“Consumer Welfare” has lost its place as the animating value and standard for modern antitrust. The standard is almost universally regarded as bunk and a “bad take” on the original goal of “competition” as rivalry. But the discussion about “what went wrong with Consumer Welfare” cannot be just about how we define the headline value, as that value ultimately found form in the tools that were developed to pursue it. The end of the Consumer Welfare Standard demands that antitrust practitioners and authorities revamp their current suite of tools, and not just use the same old tools under a new headline.

This article is a response to the papers submitted by the panelists of the recent “2023 Antitrust and Competition Conference – Beyond the Consumer Welfare Standard?” In those papers, the panelists proposed their ideas for how to revise or replace the Consumer Welfare Standard. You can find those ideas here.


From Ordoliberalism, to Chicago-lite, to “this is how we do things”

The debate on whether Consumer Welfare (“CW”) is a fit objective for antitrust remains predominantly a U.S. one because of the country’s adherence to the “efficiency” paradigm that has been the lodestar of its antitrust enforcement for the last 40 years. Europe has had a different trajectory, with the origins of its enforcement philosophy steeped in “ordoliberalism”—the notion that the purpose of antitrust is to preserve “decentralised decision making” andfreedom of economic agents,” not efficiency. These differences have elevated the importance of market structure in Europe, as with concentrated power there is no choice. Relatedly, there has been an obligation on large firms to behave “fairly” with smaller competitors because “residual competition” must be protected against strategies that would strengthen dominance at the expense of consumer choice.

Europe, too, ended up adopting the Chicago posture in the mid-late 1990s. There was a major “push,”  led by economists, to adopt Chicago School economics, which was seen as a more rigorous and sophisticated method for the enforcement of antitrust than the vaguely defined “blancmange” of European competition policy at the time. Europe discarded its public interest tests and structural presumptions in favor of a slew of Chicago-inspired axioms: “two is enough;” “monopolies are the product of superior efficiency and are just fine” (as long as they don’t actively kill rivals); “less efficient/ higher-cost competitors deserve to die” (the logic of the “As Efficient Competitor Test”); “vertical integration eliminates the double margin;” “price discrimination is output expanding;” “exclusivity is all about ex-ante competition for the market;” “exclusives are necessary to justify relationship-specific investments;” and practically every merger is “procompetitive.” 

Europe never embraced the U.S. fetishism for market definition, a manifestation of the CW-centric approach and a burden that often shapes and distorts the formulation of the theory of harm. More so, market definition in Europe is increasingly an afterthought (we pragmatically define the market at the end, to fit the theory of harm). However, European substantive economic defenses also put CW at the center. We never explicitly marked the change with a positive statement that “from now on the goal is Consumer Welfare,” but we rapidly adopted this approach. European economists pushed hard the “more economic approach” that emphasized efficiency in the early 2000s, culminating in 2007-8 in the European Commission issuing a Guidance Paper on Exclusionary Abuse, just after the 2007 Non-Horizontal Guidelines. It is much like the expression “there is an app for that”: every form of conduct by a dominant firm has been given a “pro-competitive/efficiency-enhancing justification.” A very good exculpatory narrative has been invented for every conceivable conduct.

In practice, European practitioners developed (and antitrust agencies came to accept) an almost cartoonish “phrasebook” that became used in every case, with the upshot that every deal has become “pro-competitive” because it creates a stronger rival or eliminates the double margin; or it will “not materially reduce competition” because if we combine a made-up margin and a diversion ratio into a GUPPI (Gross Upward Pricing Pressure Index), it implies a predicted price increase of “only” 3.475%; or a bidding study shows there is no systematic difference in prices/margins with two bidders instead of 10; or “breaking the One Monopoly Profit” is always “hard.”

At the more “sophisticated” end, we have used highly stylized models to pretend to “measure” CW  effects, usually with a simple Hotelling or differentiated Bertrand framework, where CW is a simple combination of prices and quantities. We like to say we also “include quality and innovation” into CW, but in practice quality and innovation end up in the measure of CW by either increasing quantities consumed, or (equivalently) reducing prices.

The U.S. debate challenging CW has been less polarized in Europe—the prevailing sentiment being that “we are doing just fine, we don’t pursue a narrow notion of CW that cares just about output and prices, we include also quality and innovation.” Yet, as mentioned, quality and innovation often end up being formally modelled in terms of prices and quantities anyway. And the European Commission’s attempt to focus on “innovation” in the agrochemical mergers (Bayer/Monsanto, Dow/Dupont in 2017) was met with strong pushback, including by many economists engaged in a “battle of the models,” which led to that effort being shut down and not pursued further. 

European agencies are (in the main, with Germany and the United Kingdom as perhaps the exceptions) still operating with a view that antitrust is a narrow economic policy tool. Perhaps this is because Europeans feel that we have moved more recently out of a past where we had “public interest” objectives confusingly mixed in with antitrust, or in part because the influence of the classic European economic schools (highly regarded seats of industrial organization theory) is still pervasive and unchallenged. Data protection is a completely distinct domain; labor issues do not touch the sides.

