Others have leveled both theoretical and empirical critiques at these studies and the proposed antitrust policy changes associated with them. A recent paper by Merritt B. Fox and Menesh S. Patel falls directly into this category. At the outset, Fox and Patel cite data (at 4 and note 2) that “in a typical industry, each of the Big Three [institutional investors Vanguard, BlackRock, and State Street] holds approximately 4-9% of the shares of every one of the industry’s constituent publicly traded firms. In other words, for each firm in an industry, a meaningful portion of the firm’s share is controlled by entities that concurrently hold shares in all the firm’s relevant competitors, a pattern ordinarily referred to as ‘common ownership’.”
The paper’s key focus is on the extent to which the firm’s managers account for the holdings of these common owners when making pricing and output decisions—a principal/agent problem. In particular, the paper focuses on frictions that can arise between principals (the common owners) who prefer a particular managerial response and the manager as the agent of the principals in implementing that response along with her own set of objectives. Fox and Patel note (at 7) that “it is decisions by firm managers—not shareholders—that in the first instance determine the firm behavior whose interaction results in an industry’s level of competition.”
Fox and Patel conclude that, among other reasons, those frictions are so substantial (but generally ignored in the literature) and the common ownership shares are currently sufficiently small that the anticompetitive concerns raised by common ownership are misplaced. Instead, Fox and Patel argue that the firm’s managers would choose to maximize the stand-alone profits of the firm without accounting for the effects of its actions on the other firms commonly owned by the investor. In particular, Fox and Patel (at 1) note that today roughly 79% of all S&P 500 shareholders are non-common owners and so dominate shareholdings in these firms. The paper finds no reliable conceptual or empirical support for concluding that there are unilateral adverse competitive effects arising from common ownership.
Theoretical Basis for Common Ownership Concerns
Fox and Patel motivate their analysis by first using a straightforward example of the effects of common ownership based on the workhorse Cournot model of firm interaction, one that has often been used in assessing the possible anticompetitive effects of common ownership. In a perfectly competitive model, each firm takes price as given and as unaffected by its output choices. In the Cournot model (and in the absence of common ownership and any managerial frictions), output choices matter for prices—the firm does not take price as given but rather chooses an output that will maximize its profits after accounting for the output choices of its rivals. Roughly speaking, each firm is aware that the output responses of its rivals to the firm’s own output choices affect the firm’s own profits and so take those rival responses into account when choosing its profit-maximizing output level.
In this model, if the firm were to lower its output level (and so increase prices), rivals would respond to these higher prices by increasing their output (and so lowering price). Roughly speaking, this will result in the firm increasing its output level because its lower output level is no longer as profitable in light of the rival responses. When each firm accounts for rival responses in their output decisions (in contrast to a perfectly competitive firm), the resulting equilibrium is characterized by an industry output level that is less than the competitive level (and so prices are higher than the competitive level) but still greater than the monopoly level of output. Note that in this model, there is no explicit or direct communication among the industry rivals. None is needed. Each firm is assumed to be aware of and to account for the reactions of its rivals in making its profit maximizing output choices. That awareness alone leads to the less-than-competitive outcome.
Fox and Patel then extend this model to illustrate how common ownership alters that Cournot outcome. They note (at 21) that:
The common ownership literature has a single key conclusion: common ownership in an oligopolistic industry diminishes managerial incentives to compete, even in the absence of collusion or communication . . . [When] the managers of a firm with common owners set their output level, they take account of the effect of that decision on the net revenues of their rival firms. This substitutes for the standard model’s assumption that each firm’s managers seek to maximize only its own firm’s net revenues. (Notes omitted.)
Common shareholders would prefer that each firm in which they have an ownership interest accounts for the effect of the firm’s output choices on the profits of each of the other rival firms in which they have an ownership interest. In these models, each firm’s managers are assumed to maximize own-firm profits after accounting for the effects on rivals’ profits, giving weight to the holdings of each common owner reflecting the common owner’s control or influence over output decisions. For simplicity, Fox and Patel (at 23) frame this in terms of a single “blended shareholder” in which the managers are assumed to maximize the value of the weighted average of the common owners’ shareholdings.
