Economic Analysis and Competition Policy Research

Home   •   About   •   Analytics   •   Videos

This piece originally appeared in ProMarket but was subsequently retracted, with the following blurb (agreed-upon language between ProMarket’s Luigi Zingales and the authors):

“ProMarket published the article “The Antitrust Output Goal Cannot Measure Welfare.” The main claim of the article was that “a shift out in a production possibility frontier does not necessarily increase welfare, as assessed by a social welfare function.” The published version was unclear on whether the theorem contained in the article was a statement about an equilibrium outcome or a mere existence claim, regardless of the possibility that this outcome might occur in equilibrium. When we asked the authors to clarify, they stated that their claim regarded only the existence of such points, not their occurrence in equilibrium. After this clarification, ProMarket decided that the article was uninteresting and withdrew its publication.”

The source of the complaint that caused the retraction was, according to Zingales, a ProMarket Advisory Board member. The authors had no contact with that person, nor do we know who it is. We would have welcomed published scholarly debate versus retraction compelled by an anonymous Board Member.

We reproduce the piece in its entirety here. In addition, we provide our proposed revision to the piece, which we wrote to clear up the confusion that it was claimed was created by the first piece. We will let our readers be the judge of the piece’s interest. Of course, if you have any criticisms, we welcome professional scholarly debate.

(By the way, given that the piece never mentions supply or demand or prices, it is a mystery to us why any competent economist could have thought it was about “equilibrium.” But perhaps “equilibrium” was a pretext for removing the article for other reasons.)

The Antitrust Output Goal Cannot Measure Welfare (ORIGINAL POST)

Many antitrust scholars and practitioners use output to measure welfare. Darren Bush, Gabriel A. Lozada, and Mark Glick write that this association fails on theoretical grounds and that ideas of welfare require a much more sophisticated understanding.

By Darren Bush, Gabriel A. Lozada, and Mark Glick

Debate seems to have pivoted in the discourse on consumer welfare theory to the question of whether welfare can be indirectly measured based upon output. The tamest of these claims is not that output measures welfare, but that generally, output increases are associated with increases in economic welfare. 

This claim, even at its tamest, is false. For one, welfare depends on more than just output, and increasing output may detrimentally affect some of the other factors which welfare depends on. For example, increasing output may cause working conditions to deteriorate; may cause competing firms to close, resulting in increased unemployment, regional deindustrialization, and fewer avenues for small business formation; may increase pollution; may increase the political power of the growing firm, resulting in more public policy controversies and, yes, more lawsuits being decided in its interest; and may adversely affect suppliers. 

Even if we completely ignore those realities, it is still possible for an increase in output to reduce welfare. These two short proofs show that even in the complete absence of these other effects—that is, even if we assume that people obtain welfare exclusively by receiving commodities, which they always want more of—increasing output may reduce welfare. 

We will first prove that it is possible for an increase in output to reduce welfare under the assumption that welfare is assessed by a social planner. Then we will prove it assuming no social planner, so that welfare is assessed strictly via individuals’ utility levels.

The Social Planner Proof 

Here we show that a shift out in a production possibility frontier does not necessarily increase welfare, as assessed by a social welfare function.

Suppose in the figure below that the original production possibility frontier is PPF0 and

the new production possibility frontier is PPF1. Let USWF be the original level of social welfare, so that the curve in the diagram labeled USWF is the social indifference curve when the technology is represented by PPF0. This implies that when the technology is at PPF0, society chooses the socially optimal point, I, on PPF0. Next, suppose there is an increase in potential output, to PPF1. If society moves to a point on PPF1 which is above and to the left of point A, or is below and to the right of point B, then society will be worse off on PPF1 than it was on PPF0. Even though output increased, depending on the social indifference curve and the composition of the new output, there can be lower social welfare.

The Individual Utility Proof

Next, we continue to assume that only consumption of commodities determines welfare, and we show that when output increases every individual can be worse off. Consider the figure below, which represents an initial Edgeworth Box having solid borders, and a new, expanded Edgeworth Box, with dashed borders. The expanded Edgeworth Box represents an increase in output for both apples and bananas, the two goods in this economy.

The original, smaller Edgeworth Box has an origin for Jones labeled J and an origin for Smith labeled S. In this smaller Edgeworth Box, suppose the initial position is at C. The indifference curve UJ0 represents Jones’s initial level of utility with the smaller Edgeworth Box, and the indifference curve US represents Smith’s initial level of utility with the smaller Box.  In the larger Edgeworth Box, Jones’s origin shifts from J to J’, and his UJ0 indifference curve correspondingly shifts to UJ0′.  Smiths’ US indifference curve does not shift. The hatched areas in the graph are all the allocations in the bigger Edgeworth Box which are worse for both Smith and Jones compared to the original allocation in the smaller Edgeworth Box.

In other words, despite the fact that output has increased, if the new allocation is in the hatched area, then Smith and Jones both prefer the world where output is lower. We get this result because welfare is affected by allocation and distribution as well as by the sheer amount of output, and more output, if mis-allocated or poorly distributed, can decrease welfare.