What should a “rethink” involve?

Defenders of the status quo appear to argue that CW is a good objective because we can be sure that practices that expand output and reduce prices are generally good and, importantly, measurable. However, this claim overstates our ability to measure anything (can we really measure with any serious reliability in an actual case what the output-expanding or price-reducing effects of a practice are? Anyone who has been doing these things at the coalface knows there is a lot of false precision and made-up numbers in all of these analyses). Furthermore, we don’t even measure price and output effects in practice but rely on indirect proxies like changes in market shares and GUPPIs, from which we infer some price and output effect. Finally, and more fundamentally, this outlook reduces the goal of enforcement to one that justifies all sorts of means. For example, we “can measure” that if a super-dominant company with major cost advantages can respond to new entrants by cutting prices it will expand output and this is “efficient,” but this may also cause the entrant to no longer be viable. Or “we can measure” that a bundle discount expands output, and for this reason we can almost never show in theory that mixed bundling (conditional discounts) is a problem. “We can measure” that a merger may reduce costs, but most cost savings do not happen and are not passed on to consumers anyway.

The reasoning is circular: “we care about these indicators because we can measure them.” But even then, we are not measuring much. We are measuring at most distant proxies, like market shares, instead of understanding how assets and capabilities are being combined and reconfigured. The reality is that a firm making an acquisition acquires assets and capabilities and combines them with what it already owns. It does not just buy the share of product X in market Y. It is the combination of assets and capabilities that can create new products and services, but also new potential for bottlenecks and hoarding assets from others.

When we have merger after merger concentrating assets in fewer and fewer hands, with cross-industry assets and capabilities coming together under the same roof, it is wrong for antitrust enforcers to look piecemeal and serially at the change in shares in the smallest product markets to inform us about what is going on. Enforcers currently do hundreds of market share cuts to infer the increment in market power. In vertical and conglomerate deals, we have models with mechanisms to show leveraging of market power from market A to market B, through some form of tying or bundling or raising rivals’ costs (RRC), but we don’t have a way of articulating a concern around a deal bringing together multiple assets. A large company “having stuff” may not sound like a theory of harm, but this very issue has emerged in conglomerate acquisitions of assets by large digital companies, where foreclosure analysis of a conventional type is not very apt and yet there is an agglomeration of assets in ways that may create barriers and bottlenecks at various places.

Overall, the idea that we must continue to do what we do because “it is measurable” is deeply misguided. That things are measurable is mostly an illusion. And what is measurable is also not so interesting or dispositive most of the time.

What should we be doing then?

At the Antitrust, Regulation and the Political Economy conference last month, Aviv Nevo, the new chief economist for the U.S. Federal Trade Commission, said enforcers working on cases need commentators not to fantasize idly about impractical rules but to “roll up their sleeves” and be clear on what criteria enforcers should be adopting. Companies and the bar also constantly clamor for “predictability.” I am in favor of bright lines if we also accept that the task cannot be to design something that “optimally deals” with all trade-offs: we are where we are because the obsession with avoiding Type 1 errors has led to systematic underenforcement. We need to accept that part of the necessary correction is that we will make Type 2 errors. The dread of “protecting inefficient competitors” as the guiding light for enforcement needs to be softened. We have been there, done that, with the result that in multiple markets there are no competitors left standing, however inefficient. At the same time, we need to articulate theories of harm beyond the narrow set of “dance moves” we know so well and broaden them to incorporate insights from outside the narrow church of IO. This is anathema to many of the economists who dominate the profession, but it is something we need to do.

Note, this does not mean we should ignore all economic insights, nor that we should burden agencies and judges with making trade-offs across disparate societal objectives (a standard “go to” critique of defenders of the status quo). If we “agree” that high levels of concentration are associated with market power, which is associated with multiple societal harms, we need to focus more on concentration per se. In practice:

Antitrust enforcement needs to overcome its denial of the significance of major underlying shifts in industry structure and performance indicators. The reaction to the early evidence (circa 2015) of an increase in margins and concentration across markets was scorn and disbelief. The recent work of De Loacker and Eeckout and Pellegrino about how merger activity and market concentration are shaping economic performance should convince everyone that we cannot continue to use the same narrow tools focused on the share of product sales in a particular market to measure whether there is market power, and whether conduct is therefore good or bad. We need the assessment of conduct and mergers not to be independent of the underlying drift.

Antitrust enforcement needs to get out of the horizontal/vertical/conglomerate paradigm and recognize that what matters for the presence and creation of market power is the distribution of assets and capabilities firms have, and how a deal or a piece of conduct changes that. The incipient interest for “ecosystem theories of harm” in acquisitions by digital giants is a recognition that companies are not just buying a “product” with some revenue. In the main, they are buying assets and capabilities that will be placed in the existing network or assets and capabilities they already own. This can create opportunities for new products but also incentives to hoard assets and deny capabilities to others.