Thus, in these Cournot-based models, the managers of each firm are assumed to maximize the value of the blended shareholders’ portfolio. That is, the firm accounts for the effects of its output decision not only on the profits of the firm itself but also on the profits of rivals that are included in the blended portfolio. When the firm increases its output, that increase has the effect of depressing the market price, reducing the profits of its rivals and so harming the value of the blended portfolio. With common ownership, those effects on the rivals’ profits are now accounted for in the firm’s profit calculus. As a result, each firm produces less output and charges higher prices than in the standard Cournot model. And as with the standard Cournot model, the model’s assumption is that this outcome can be attained without any explicit collusion or direct communication with the firm’s rivals included in the blended shareholder’s portfolio.
Fox and Patel’s Criticisms of the Cournot Foundations of the Model
But Fox and Patel question the basis for that assumption. First, their paper highlights the omission of the preferences of non-common owners. In particular, Fox and Patel argue that the non-common owners, accounting for 79% of shareholdings, would prefer that the firm maximize its own profits as in the standard Cournot model rather than the maximize the value of the blended shareholder’s portfolio. They note (at 32) that at the equilibrium arising from the assumption that each firm maximizes the wealth of the blended shareholder, the
non-common shareholders would be better off if their own firm takes advantage of this opportunity to increase market share and has a level of output that actually is in excess of what it would be in the total absence of common ownership. (Note omitted.)
To be sure, acting in concert, the non-common shareholders of all rivals are better off if the firm managers account for the holdings of the blended shareholder. But acting alone, the non-common shareholders of each firm would view themselves as better off if the firm managers expanded output beyond that arising from a sole focus on the blended shareholder. Thus, absent some coordination among the non-common owners of all the rival firms, the managers confront an apparent conflict between the objectives of the common owners and those of the non-common owners.
Fox and Patel (at 22) contend that the use of the Cournot model as the backbone for evaluating the competitive effects of common ownership “glosses over this conflict by assuming, without serious exploration, that managers will make output decisions based on some kind of averaging of these differing interests.” Indeed, the paper highlights observations by O’Brien and Salop in their seminal paper which are worth repeating here:
[W]here the owners have conflicting views on the best strategy to pursue, the question arises as to how the objective of the manager is determined. Ultimately, the answer turns on the corporate-control structure of the firm, which determines each shareholder’s influence over decision making within the firm.
Can Common Owners Create Incentives for Management to Abide by the Preferences of the Common Owners?
The differences between the preferences of managers and those of shareholders, i.e., the principal-agent frictions, are key to understanding why Fox and Patel contend that common ownership is not likely to explain the adverse competitive effects found empirically. In particular (and without any regard to common ownership), Fox and Patel (at 38) note that the managers’ position in the firm may provide benefits valued by managers: “compensation, prerequisites, power, prestige, the pleasure of benefitting their associates in the firm, a sense of doing social good, and so on.” But the ability to attain these benefits depends on the profits the firm makes. The greater those profits, the more that managers may be able to engage in this kind of behavior. The profits earned by the firm’s rivals are irrelevant to the firm’s managers. Those profits cannot be used to “subsidize” managerial preferences for those benefits. So, at the outset, Fox and Patel argue that the firm managers would maximize each firm’s own profits and, in that way, the managers’ output preferences and those of the large pool of non-common owners can be more aligned.
The common owners, of course, would prefer instead that management respond to their strategy of considering the rivals’ profits in the manager’s output decision. Given these differences in preferences between the managers and the common owners, Fox and Patel (at 40) consider whether the common owners can use a “sticks and carrots” managerial incentive structure to persuade management to account for all the shareholding interests of the common owners.
These “sticks” include proxy fights (which can result in a new management team), a hostile tender offer for enough shares to replace existing management, the threat to sell sufficient shares to depress the share prices, investor activism to pressure management, and the threat of a suit claiming the existing management team is not upholding its fiduciary obligations to act in the best interests of the corporation. An important carrot could be changes in the manager’s compensation package to incentivize cooperation with the common owners’ objectives.
