Over the past several years, the case against conduct remedies seems largely to have prevailed. Experience has shown that supposed remedies that are essentially rules seeking to prohibit profitable actions by merged companies will predictably elicit efforts to circumvent them. Both evidence and anecdotes confirm that firms routinely succeed in these efforts, resulting in failures of competition policy relying on such remedies. This should not have been a surprise to anyone familiar with industry regulation. The very point of many regulatory rules is to prevent firms from unconstrained maximization of their profit. Decades of research have shown how often this is futile or, worse, counterproductive.
As a result, most competition authorities worldwide have now declared their strong preference for structural remedies. Done correctly, a divestiture creates a separate entity, with its own profit-maximizing objectives similar to those in the premerger market. This harnesses a firm’s independent incentives and relies on competitive forces to restore market competition. At least in principle, this should allow competition authorities to avoid regulatory-type oversight.
But recently, both the Federal Trade Commission and the Antitrust Division of the Department of Justice have gone further and indicated their determination to avoid the use of divestitures as well as conduct remedies. The head of the Antitrust Division, Jonathan Kanter, has said:
Merger remedies short of blocking a transaction too often miss the mark. Complex settlements, whether behavioral or structural, suffer from significant deficiencies. When the Division concludes that a merger is likely to lessen competition, in most situations we should seek a simple injunction to block the transaction.
This critical view has been seconded by FTC Chair Lina Khan. In commenting on what had often been protracted negotiations between merging parties and the agencies seeking acceptable divestiture settlements, she stated:
This is not work that the agency should have to do. That’s something that really should be fixed on the front end and by parties being on clear notice about what are lawful and unlawful deals. We’re going to be focusing our resources on litigating rather than settling.
There is now good evidence of changes in agency practice. In 2021, before these announced policy changes, the FTC challenged 18 mergers, five of which were settled with some type of remedy, seven abandoned, and six litigated. The DOJ challenged 14 mergers, settling nine, while three were abandoned and two litigated. After these declarations by the agency heads, the data begin to show a different pattern. DOJ has entered into a settlement only once (and that under unusual conditions), while the FTC has done so in ten instances but has refused to accept remedies in certain cases where it would have done so in the past.
These changes in policy and practice have been to the surprise of many and the consternation of some, but as will be described, these changes in fact represent the continuing evolution of the agencies’ remedies policy. There have been sufficient troublesome experiences—including outright failures—of divestitures to prompt increasing caution about their use, as well as concern that the necessary conditions for their success have not been sufficiently attended to and that their inherent limitations have not been fully appreciated. Transactions such as Haggens-Safeway and Hertz-Dollar are well-cited examples, but divestitures in T-Mobile/Sprint, Ticketmaster-Live Nation, and many pharmaceutical mergers also illustrate their limitations and failures and demonstrate how accepting inadequate divestitures can result in effectively allowing anticompetitive mergers to proceed with few if any constraints.
Two major forces have converged to produce this policy reckoning. The first is the recognition that so-called “structural remedies” are not as free of adverse incentives and opportunities for mischief as has often been thought, or at least hoped. Merging firms’ contrary incentives can play a role in their design and implementation, compromising—and at times sabotaging—what may appear to be straightforward divestitures. Examples demonstrate that the agencies have not always prevented these forces from taking effect.
The second factor stems from economic research on structural remedies. That research has often been interpreted as generally supportive of divestiture remedies. A careful reading, however, reveals weak methodology and inconsistent standards of what constitutes “success” in some widely cited studies. In reality, the track record of structural remedies in practice is considerably more mixed than some authors claim and many advocates would like to believe.
While these considerations have resulted in a more cautious—indeed, skeptical—view of divestitures, they do not imply the absence of any possible role for structural remedies. Experience, however, suggests a far narrower role than in recent practice. A useful complement to this new policy would be further studied by the agencies of criteria for when structural remedies are most problematic and when they are more likely to succeed in preserving or restoring competition.