Antitrust enforcement needs to also recognize it is in principle an immensely powerful tool for wealth and power redistribution. There is no escaping this. We debate the subtle differences between protecting “competition,” the “competitive process,” or “a process rivalry” (all semantic differences) and have been taming enforcement into a set of technocratic rules to protect it from becoming a political weapon. But antitrust has, in principle, enormous potential for inducing change that directly benefits consumers and other stakeholders. At the most basic level, antitrust can stop companies from adopting practices that exploit consumers and workers, from “right to repair” to “non competes” to targeting “junk bank fees.” The delineation between antitrust, consumer protection and sectoral regulators is irrational when everything emanates from the same market power.

I do not have an alternative standard or list of rules, but some progress can perhaps be made thinking of the following:

  1. In each case, we should try to understand industry developments (concentration, margins, barrier to entry) over the previous 10-20 years, not just the past three years. This is not a prescription for a lengthy academic study, which enforcers don’t have time to do. But we need to understand how competitive advantages have shifted, how the distribution of technology has changed relative positions, how rents have redistributed, and how innovation is trending. Concentration of assets has taken place serially and incrementally in many industries. Current structures have not emerged “fully formed” in the last three years but are the product of longer-term dynamics, previous deals and conduct. Yet, we operate often under total oblivion of the past. Each deal is incremental, yes, but it matters to the assessment how the distribution of assets and capabilities has evolved over the longer run.
  1. We should develop a common understanding of how to measure “margins” if it is to meaningfully gauge market power. There is too much leeway in practice for consultants and economists to “cook” margins in the most bizarre ways. Anyone who has been involved in margin calculations in real life knows there is great scope for obfuscation. Margins ought to be part of a composite set of indicators, but only if we come to a common view on how to measure them.
  1. We should actively seek to develop an understanding of how to map capabilities and assets for a firm, its target, and rivals. We need to form a view on what a deal does and what a piece of conduct does. Enforcers need to work not just with IO economists but also, for example, network economists and data protection experts. Technology (AI/ML) can help accomplish this at scale by using automated reviews of internal documents to map out the firm, for example.
  1. We can also adopt some presumptions, especially in mergers where damage is prospective, and undoing it ex post is much worse than preventing a deal from going forward in the first place. We abandoned all structural presumptions in the late 1990s because we fell for the line that everything had to be case-by-case or we risked massive Type 1 errors. This has failed and has led us to a cumbersome case-by-case approach that sucks up resources and delivers nothing. I would favor a rebuttable presumption of harm for deals (horizontal, vertical, conglomerate), where the buyer has significant market power in its core markets. This is close to the proposals made in ProMarket by Filippo Lancieri and Tommaso Valletti. The “rebuttable” part would require firms to provide concrete plans that the deal would enable the creation of new services and new offerings AND simultaneously would not lead to the “killing” of an equivalent innovation effort internally. Conversely, a deal where there is evidence that the incumbent/buyer is “building” an equivalent capability internally and there is a realistic possibility it is going to cannibalize the target instead and deny access to the capability to others would not be a candidate for rebuttal.
  1. Presumptions are harder to prescribe for conduct: although discriminatory rules, exclusivities, discounts, tying/bundling, and input restrictions are often harmful when practiced by a dominant firm and discourage entry. A presumption will inevitably catch situations where there is in fact a benefit or a justification. So be it. As mentioned, we need to compensate for overindulgence the other way. The circumstances in which output expansion saves the day in discrimination cases, for example, are few and far between, and mostly based on heroic assertions, not measurement. The circumstances in which the market power conferred by exclusivity is going to be dissipated ex ante by competition for the exclusive relationship are again few and far between.
  1. We should certainly abandon benchmarking conduct on the incumbent’s performance. Anything premised on “as efficient” (AECT, margin squeeze) should go.This is a monopolist’s charter.

We have, in some way, come full circle. I supported the “more economic approach” in the mid-1990s, thinking this was a much-needed way of bringing antitrust in Europe into the modern age. Yet, the concept of CW that we implicitly came to adopt has embodied a vision of enforcement that has protected incumbents from antitrust intervention. It has done so by associating an increase in CW to situations where a firm adopts practices that purportedly lower prices and eliminate “inefficient” competitors. But this ignores so much of what we should be worrying about. Competition involves “choice”—if there is no choice there is no competition. Yes, we will sacrifice economies of scale and scope. Ironically, for Europeans, there is more than a little ordoliberal flavor in this posture. But then, this is not so dissimilar from the U.S. calls for a return to the original animating values of antitrust: freedom of choice and curbing of corporate power. It is just a pity we do not have quite the same poetic flourish in the European narrative right now around protecting democracy and freedom from all masters…

Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.