Fox and Patel conclude that these sticks and carrots would not be sufficient to alter management’s focus on own-firm profitability. Moreover, the paper argues that many of the channels the common owners might use to align the preferences of management and those of the common owners (for example, compensation changes discussed below) are not practical in the real world.
Fox example, Fox and Patel argue that a tender offer or a threat of such an offer to impanel a more sympathetic management slate would be unlikely. They argue (at 45) that a tender offer by a common owner would be very costly, given “the tender offer’s considerable transactions costs plus the share price premium needed to acquire enough of the target’s shares” to succeed. Against that background, they conclude that such an offer is likely to succeed only at common ownership levels considerably larger than current holdings.
Fox and Patel similarly argue that proxy fights, sales of share blocks by the common owners, or lawsuits by common owners alleging that the managers are violating their fiduciary obligations are equally unlikely to succeed in convincing management to curtail output.
On available “carrots”, the paper (at 52) dismisses the claim that common ownership could result in compensation packages that are less tied to the firm’s own profitability and so could encourage the managers to soften competition. In particular, Fox and Patel observe that these common owners “have no means of directly dictating executive compensation, as the board, not shareholders, sets executive compensation.”
Fox and Patel conclude that neither the sticks nor the carrots are likely to be successfully used to coax management to account for all the shareholding interests of the common owners for a different reason. Common owners themselves may have divergent interests and would not agree on how the managers should account for the interests in rival firms. In the prior analysis, Fox and Patel typically assume that each common owner has the same interests in the rival firms, but that is unlikely to be the case. As they note (at 42), the percentage ownership interest by common owners in rival firms differs across industries and across firms.
As a result, while common owners would prefer an output level lower than the Cournot outcome, those owners may have different strategies for accounting for the rivals’ profits. Depending on the ownership configuration, one common owner may find it more profitable if output is restricted in one firm more than that which would be preferred by others. Thus, heterogeneity in the common owner’s shareholdings in rivals may result in disagreements over their price-raising strategy in fashioning dictates to management.
In short, Fox and Patel argue that the assumption that managers will account for all the shareholding interests in rival firms is unrealistic. They argue that much of the existing analysis assumes that it is reasonable to characterize managers as implementing a strategy that will serve the interests of the common owners through a reduction in the intensity of competition. But that approach ignores conflicts between the goals of the common owners and the goals of managers, the conflicts between common owners and non-common owners, and indeed, the conflicts among the common owners themselves.
Given those views, Fox and Patel (at 56) conclude that the output levels chosen by the firms will be a standard Cournot outcome even in the presence of common ownership given those conflicts, the large shareholdings of non-common owners, and the manager’s preferences for “perks” generated by own-firm profitability.
Appraising the Empirical Findings of Adverse Competitive Effects Arising
from Common Ownership
Against that backdrop, Fox and Patel consider the empirical analyses of the effects of common ownership on competition. At the outset, Fox and Patel (at 62) clearly articulate their criteria for evaluating the empirical literature:
All else equal, where two bodies of empirical work respectively support opposing hypotheses, but one hypothesis is the more plausible of the two, the work supporting the more plausible hypothesis is more likely to be the correct one. Our analysis suggests that the hypothesis that common ownership at current levels reduces competition is highly implausible. The more implausible a hypothesis, again all else equal, the more likely that results in a study purporting to support the hypothesis, though consistent with the hypothesis, are in fact due to something else. (Note omitted.)
Specifically, there have been a number of studies that have failed to find a statistically meaningful effect of common ownership on increasing prices. These are studies that have considered such effects in industries other than banking and airlines as well as studies that re-examine the initial papers finding anticompetitive effects from common ownership. On the latter in particular, the contrary studies have used different data sources and different methodologies from those used in the original papers and found no anticompetitive effects in either airlines or banking.
Fox and Patel conclude that the underlying failure of these other studies to find any anticompetitive effects from common ownership is the expected outcome, given the manager’s incentives to maximize own-firm profitability rather than the returns to a blended shareholder. Thus, Fox and Patel conclude (at 63) that “our analysis showing the implausibility of the common ownership literature’s hypothesis of common ownership reducing competition itself has affirmative empirical support.”