The purpose of all remedies is to preserve or restore the degree of competition otherwise lost as a result of the merger. This also represents the criterion against which to judge whether or to what degree a particular remedy is in fact successful. A useful starting point in this discussion is a description of the canonical case of a divestiture remedy. Suppose two firms proposing to merge each have numerous geographic locations (as in retailing) or numerous products (as with many if not most manufacturers). Suppose further that the two firms overlap in only a handful of their many locations or products. Requiring the firms as a condition of their merger to divest to a capable buyer one of their two operations in each of the overlap markets will preserve the same number of independent and viable entities and allow the remaining large parts of both businesses to merge free of competitive concerns.
This scenario captures the key attraction of a divestiture remedy, and it is for this reason that the Supreme Court has unequivocally endorsed divestitures. It characterized such remedies as “simple, relatively easy to administer, and sure” as a method for preserving competition without having to block an entire merger (or resorting to conduct remedies). Lying behind this endorsement, however, are some conditions necessary for divestitures to be successful, conditions that over time have not been adhered to as the agencies have used divestitures in a wider array of circumstances than warranted.
Consider, for example, the 2013 merger of Safeway and Albertsons, two large grocery chains, both with substantial presence in the northwest U.S. Their overlaps were not incidental: the FTC identified 146 locations where both chains had stores. In the absence of more suitable firms to which the overlap stores could be divested, the agency settled on a small regional chain of 14 stores—Haggens—as the buyer. But the result of divestiture was that Haggens immediately grew tenfold in size and was overwhelmed in the effort to expand its footprint and operations. It soon filed for bankruptcy, leaving the FTC with little choice but to return many of the overlap stores to the now merged Safeway-Albertsons chain. The Haggens case illustrates two necessary but unfulfilled conditions for a sound divestiture—the need to ensure the quality of divested assets and the operational capabilities of the buyer.
In the 2018 proposed merger of T-Mobile and Sprint, the DOJ scarcely made a serious effort to satisfy either of those conditions. Sprint and T-Mobile were two of only four national wireless carriers, and the DOJ acknowledged that four were necessary for competition in the market. It approved the merger anyway by promising the emergence of—indeed, attempting to create—a new competitor to replace the one being eliminated. That new competitor was to be Dish, a satellite TV operator with no wireless operations or experience, but which would get various assets and “transition assistance” from the merged company so as to become, supposedly, a new robust national wireless carrier within six or seven years. In reality, asset transfers have been halting, Dish has faced problems with financing, there have been disputes with T-Mobile regarding continuation of certain services, and more. The supposed replacement fourth national wireless carrier is nowhere in sight.
The foregoing cases bear no relationship to the canonical model of divestitures. Rather, they illustrate a few of the many pitfalls that arise in the process of attempting divestitures in settings for which they were not intended. A fairly comprehensive list would include the following concerns:
First, the merging parties have every incentive to find the weakest possible purchaser of the assets to be divested—the exact opposite of the agency’s objective. The merging parties have the advantage of superior information about the assets, the requirements for their effective operation, and the potential buyers. Moreover, if the divestiture process involves soliciting bids, the highest bid is likely to come from the firm least likely to bring tough competition to the post-merger market since the tough competitor will drive down profits at the incumbent.
Second, the merging parties have every incentive to offer up assets that will not be sufficient to restore or preserve competition. Again, the merging parties have better information about requirements and every incentive to offer a package that may appear sufficient but in actual practice turns out to require more. This strategy is especially likely to succeed in cases where the production technology is complex.
Third (and related), the merging parties may be able to take actions that diminish the competitive effect of the assets to be divested. In some cases, this may be as simple and perhaps as obvious as transferring personnel, assets, or inventory out of the operation to be divested, or failing to maintain the plant or other assets in the run-up to divestiture. More subtle would be to use the divestiture process to retain the better of every two overlapping locations or products so as to enhance the merging parties’ strength even while ostensibly leaving the same number of stores or products in each market. Once again, the agencies are at a distinct disadvantage in second-guessing these choices.
A fourth—and often unappreciated—concern is that a great many divestitures require at least some limited form of engagement between the merged firm and the buyer of the divested assets, all with the stated purpose of ensuring that the buyer can operate the divested assets effectively. One common form of engagement is so-called “transition services,” which include a variety of support activities ranging from periodic consultations to placement of personnel in the buyer’s firm. These are intended to be temporary assistance with the acquired operations, which in practice typically means one or two years. During that time, the buying firm is dependent on its dominant and direct competitor for information and assistance. The latter, of course, has no incentive to turn its nascent rival into the strongest possible competitor—again, the very problem that infects conduct remedies.
