare. Within economics, 19 See U.S. DEP’T OF JUSTICE & FED. TRADE COMM’N, HORIZONTAL MERGER GUIDELINES § 10 (2010). 20 In particular, the Merger Guidelines note that the agencies may “consider efficiencies not strictly in the relevant market, but so inextricably linked with it that a partial divestiture or other remedy could not feasibly eliminate the anticompetitive effect in the relevant market without sacrificing the efficiencies in the other market[s].” See id. at § 10 n.14. 21 Ginsburg, supra note 1, at 835–38. 22 Id. 3_2019.3_ARON (DO NOT DELETE) 12/25/2019 11:20 PM No. 3:849] AN ECONOMIC PERSPECTIVE ON BALANCING 865 we take preferences as given. Dividing preferences into “antitrust relevant” and “antitrust irrelevant” groups runs the risk of attributing normative value to some preferences at the expense of others, rather than taking them as given. Even within a broadly conceived notion of consumer welfare, however, it need not be the case that all qualitative effects must be considered. As discussed earlier, how to aggregate consumer welfare across people is highly subjective with no correct answer (at least, none provided by economic science) of how this might be accomplished. One potential rationale for excluding non-competition public interest considerations is that they often involve welfare effects of consumers outside of the relevant antitrust market. If, for example, the benefits of reducing income inequality largely fall to consumers who do not purchase the product at issue, then such effects would have limited importance when applying a social welfare function that focuses on the welfare of within-market consumers. Thus, excluding such effects as beyond the purview of traditional antitrust analysis could be justified as a proxy for using a narrow definition of “relevant consumer.” We recognize, however, that there is a range of opinions regarding who is the relevant consumer about whose welfare antitrust should be concerned. Some believe that antitrust’s scope should be narrowly focused on consumers in the antitrust market at issue, while others take a more expansive view.23 A narrower definition of “relevant consumer” results in the potential exclusion of certain effects such as social and political considerations—a position more closely aligned with 23 See The Consumer Welfare Standard in Antitrust: Hearing Before the Subcomm. on Antitrust, Consumer Prot. and Consumer Rights of the S. Comm. on the Judiciary, 115th Cong. 3 (2017) (opening statement of Professor Carl Shapiro) (advocating for a consumer welfare standard that focuses on harm to trading parties on the other side of the market). See also Tim Wu, After Consumer Welfare, Now What? The “Protection of Competition Standard” in Practice, ANTITRUST CHRON., Apr. 2018, at 12, 18– 19 (advocating for the protection of the competitive process generally). See generally ANTITRUST CHRON., Apr. 2018 (including several articles discussing “Hipster Antitrust”). 3_2019.3_ARON (DO NOT DELTE) 12/25/2019 11:20 PM 866 COLUMBIA BUSINESS LAW REVIEW [Vol. 2019 Judge Ginsburg’s view on the range of effects that should be properly included in an antitrust analysis.24 Without a doubt, weighing the benefits to one set of consumers against costs to others is a fraught exercise. It is not clear, however, that the challenge facing the decisionmaker is greater when the winners and losers are entirely separate consumer groups than when they fall within the same consumer group, but with varying effects for different consumers. As a result, it is not clear that ruling out “non-traditional” considerations on the grounds that they are outside the realm of traditional antitrust analysis is justified by the observation that the tradeoffs are hard to make or are potentially subjective. Indeed, ruling out such considerations may result in ignoring effects that are central to social welfare determinations, and refusing to consider them may amount to promoting tractability at the expense of consumer welfare. As noted earlier, it may well be that the evidence in a given case does not permit a disciplined and rigorous consideration of effects outside of those directly impacting the consumers in the market at issue. In such cases, these more nebulous effects, while potentially of great importance to consumers, may be best addressed in a different venue than the antitrust court, consistent with Judge Ginsburg’s view.25 IV. CONCLUSION Qualitative evidence and effects pose significant challenges, but economists must, and routinely do, apply rigorous, evidence-based analysis to overcome them. In making judgments about countervailing effects of alleged anticompetitive mergers or other conducts, we do not serve the objective of maximizing consumer welfare by establishing a rule of ignoring certain factors that may affect consumer welfare. Indeed, one may argue that by ignoring them we may in some cases affirmatively expunge from the consumer welfare calculus factors that are central to it, to the detriment of the goals of maximizing consumer welfare. 24 See generally Ginsburg, supra note 1. 25 Id.