The Efficacy of Communication Channels between Common Owners and Management
Fox and Patel (at 63-64) also address the communication “mechanism” question. The modeling of the competitive effects of common ownership assumes that the firm manager will maximize the value of the common owners’ portfolios. Under that assumption, there need not be any coordination between the portfolio owners and managers. Managers are assumed to behave that way. But if that assumption is incorrect, the literature considers the question of how the common owners transmit their preferences to managers and whether those transmission mechanisms are sufficient to convince managers to manage in the interests of the common owners. As noted above, however, Fox and Patel argue mechanisms such as proxy fights and tender offers are unlikely to be effective communication channels between common owners and the firm’s managers.
One mechanism discussed in greater detail (and dismissed by Fox and Patel) is the possibility that common owners could have sufficient influence to alter the management compensation package to make the manager’s compensation less sensitive to own-firm performance. As Fox and Patel explain (at 67), the hoped-for effect would be to cause “managers to become less incentivized to engage in conduct that improves firm productivity and in turn causes prices to rise.” (Note omitted.) Fox and Patel note the analyses of this mechanism assume counterfactually that the common shareholders can directly dictate the terms of the compensation package, which they cannot.
Further, the empirical findings that common ownership is associated with compensation structures less sensitive to own-firm profitability are not plausible, according to Fox and Patel (at 68), because of the lack of a “plausible mechanism connecting common ownership at current levels with the sensitivity of managerial compensation to firm profitability. . . .” So, Fox and Patel (at 68) conclude that any empirical relationship between common ownership and the sensitivity of managerial compensation to own-firm profitability is likely spurious and not causal, given the differing preferences of common owners, non-common owners, and managers.
The Validity of the Metric Measuring Common Ownership’s Effects on Competition
One other notable criticism of the empirical analysis of common ownership is the particular metric used in many of the studies to account for the firm’s accounting of the interests of common owners. Described in more detail by Fox and Patel (at 80-82), that metric is the Modified Herfindahl-Hirschman Index or MHHI. This in turn can be decomposed into the ordinary HHI plus the MHHI Delta (the difference between the MHHI and the HHI) where the MHHI Delta reflects the effect of common ownership in the competition metric. Thus, when using only the HHI as a structural gauge of competition, the HHI could understate the degree of concentration by not accounting for the MHHI Delta.
A key concern when using the MHHI in a statistical analysis is that it can generate biased results. The MHHI is calculated (in part) by using the market shares of the firms. As Fox and Patel highlight, the MHHI can increase without any change in common ownership. They note in particular (at 83) that an increase in demand can lead to higher prices and a higher MHHI “without there being any causal relationship between [common ownership and prices]”. Further, as discussed above, Fox and Patel dismiss the underlying use of the “blended shareholder” assumption on which the MHHI is based, meaning that in their view, any analysis relying on the MHHI is simply misplaced.
Fox and Patel provide an insightful analysis of the frictions that can obstruct efforts by common owners to reduce market competition. The paper provides a detailed evaluation (with copious cites to the literature) of the principal-agent issues that work against the manager’s maximization of the value of the common owners’ portfolio and explains why those frictions should be accounted for in the competitive effects analysis.
Further, the paper provides a useful discussion of mechanisms (again with copious literature cites) by which common owners might attempt to communicate to management a policy that accounts for the effect of management decisions on the profits of the firm’s rivals. Based on the literature, Fox and Patel conclude that those mechanisms are likely insufficient to convince managers to fully consider the interests of the common owners. Indeed, Fox and Patel argue that many of the channels that are described as available to common owners are not in fact available to them. As a consequence, Fox and Patel conclude that, at current levels of common ownership, the default analysis should be one that assumes managers focus solely on the profits of their firm and their own managerial objectives, not the profits of their firm’s rivals.
Fox and Patel may have understated the extent of common ownership.