Fifth, in many cases the divesting firm has a critical input—a physical asset, intellectual property, or (less frequently) a function—which is critical to successful operation by the acquiring firm but not easily or quickly undertaken by the latter. The divestiture agreement may require the divesting company to provide the input to the purchaser for an extended period of time. By its very nature, this arrangement creates critical dependence of the purchaser on its direct competitor, mimicking a key weakness of conduct remedies.
The canonical case—seemingly as straightforward as possible—illustrates how the unrecognized critical issue may jeopardize or doom a simple divestiture if overlooked or not fully addressed by the agency. Separate products or geographies may be tied together by important distribution, marketing, R&D and other economies that are the very reason that economically efficient multi-operation firms exist in the first place. In such cases, preserving the headcount of sellers in each consumer market will not by itself maintain the competitive impact of the divested assets, as any number of failed divestiture remedies have shown.
One final issue is the risk that post-divestiture competition may not be as robust as it was premerger, even if the same number of independent producers has been preserved. The search for the best buyer may inadvertently settle on a potential entrant. The divested assets may be accompanied by personnel who can communicate much about the business strategy of their prior affiliation, thereby facilitating coordination. These and other factors may result in diminished competition in the market regardless of the number of sellers in the market. Competition is simply not a guaranteed outcome.
Of course, not all—and in some cases, not any—of these problems might arise. Instead, a divestiture may work very much as set out in the canonical case. Yet, as these examples make clear, that outcome is not assured. Moreover, as divestitures have become used in ever-more ambitious circumstances, the problems of under-resourced purchasers, compromised assets, virtual creation of competitors, etc., make adverse outcomes correspondingly more likely. It is in part for this reason that policy has recently shifted from acceptance of divestitures to caution about their use to avoidance wherever possible. The other reason, as noted at the outset, stems from a more careful reading of the evidence about the effectiveness of divestitures.
Although the effectiveness of divestitures is not necessarily straightforward, first principles cannot tell us whether their limitations outweigh their benefits. That question is ultimately empirical, and there are now a number of useful compilations of merger remedies, most of which focus on divestitures. The first of these was undertaken by the FTC in 1999, with a major follow‑up in 2017, but there are other studies examining U.S. remedies as well as several remedies in other jurisdictions.
The FTC in the 1990s undertook what is generally regarded as the first systematic study of merger remedies, evaluating all of its divestiture orders between 1990 and 1994. There were a total of 35 orders, covering 50 specific divestitures across numerous industries. The key question in this study was said to be whether or not “the approved buyer acquired the assets, began operations, and was operating in the relevant market within a reasonable period.” The agency’s method of evaluation was to seek out buyers of the divested assets in each case, and then to have agency staff interview those buyers.
The relevant buyers could be identified in about three-fourths of cases. Of those, the divested assets were in operation and viable at the time of inquiry in 76% of cases. Based on this, the report headlined that “The Study supports the view that divestitures have been successful remedies for anticompetitive mergers.” Accordingly, many later discussions cite this study as supporting the proposition that divestitures are usually successful remedies for competitively problematic mergers. The OECD, for example, notes that the FTC report found “a generally high success rate of divestments.”
This conclusion, however, is subject to some often overlooked but critical caveats. One threshold issue is simply that the 76 percent “success” rate implies that in nearly one-quarter of cases, the divested assets either had disappeared or were no longer functionally viable. In addition, the relevant time frame was simply stated to be a “reasonable period,” but no definition of that term was offered. The rate of success can be quite different depending on the length of the allowed time period. Moreover, as noted, 13 of the original 50 divestitures could not be investigated because the buyers of the divested assets could not even be identified. The inability even to locate the buyers suggests that those cases may represent instances of asset disappearance. If so, then the failure rate might be considerably greater than the reported 24%, potentially as high as 44%.
