In new research, Alejandro Herrera-Caicedo, Jessica Jeffers, and Elena Prager find that firms that share a C-suite executive or board director are much more likely to collude.
Communication facilitates coordination. For a long time, overlapping corporate boards and other shared leadership structures between companies were praised as tools to facilitate the flow of information across firms. Information flow can be good for markets when it spreads best practices and yields more efficient matching, such as suppliers to retailers. But it can also hurt markets. When firms coordinate instead of competing, they tend to raise prices and reduce the quantity and variety of available products and services, to the detriment of the public. Under certain circumstances, this is considered collusion and violates antitrust law.
For over a century, United States antitrust law has recognized that the risk of collusion may increase when representatives of one firm serve in key roles at a competing firm. Section 8 of the Clayton Act explicitly prohibits two competing firms from simultaneously sharing the same person in high-level leadership positions. Typically, such “common leaders” take the form of the same person sitting on two boards of directors, or a C-suite executive of one firm sitting on the board of another. The prohibition stems from the concern that common leaders may help align the incentives of the competing firms, making them more likely to want to collude. It may also help them communicate about and enforce collusive agreements, making them more likely to succeed at colluding.
Our research asks a simple question: are these fears justified? In other words, do common leaders really facilitate collusion? To answer this question, we studied the largest known modern case of collusion in the U.S. labor market, which involved several dozen firms, mostly in the tech industry, that agreed not to poach each other’s employees. In this sample, the arrival of a common leader raised the probability of subsequent collusion by a factor of nine (up 11 percentage points from a baseline of 1.2 percent). Although the effect is likely smaller in the economy as a whole, this finding demonstrates that legislators had good reason to fear common leaders.
Increasing attention to common leadership
To grapple with the potential scope of the problem, it helps to know that common leaders are, well, common. Among publicly traded companies, more than one third share a common leader with at least one other publicly traded company—and that doesn’t include leaders shared with privately held companies we cannot readily observe in data. Moreover, a disproportionate share of common leadership links are to companies in the same industry.
U.S. antitrust law prohibits such common leadership, but the existence of a prohibition does not necessarily imply its enforcement. After its passage in 1914, Clayton Act Section 8’s prohibition of common leaders lay nearly dormant for over a century. But since 2022, federal antitrust agencies have taken up its enforcement with vigor. The Department of Justice (DOJ) Antitrust Division and the Federal Trade Commission (FTC) have launched investigations subsequently connected to the resignations of members of the boards of directors of dozens of firms, including well-known firms such as Pinterest and Warner Bros.
These actions are part of a broader shift toward more aggressive antitrust enforcement in the last decade. However, antitrust enforcement resources are scarce, so it is important that they be targeted effectively. Whether targeting common leaders would reduce collusion was, until our work, an open question. Firms have many other avenues for coordinating their actions, so the presence or absence of common leaders may be irrelevant. Enforcing against common leadership would then be ineffective. It would also place unnecessary restrictions on the leadership talent pool. Good enforcement therefore requires an understanding of whether common leaders are, on average, indicative of collusion.
Measuring the link between common leaders and collusion
Studying the relationship between common leaders and collusion requires overcoming a data roadblock: Collusion is illegal. Colluders try to hide their actions. This makes collusion very difficult for researchers to measure.
Our study used unsealed court records from the largest known modern case of labor market collusion in the U.S.: the Silicon Valley no-poaching case. In the mid-2000s, 43 firms entered into a series of agreements not to recruit one another’s employees. The no-poaching agreements barred firms from cold-calling each other’s employees, then (and now) a key recruitment channel for software engineers. Most of these firms were in tech, but not all. Apple, one of the firms at the epicenter of the agreements, had agreements with Google, Intel, and Nvidia, but also Nike (sports apparel) and Genentech (biotech). Refusing to compete aggressively for talent by effectively averting bidding wars is a violation of antitrust laws. But until recently, the DOJ and FTC rarely enforced antitrust laws in labor market contexts, and the colluding firms evidently concluded that the legal risk was worth taking.
This may have been a bad bet, because in 2010, the DOJ brought suit against eight of the involved firms. Several civil lawsuits followed, implicating additional firms. By the time the legal dust settled, 43 firms were revealed to have been involved when the presiding judge released an unusually detailed trove of evidence from multiple court cases. Our study uses the released evidence to measure which pairs of firms had collusive no-poaching agreements and which did not. The resulting firm-pair panel allows us to investigate the relationship between collusion and common leadership.
The onset of common leadership increases the probability of collusion
We estimate that relationship using difference-in-differences and event study methods, comparing pairs of firms that recently began to share a common leader against other pairs of firms that did not. Moreover, we can control for factors that might drive an individual firm to collude more or less at a given point in time, such as new product launches that make collusion difficult to sustain or the presence of a risk-tolerant CEO, by including firm-year fixed effects. In what we view as our most compelling analysis, we also control for any factors that might drive collusion between a pair of firms, such as the similarity of their products or target markets, by including firm-pair fixed effects.
We find that the onset of common leadership raises the probability of a collusive agreement by 11 percentage points. The baseline rate of collusion in the absence of common leaders in our sample is 1.2 percent, so this represents a ninefold increase in the probability of collusion. For an outcome that is normally difficult to detect due to its secrecy, the magnitude of this predictive relationship is very large.
Generalizing beyond one case study
The court documents released from the Silicon Valley no-poaching cases are a treasure trove of information of the sort rarely available to researchers. They are what allow us to construct measures of secretive, illegal agreements at the granular level of firm pairs. Without that, we would not have a clean test of whether common leadership actually increases collusion.
Our sample consists of over 900 pairs of firms constructed from all possible pairs of firms implicated in the court documents. For example, if Pixar colluded with Apple (which it did) but not with Google (which it did not), then Pixar-Apple and Pixar-Google will both be in our sample, but only the first will be marked as having colluded. In this way, we can focus on pairs that consist only of firms that faced legal scrutiny. This minimizes the risk that we include in the sample a pair of firms that did collude, but we cannot know that they colluded. For example, Facebook is not implicated in any of the court documents, and as far as we know, it was not investigated. Even if Facebook and Pixar had an agreement, we likely would not know about it.
The resulting clean measure comes at a cost: We can only study a particular set of firms operating in a particular environment at a particular time. Each of these firms is sufficiently risk-tolerant to engage in some illegal activity; perhaps the average firm is more risk-averse and less likely to collude, even in the presence of common leaders. The enforcement environment has also shifted since our study period, likely making firms more acutely aware of the legal risks of collusion. It is entirely possible, even likely, that the relationship between common leadership and collusion in the broader, modern economy is smaller than an 11 percentage point effect.
What, then, can we take away from this case study? Enforcing Clayton Act Section 8’s prohibition on common leadership should reduce collusion. We now know that when firms have unfettered access to common leadership, as they did during our sample period, some of them will use it to collude. Some amount of collusion would, of course, take place regardless. But the fact that firms choose to collude through common leadership when that path is available should tell us that common leadership is an easier and/or more effective path to collusion than others.
Authors’ Disclosures: The authors report no conflicts of interest. You can read our disclosure policy here.
Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.
Subscribe here for ProMarket’s weekly newsletter, Special Interest, to stay up to date on ProMarket’s coverage of the political economy and other content from the Stigler Center.