A key to the analysis of Fox and Patel is the presumed influence of the much larger group of non-common owners over managerial behavior. Fox and Patel argue that, with a large group of non-common owners, managers would choose strategies that will serve the interests of this dominant group of non-common owners (who prefer maximization of own-firm profits), given an inability to reconcile the different objectives of the common owners and non-common owners.
In their illustrations, Fox and Patel focus on Vanguard, BlackRock, and State Street because their holdings across rival firms are more complete than other investors and (at 40) “because they have been the primary subject of academic and other discussions of the common ownership issue.” However, the pervasiveness of institutional holdings beyond the “Big Three” is substantial.
For example, Einer Elhauge observes (at 1267) that, “from 2013 to 2015, seven shareholders who controlled 60.0% of United Airlines also controlled big chunks of United’s major rivals, including 27.5% of Delta Airlines, 27.3% of JetBlue Airlines, and 23.3% of Southwest Airlines.” More generally, one of the papers that gave rise to this new strain of research found that institutional investors account for 70% to 80% of all the equity of publicly traded firms.
In the airline example highlighted by Fox and Patel, the Big Three account for about 19% of the equity of United and Delta and about 18% of American Airlines. But the data provided by Fox and Patel (at 53 and note 110) indicate that Primecap has a greater share than BlackRock and State Street in United and American Airlines and a comparable share of United as Vanguard. Yet, Fox and Patel consider Primecap to be a non-common owner. More generally, if these other investors share the same interest as the Big Three in restraining competition, then any constraint on management’s maximization of own-firm profits imposed by so-called non-common owners may be reduced.
Understanding where the line should be drawn between common and non-common owners would seem significant for both assessing constraints imposed on management by these owners to satisfy the preferences of common owners and any antitrust monitoring of the expansion of common owners. After all, Fox and Patel repeatedly note that their analysis may apply to the current state of the world with what Fox and Patel apparently regard as a relatively small common-owner share. But if the extent of competitively-significant common ownership is greater than that identified by Fox and Patel, it would be useful to identify the relevant set of common owners (by industry, perhaps) so as to trigger some form of antitrust review if the common-ownership share does become significant for antitrust purposes.
Fox and Patel do not opine on what such a threshold should be. But if Fox and Patel have defined the scope of common owners too narrowly, a more appropriate definition (if required) might indicate we have already exceeded such a threshold.
Differing preferred strategies among the Big Three.
As discussed above, Fox and Patel argue that investors outside of the Big Three will have different incentives from the Big Three and so Fox and Patel argue that management will treat them as non-common owners. If that were sufficient to regard these other institutional investors as non-common owners, then Fox and Patel recognize that the same logic could apply to the Big Three itself. Fox and Patel (at note 110) observe that those three investors (Vanguard, BlackRock, and State Street) have different ownership interests in rivals and management might treat these common owners as non-common owners if those divergent interests cannot be reconciled. In that case,
the common owners will be even less likely to coalesce around an agreed competition strategy since, for every firm in the industry, the common owners likely will have differing preferences about the manner and extent to which that firm should separately compete with each of the firm’s rivals.
Thus, in its most basic form, as long as institutional investors have differing preferences on output levels by the firms in the market, managers will maximize own-firm profits. Fox and Patel argue that, absent collusion among the investors, common ownership has no effect on output choices of the firm at current levels of common ownership. In her paper, Tecu (at 57) made a similar but more cautious observation: “. . . real world costs of implementing the anticompetitive incentives created by common ownership could be so high that one might not expect them to have any meaningful or measurable effects.”
Fox and Patel’s critique of evidence supporting a finding of adverse competitive effects from common ownership.
Of course, Fox and Patel do not rely on mere conjecture. Fox and Patel place substantial weight on studies that cannot reject the null hypotheses of no competitive effect from common ownership. And against the background of that evidence and the dismissal of studies with contrary results, Fox and Patel (at 62) opine (as previously noted) that:
All else equal, where two bodies of empirical work respectively support opposing hypotheses, but one hypothesis is the more plausible of the two, the work supporting the more plausible hypothesis is more likely to be the correct one.