All of this suggests that the agency’s declaration that “divestitures have been successful” does not necessarily follow from these data. But there is a further, and crucial, methodological issue with this study. The fact that the divested assets have remained in operation as of the date of the study is not a measure of whether the remedy has in fact preserved the competition that would otherwise be lost due to the merger. Assets may remain in the industry and continue to be viable but still not represent an effective competitive force equivalent to that extinguished by the merger. Viability, in short, is a necessary but not a sufficient condition for a successful remedy, so that the actual success rate is almost certainly less than the survivorship percentage. All that can be correctly concluded from the FTC study is that the rate of successful (i.e., effective) remedies is less than 76% by some indeterminate, probably nontrivial, amount.
Despite these limitations, it should be acknowledged that the FTC Divestiture Study was significant in several respects. It broke new ground in its effort to evaluate remedies. It highlighted factors useful in strengthening remedy policy. And it prompted competition agencies in other countries to examine their remedy policies and in some cases to conduct similar evaluations of their practices.
The FTC conducted a more comprehensive follow-up study published in 2017, but its methodology was surprisingly uneven—better than its earlier study in one area, distinctly weaker in others. This study covered all 89 remedy orders that the agency had entered into between 2006 and 2012. It began by dividing the 89 remedies into three distinct groups and then proceeded—rather startlingly—to evaluate them using three entirely different methodologies and three different definitions of “success.” As a result, its headlined and widely cited finding that “in general remedies set by the Commission were effective” needs reconsideration.
This study’s flaws can be grouped as follows. First, of the 89 remedies, 50 were analyzed through the use of case studies, 15 were assessed based on questionnaires to “market participants,” and the remaining 24 (all in the pharmaceutical industry) were judged based on the FTC’s own internal records of oversight of the relevant markets. The FTC argued that use of questionnaires and reliance on its own records were appropriate given its expertise in crafting remedies in, and close monitoring of, certain industries. The failure to use a consistent methodology throughout the study, however, is problematic since it limits sample size, prevents comparisons, and risks selection issues.
Second, none of the methods constituted objective state-of-the-art techniques for policy evaluation. Questionnaires tend to produce formalistic responses to standardized queries. Records-based assessments can identify only whatever was previously recorded by agency personnel themselves. Even case studies—while commonly used in the past—are inferior to standard statistical techniques, such as difference-in-differences, for policy evaluation. The difference-in-differences technique relies on actual data, better controls for other possible causes, and allows for tests of statistical significance. For these reasons, it had become standard in ex‑post reviews of merger policy.
Third, the FTC study inexplicably defined “success” in three altogether different ways, one for each of the remedy groups. For the 50 case studies, the criterion for success was whether competition persisted or returned within three years; if it took longer than three years, it was termed a “qualified success.” For the 15 remedies assessed via questionnaires, the criterion was only whether the divested product continued in production. For the remaining 24 remedies, it was whether both the divested and the original pharma products were still being produced. (For products in the development stage, the remedy was said to be successful based simply on whether the assets designated for divestiture were in fact transferred as ordered to the buyer.)
This use of multiple definitions of success is itself unacceptable, especially since two of the definitions are inconsistent with the agency’s own stated objective for remedies, namely, the preservation or restoration of competition. As noted before, continued production is a necessary but not sufficient condition for achieving the actual purpose of a remedy. Mere asset transfer does not even measure viability, much less competition, and should not be represented as suggesting anything remotely like a successful divestiture.
Fourth, even with this inconsistent study methodology and varied criteria for “success,” the FTC study reported results that were in fact considerably more mixed than advertised. For the 50 remedies evaluated by case studies, the FTC reported successful preservation of competition within three years for 69% of cases. Put differently, of course, that implies that nearly one-third were failures for as much as three years. By the lower bar of mere continued production (which is not the right test), 91% of the remedies assessed by questionnaires were deemed successful, as were 75% of the pharma cases that were judged successful based on either continued production or simply asset transfer as the criterion. It is hardly surprising that the rate of stated “success” is higher when the bar for “success” is lowered, but that method of reporting casts no real light on the effectiveness of these divestitures.