GDP also does not measure aggregate Welfare 

The argument that “output” alone measures welfare sometimes refers not to literal output, as in the two examples above, but to a reified notion of “output.” A good example is GDP.  GDP is the aggregated monetary value of all final goods and services, weighted using current prices. Welfare economists, beginning with Richard Easterlin, have understood that GDP does not accurately measure economic well-being. Since prices are used for the aggregation, GDP incorporates the effects of income distribution, but in a way which hides this dependence, making GDP seem value-free although it is not. In addition, using GDP as a measure of welfare deliberately ignores many important welfare effects while only taking into account output. As Amit Kapoor and Bibek Debroy put it:

GDP takes a positive count of the cars we produce but does not account for the emissions they generate; it adds the value of the sugar-laced beverages we sell but fails to subtract the health problems they cause; it includes the value of building new cities but does not discount for the vital forests they replace. As Robert Kennedy put it in his famous election speech in 1968, “it [GDP] measures everything in short, except that which makes life worthwhile.”

Any industry-specific measure of price-weighted “output” or firm-specific measure of price-weighted “output” is similarly flawed.

For these reasons, few, if any, welfare economists would today use GNP alone to assess a nation’s welfare, preferring instead to use a collection of “social indicators.”


Output should not be the sole criterion for antitrust policy. We can do a better job of using competition policy to increase human welfare without this dogma. In this article, we showed that we cannot be certain that output increases welfare even in a purely hypothetical world where welfare depends solely on the output of commodities. In the real world, where welfare depends on a multitude of factors besides output—many of which can be addressed by competition policy—the case against a unilateral output goal is much stronger.


The Original Sling posting inadvertently left off the two proposed graphs that we drew as we sought to remedy the Anonymous Board Member’s confusion about “equilibrium.” We now add the graphs we proposed. The explanation of the graphs was similar, and the discussion of GNP was identical to the original version.

The Proof if there is a Social Welfare Function (Revised Graph)

A diagram of apples and apples

Description automatically generated

The Individual Utility Proof (Revised Graph)

A diagram of a graph

Description automatically generated

The New Merger Guidelines (the “Guidelines”) provide a framework for analyzing when proposed mergers likely violate Section 7 of the Clayton Act that is more faithful to controlling law and Congressional intent than earlier Guidelines. The thirteen guidelines presented in the new Guidelines go quite a long way in pulling the Agencies back from an approach that placed undue burden on plaintiffs and ignored important factors such as the trend in market concentration and serial mergers that were addressed by earlier Supreme Court precedent. The Guidelines also incorporate the modern, more objective economics of the post-Chicago school of economics. For these reasons, and others, the Guidelines should be applauded.

Unfortunately, remnants of Judge Bork’s Consumer Welfare Standard remain. In several places the Guidelines refer to a merger’s anticompetitive effects as price, quantity (output), product quality or variety, and innovation. These are all effects that can shift demand curves or equilibrium positions in the output market and thus increase consumer surplus, the only goal recognized by the Consumer Welfare Standard.

To their credit, the Guidelines also mention input markets, referring to mergers that decrease wages, lower benefits or cause working conditions to deteriorate. Lower wages reduce labor surplus (rent), a consideration that would come within a Total Trading Partner Surplus approach. However, the traditional goals of antitrust as articulated by Congress and many Supreme Court opinions, including protecting democracy through dispersion of economic and political power, protection of small business, and preventing unequal income and wealth distribution, are conspicuously absent.

The basis for these traditional goals is well known. Prominent economist Stephen Martin has documented the judicial and congressional statements concerning the antitrust goal of dispersion of power. The historical support for the goal of preserving small business can be found in a recent paper by two of the authors of this piece. Lina Khan and Sandeep Vaheesan, and Robert Lande and Sandeep Vaheesan, have laid out the textual support for the antitrust inequality goal. Moreover, welfare economists have empirically demonstrated significant positive welfare effects from democracy, small business formation, and income equality.

Indeed, the Brown Shoe opinion, on which the Guidelines heavily rely, examined whether the lower court opinion was “consistent with the intent of the legislature” which drafted the 1950 Amendments, and the opinion itself refers to the goal of “protection of small businesses” in at least two places. The legislative history of the 1950 Amendment deemed important by the Brown Shoe Court evinced a clear concern that rising concentration will, according to Senator O’Mahoney, “result in a terrific drive toward a totalitarian government.”

The remnants of the Consumer Welfare Standard are most evident in the Guidelines’ rebuttal section on efficiencies. The Guidelines open the section with the recognition that controlling precedent is clear that efficiencies are not a defense to a merger that violates Section 7; accordingly, the section is offered as a rebuttal rather than a defense. In essence, if the merging parties can identify merger-specific and verifiable efficiencies, it can rebut a finding that the merger substantially lessened competition. The Guidelines do not define “efficiencies.”  However, the context makes clear that Guidelines mean to follow previous versions of the Merger Guidelines, that assume “efficiencies” are primarily cost savings. A defendant can offer a rebuttal to a presumption that a merger may significantly harm competition, if such cost savings are passed through to consumers in lower prices, to a degree that offsets any potential post-merger price increase. There are at least six reasons why the Agencies should jettison this “efficiency” rebuttal.

