/>iWe interview Professor Matthew Weinberg of The Ohio State University to gain his insights on merger retrospectives: what they are, when they are used, how they are evaluated, and why they continue to evolve. Professor Weinberg is an authority on merger analysis and the economic effects of regulation. His research covers competitive effects, coordinated behavior among competitors, enforcement, and the accuracy of merger simulations.
What are merger retrospectives?
As one might guess from the name, a merger retrospective involves looking at past mergers and assessing their effects. A typical retrospective study analyzes cases that were reviewed by the agencies—the Department of Justice (DOJ) or the Federal Trade Commission (FTC)—but eventually were cleared and consummated. Merger retrospectives examine what actually happened after the merger to determine whether the theories, tools, and techniques used during the review accurately predicted outcomes that can now be measured. As such, lessons from merger retrospectives can inform how experts and competition authorities approach the evaluation of proposed mergers.
Lately there has been significant interest in merger retrospectives, particularly with the FTC looking more closely at the impact of past mergers. What purpose do these studies serve and what can the agencies learn from them?
The FTC is always interested in evaluating the success of the antitrust enforcement process. On the one hand, there is a risk of challenging mergers that would lead to efficiencies that benefit consumers. On the other hand, there is a risk of approving mergers that harm consumers. The agencies’ ability to screen proposed mergers in a way that consistently benefits consumers depends on the accuracy of the theories and techniques used to assess them. A straightforward way to determine the accuracy of these predictions is to contrast them with detailed studies of the aftermath of particular mergers.
The FTC uses retrospective analysis to better understand whether its analytical tools function as intended, and whether certain evaluation criteria should be adjusted. I recently addressed this topic on a panel at the FTC, as part of the agency’s ongoing Hearings Initiative on competition and consumer protection in the twenty-first century.
What techniques are used to study past mergers?
Merger retrospectives estimate what market outcomes would have been if the merger had been blocked, and then compare those hypothetical outcomes to actual observed outcomes. Usually merger retrospectives focus on price, but they can also take into account product quality, innovation, levels of service, and other factors.
The key challenge in estimating the effect of a merger is developing a convincing counterfactual. How would market outcomes have changed had the merger not occurred? In some cases, an approach called “differences-in-differences” analysis can be used to answer this question. With this approach, economists identify a control group of markets or firms not directly affected by the merger. Then, they examine the differences in outcomes between the groups affected by the merger and the control groups. It is also possible to compare the results of merger retrospectives to the predictions made by the agencies’ models prior to the merger.
What have we learned from existing merger retrospectives? What has your research shown about the accuracy of merger simulations?
In 2002, after the agencies had lost seven consecutive hospital merger challenges in court in the preceding years, the FTC conducted a particularly well-known merger retrospective. Looking at several past hospital mergers, the FTC analyzed the transactions’ effects on markets, and discovered several flawed assumptions in the merger review process. For example, contrary to a commonly made argument, the FTC found that nonprofit hospitals and for-profit hospitals responded in a similar manner. In response to these and other findings, the FTC modified its analytic tools.
Another FTC retrospective examined the impact of divestitures. In some circumstances, an agency will allow a merger to proceed, with the condition that the merging parties sell an asset to a third party in order to preserve competition. In 2017, to determine whether this approach was effective, the FTC updated a 1990s study that evaluated past divestiture orders. The newer study concluded that 17 percent of the divestitures reviewed had failed to introduce buyers of the divesting assets to the relevant market as viable competitors. Based on these findings the FTC subsequently defined best practices for divestitures. These include favoring divestitures of full, freestanding business units and requiring that proposed buyers take steps to establish their competitive and financial viability.
I have researched both the effects of consummated mergers and the efficacy of the forecasting tools the agencies typically use to predict these effects. I have looked at mergers across a range of consumer goods, as well as in other industries, and found that the performance of the models varied. Sometimes, they predicted mergers’ actual price effects with reasonable accuracy; sometimes they over- or underestimated them.
The accuracy of merger simulation models often depends on how well a model’s assumptions match the industry’s characteristics. For example, does competition exist primarily over price? Do elements of quality or investments in capacity also play a role? Are prices set by producers or bargained over? In addition to the answers to these questions, much depends on the data available for analysis.
Where do you see the FTC’s retrospective program going? How do you think retrospective analyses will influence merger enforcement over the next five to ten years?
Overall, I would expect evolution, not revolution. The hospital merger and divestiture examples, though they occurred at different times, offer useful cases in point. In both instances, the FTC undertook no radical shift to achieve its end goals. Instead, the agency adjusted its analytical approaches and amended its policy tools. The information gained in the merger retrospective may change the kinds of assumptions that the agencies find plausible going forward, as well as the procompetitive features they consider most relevant.
One area the FTC might consider more deeply is the performance of merger simulations. Recently, both agencies and parties have been deploying sophisticated models to predict the likely effects of mergers on prices. These are powerful models, but they often require simplifying assumptions. I have published several research papers that evaluate the performance of merger simulation models, and I conclude that such simulations reasonably predict post-merger prices in some cases, but fail to do so in others.
There is room for the FTC to learn more about the circumstances in which merger simulation models are most reliable, and how to adjust them in cases that do not conform to standard assumptions.