These disparate methodologies and varying standards in the FTC study caused concern about its soundness, its findings, and its interpretation of those findings. One FTC Commissioner, after conducting his own evaluation of the study, issued a statement declaring the study to be “neither rigorous nor reliable” and advised that “it should not be cited by the public.” He went on to urge the full Commission to determine “whether additional information should be released regarding the study, whether the entire study should be subjected to a more complete independent review, or whether it should be rescinded altogether.”
This very public criticism by one commissioner of the agency’s own 2017 study underscored the extent of doubt about its conclusions. This review shows that a more accurate statement of conclusions is that for just over half of the remedies, simple non-quantitative case studies found that divestitures restored competition within three years in about two-thirds of the cases. For the other remedies, nearly half of all observations, no investigation of competitive effects was conducted.
Other research on divestiture policy in the U.S. has generally come to similarly cautionary—sometimes more negative—conclusions. The studies vary in methodology and scope, numbers of observations, and other important details, but they do cast further light on the key question of the effectiveness of divestitures.
My own study compiled data from all published merger retrospectives, a total of 60 individually studied mergers in 16 industries. All of these mergers were investigated and cleared, some with remedies, by either the FTC or DOJ. The first part of this study examined whether these mergers, overall and after agency review and possible remedy, in fact achieved their objective of eliminating any anticompetitive effects. The test involved difference-in-difference analysis—as discussed above, a statistical method for isolating the effects of the mergers. It found that the mergers on average resulted in price increase of 7.2%, with more than 80% of mergers having higher prices.
The next step was to determine whether those mergers subject to remedies resulted in smaller (or no) price increases, relative to those cleared outright. This further comparison could be made for 42 of the mergers. Twenty-seven of these cases, or 64%, were cleared outright, that is, without any remedy. They resulted in an average price increase of 6.1%, while the remaining 15 mergers subject to any type of remedy were also associated with a substantial price increase, averaging 7.7%. These price increases were insignificantly different from each other, implying that remedies overall failed to improve on the anticompetitive outcome found for mergers that were cleared outright without any remedy at all.
The third and final disaggregation examined the effectiveness of divestiture remedies versus conduct remedies. Divestitures were found to have resulted in an average price increase of 7.1%—again essentially identical to the average for those cleared outright. Conduct remedies were associated with much larger price increases, averaging 13.4%, but this result was based on only two or three observations, so that no real test of statistical reliability was possible.
Focusing on divestitures, these results implied that structural remedies are not as straightforward and successful as often stated. To the contrary, the data suggest that in practice they generally failed in their mission of preserving or restoring competition since statistical tests found outcomes little different from mergers cleared without remedies. In addition to these substantive findings, the Kwoka analysis is notable for its methodology. It relied exclusively on published studies using difference in differences analysis, which, as noted before, is statistically sounder than the case study method.
In a more recent 2023 study, Tommaso Valletti and I analyze all FTC and DOJ investigations of consummated mergers between 2001 and 2023. There were 51 such investigations, of which 49 resulted in some form of remedial action, almost all structural. These cases had some notable features. Often the remedial actions were taken soon after the consummation of the merger, suggesting that the investigations may well have been underway even as the parties proceeded to merge. This short time frame also implies that a lesser degree of integration (if any) had likely occurred in the period after consummation.
The original database recorded various facts about each merger investigation and remedy from publicly available agency documents at the time the matter concluded. In order to be able to assess the remedy, we undertook further investigation of the cases in their post-remedy period and recorded key information about the divested assets, their specific identity, the entity to which they were divested (original owner, third party, new entity), and the viability of the assets in a period of two to three years after divestiture. Viability was defined as continued operation at a level substantially similar to the pre-divestiture/pre-merger level. As previously emphasized, viability is a necessary but not a sufficient condition for a successful remedy, so that any remedy that does not achieve viability necessarily has failed to protect competition.
The findings of this study were striking. Divested assets were “probably or likely viable” for at least two to three years in only 56% of cases. That is, in nearly half the cases, divested assets did not remain viable for as much as three years. Since the asset viability rate is a lower bound on the rate of failure of remedies to restore competition, this implies a failure rate of remedies greater than 44%. Interestingly, this outcome is broadly consistent with evidence about viability from the FTC studies, properly interpreted.