First, lower prices resulting from cost savings are quite a bit different than lower prices resulting from entry (rebuttal by entry). New entry reduces concentration, but cost savings at best will only lower output price, and higher prices (or reduced output) is not the sole problem that results from high concentration except under a strict Consumer Welfare Standard.

Second, to the extent the Guidelines equate efficiencies with cost savings (as in earlier merger guidelines), they have adopted the businessman’s definition of efficiencies. In contrast, economic theory suggests that some cost savings lower rather than raise social welfare. For example, cost savings from lower wages, greater unemployment, or redistribution between stakeholders can both lower welfare and reduce prices. An increase in consumer or producer surplus that comes at the expense of input supplier surplus can also lower welfare.

Third, only under the output-market half of a surplus theory of economic welfare, which is the original Consumer Welfare Standard, can one clearly link cost savings to economic welfare, because lower cost increases consumer and/or producer surplus. As we show elsewhere, this theory has been thoroughly discredited by welfare economists. In fact, for economists, “efficiency” only means Pareto efficiency. As discussed by Gregory Werden and by Mas-Colell et al.’s leading Microeconomics textbook (Chapter 10), the assumptions necessary to ensure that maximizing surplus results in Pareto Efficiency are extreme and unrealistic. These assumptions include quasilinear utility, perfectly competitive other markets, and lump sum wealth redistributions that maximize social welfare. This discredits the surplus approach, which is the only way to reconcile Pareto Efficiency, which is what efficiencies mean in economic theory, with cost savings, which is the definition implied in the Guidelines.

Fourth, the efficiency section is superfluous. As many economists have recognized, most recently Nancy Rose and Jonathan Sallet, the merging parties are already credited for efficiencies (cost savings) in the “standard efficiency credit” which undergirds Guideline 1. After all, absent any efficiencies, why allow any merger that evenly weakly increases concentration? A concentration screen that allows some mergers and not others must be assuming that all mergers come with some socially beneficial cost savings. Why do we need another rebuttal section when cost savings have already been credited?

Fifth, there is no empirical research to suggest that mergers that increase concentration actually lower costs and pass on sufficient benefits to consumers to constitute a successful rebuttal. As one district court commented, “The Court is not aware of any case, and Defendants have cited none, where the merging parties have successfully rebutted the government’s prima facia case on the strength of the efficiencies.” We have identified nine studies measuring either cost savings or productivity gains or profitability from mergers spanning industries like health insurance, banking, utility, manufacturing, beer, and concrete industries. Five of these studies find no evidence of productivity gain or a cost reduction. The other four studies find productivity gains in terms of cost savings; but three of these four studies report a significant increase in prices to the consumers post-merger, and the remaining study does not report price effects post-merger. In other words, we have not been able to find any empirical study showing post-merger pass on of cost savings to consumers. These results are consistent with those of Professor Kwoka, who performs a comprehensive meta-analysis of the price effects of horizontal mergers and finds that the post-merger price at the product-level increases by 7.2 percent on average, holding all other influences constant. More than 80 percent of product prices show increases, and those increases average 10.1 percent.

Sixth, even if there were cost savings from mergers it is unlikely that they would be merger- specific and verifiable. Earlier versions of the Merger Guidelines expressed skepticism that economies of scale or scope could not be achieved by internal expansion (1968 Merger Guidelines) or that cost savings related to “procurement, management or capital costs” would be merger specific (1997 Merger Guidelines). In their article on merger efficiencies, Fisher and Lande write that “it would be extremely difficult for merging firms to prove that they could not attain the anticipated efficiencies or quality improvements through internal expansion.” Louis Kaplow has argued that the ability to use contracting to achieve claimed efficiencies is seriously underappreciated or studied. Verification of future efficiencies is also inherently problematic. The 1997 Merger Guidelines state that efficiencies related to R&D are “less susceptible to verification.” This problem and other verification hurdles are discussed by Joe Brodley and John Kwoka. In summary, the New Merger Guidelines could be improved by a footnote in Guideline One clarifying the multiple antitrust goals Congress sought to achieve by preventing concentrated markets through mergers. In addition, the Agencies should take seriously the holdings of at least three Supreme Court Opinions, none of which have been overturned (Brown Shoe, Phila. Nat’l Bank and Procter & Gamble Co.) that (as quoted in the Guidelines) “possible economies [from a merger] cannot be used as a defense to illegality.” There are good reasons to abandon an efficiencies rebuttal as well.

Mark Glick, Pavitra Govindan and Gabriel A. Lozada are professors in the economics department at the University of Utah. Darren Bush is a professor in the law school at the University of Houston.