One might be concerned that Fox and Patel may be too tilted toward that finding of no effect. If there is evidence that supports alternative hypotheses, one might expect that, given the underlying theory, the issue is whether one set of empirical analyses clearly dominates the other. While one theory may be viewed by some as more plausible and by others as less plausible than a competing theory, only empirical analysis can distinguish between the two (at least with some underlying probability).
To be sure, Fox and Patel argue that the literature and empirical analyses consistent with a finding of adverse effects of common ownership are flawed and more weight should be attached to the contrary studies. The results of these kinds of analyses are conditional on the model itself and the data used. Thus, the models and/or the data generating results both supportive and not supportive of the adverse effects must be assessed for validity. In the midst of that disagreement, rejecting a finding of adverse competitive effects because in principle it is inconsistent with the managers’ output preferences seems too swift a judgment. After all, given repeated findings of adverse effects of common ownership, one might not reject the underlying model out of hand. Similarly, one might not reject the findings of no competitive effect out of hand, given the results from other papers and the obstacles to aligning the differing incentives of common owners and management as discussed by Fox and Patel.
Nonetheless, while Fox and Patel’s key conclusion that the findings of an adverse competitive effect are conceptually and empirically flawed, the paper (at 65) does not reject the possibility that common ownership could harm competition at higher levels of common ownership:
Thus, to the extent common ownership is generating any appreciable competitive harm, the causal mechanism . . . must be the product of additional collusion or coordinated conduct among an industry’s firms that is enhanced by the presence of common ownership.
As a result, the paper suggests (at note 230) that, in light of that possibility, “the optimal policy response is for the antitrust agencies to follow a case-by-case approach to common ownership. . . .”
This paper is well worth reading for its exposition of the common-ownership theory, the highlighting of conflicts between common and non-common owners, and the principal-agent friction between managers and common owners, all of which would lead managers to maximize own-firm profits. Indeed, the paper provides an important discussion of the principal-agent frictions in particular that should be addressed by those exploring any adverse effects from common ownership.
This review has not addressed all of the issues raised by Fox and Patel, such as alternatives to the MHHI, how a common owner would be affected by (among others) interests in upstream markets where the owner also has an interest, and whether managerial inertia is a plausible mechanism for lowering output to benefit the common owners. Most importantly, this review has focused on the conceptual and empirical foundations of the common ownership theory and its critics, and did so in an abbreviated fashion. This review has not done justice to the complete analysis of Fox and Patel and so urges readers to absorb all of this analysis.
It comes as no surprise that Fox and Patel urge against the adoption of any general enforcement policy (as opposed to a case-by-case approach) against common owners. The paper nonetheless provides a detailed review of proposals to limit the influence of common owners and the costs of such policies, which is worth the read.
Antitrust scholars and practitioners should take note of potential shortcomings of the common ownership theory and empirics addressed by Fox and Patel. The extent to which principal-agent conflicts can diminish the likelihood of adverse effects of common ownership should be a necessary component of the analysis. But simply rejecting opposing empirical results based on what Fox and Patel regard as an implausible theory may be jumping the gun, given the empirical support for adverse competitive findings and the underlying theory. Future research may resolve these differences one way or another. At this point, a reasonable response might be to gather additional evidence before designing an antitrust enforcement program to address common ownership. As Fox and Patel suggest, the costs of such a policy in the absence of any competitive effects could be substantial, and as noted, they urge the agencies to adopt a case-by-case approach in evaluating claims of common-ownership harm by investigating possible coordinated behavior.
Nonetheless, there does seem to be increased momentum at the U.S. Department of Justice and the Federal Trade Commission to account for common ownership in antitrust enforcement policy based on the studies finding adverse competitive effects. Indeed, both antitrust agencies have announced that they are revisiting merger enforcement policy and are soliciting comment on (among other issues) the role common ownership might play in that policy (at 12(h)). The development of an enforcement policy would likely (or hopefully) account for the issues that have been raised here and elsewhere—and will likely also be raised by scholars and practitioners in response to the agencies’ questions. Something to look forward to.