Apart from these independent studies of the U.S. experience with divestitures, there are some notable studies conducted elsewhere. These include studies by the EU, the UK competition authorities, and the Canadian Competition Bureau. With due allowance for significant institutional and practical differences, none specifically tested for the success of the remedies in preserving or restoring competition. The focus of each was on persistent operation of the assets—what here has been termed “viability”—and all have relied on some combination of interviews and case filings. All have reported significant shortfalls in viability of the divested assets or operations, a result consistent with the U.S. experience.
As a method for resolving mergers, divestitures have a reputation for practicality and certainty. In contrast to conduct remedies, this reputation is not altogether undeserved, but the correct policy question concerns their rate of success in preserving competition compared with blocking anticompetitive mergers outright. In some cases, with sufficient attention to supportive factors, a divestiture might prove successful—for example, the canonical case of a small number of overlapping but distinct assets (locations or products) between two firms intending to merge. But divestitures have come to be used in a considerably wider range of settings, many of which are different and seem less promising.
As discussed above, a very substantial fraction of divestiture remedies were outright failures—they did not preserve or restore competition in the affected markets. The case study method has twice reported that 30% to 35% of such remedies do not even result in viable assets over a “reasonable” period of time, and this fraction has remained remarkably steady over time. This implies a higher, perhaps a considerably higher, rate of failure to preserve competition. The one empirical study using difference-in-difference analysis confirms a much higher rate at which post-merger prices rise. Only by misrepresenting “success” have any studies come to more favorable conclusions.
These results leave little doubt that a large fraction of mergers that are deemed to be anticompetitive ex ante have been dealt with by divestitures that have failed to prevent the behavior that raised initial concern. The reasons are not difficult to identify. It is often impossible for non-expert agency personnel to adequately understand the operational and financial considerations that come into play when altering the structure and behavior of the merging firms. It is also important to recognize that the merging firms are not partners of the agency in this process. Their incentives are to secure the weakest possible remedy and then to exploit it after it is in place.
Remarkably, while these issues have been recognized for at least 25 years, efforts to improve divestitures or avoid altogether circumstances where they are likely ineffective have not made much difference, since divestitures have continued to experience high and unchanged failure rates. This body of evidence and experience has led to the change in remedy policy at the FTC and the DOJ that was noted at the outset. The current policy change is rooted in the evidence and is the logical outgrowth of experience. Whether this policy change survives the new administration is uncertain, but it is the hope of many companies that it does not. The president of Live Nation was recently quoted as expressing the hope “that we’ll see a return to the more traditional antitrust approach, where the agencies have generally tried to find ways to solve problems they see with targeted remedies that minimize government intervention.”
If divestitures are to be used at all, it is essential to identify correctly the specific circumstances where they have a very high probability of true success. That can only be done with studies that go beyond existing efforts. While the latter have emphasized the issues of assets and buyers, the next steps should be to test empirically additional industry, product, financing, timing, and other conditions that are related to better outcomes.
It is also important that any such studies be conducted with appropriate standards and sound methodology. The most careful and objective approach is difference-in-differences, whereby the effect of the remedy on price can be isolated from other factors possibly causing the change in price change before and after the merger. Difference-in-differences allows for statistical testing of the direction and magnitude of measured effects, assuming that there are enough observations to conduct such testing.
Case studies are not without their value, especially if they are used as a supplement to, and not a substitute for, statistical testing. They can be useful in capturing the nuances of individual experiences in ways that statistical testing cannot. But their limitations—issues of selection and causation, difficulties in conducting statistical testing, other possible ambiguities—make case studies by themselves an inadequate method for determining whether competition has been preserved by a remedy.
In short, methodology matters, and matters a lot. Weaker analytical techniques and inexcusably lower bars for “success” produce higher rates of apparent—but not real—success in preserving competition. Failure rates of 30% or 40%, or possibly more, indicate a policy that is neither as straightforward nor as reliable as often described—one that cries out for a detailed and well-designed study of divestiture remedies. The result of such a study would undoubtedly be a more careful remedy policy than common in the past, but one with great benefits to consumers and the competitive process.
