12.2.1 How Do Agency Frictions, Informational, and Organizational Complexities Affect How Common Ownership Affects Product Market Outcomes?

At the 2018 FTC Hearings on Common Ownership, FTC Commissioner Noah J. Phillips remarked that ‘… areas of research that I, as an antitrust enforcer, would like to see developed before shifting policy on common ownership [are]: Whether a clear mechanism of harm can be identified …’Footnote ¹² At the same event, SEC Commissioner Robert Jackson Jr. added: ‘The organizational complexity of today’s largest public companies makes it far from clear how – even if top managers receive an anticompetitive signal from their pay packages – those incentives affect those making pricing decisions throughout the organization. […] For these reasons, I worry that the evidence we have today may not carry the heavy burden that … I would require imposing costly limitations’.Footnote ¹³ I focus on two recent contributions to the literature that have pushed the boundaries of knowledge in these dimensions.

My recent co-authored workFootnote ¹⁴ offers both an economic model and empirical analysis identifying a mechanism of harm. Our ‘Antón and others’ model makes two modifications to a standard model of optimal executive compensation amid a moral hazard problem. The first modification is that we do not restrict the firm to be a price taker. Instead, firms strategically interact, as in a standard model of industrial organization. The second modification is that we allow shareholders to hold more than one firm, as standard diversification motives would dictate. This latter assumption contrasts with arrived models in industrial organization, which implicitly assume that shareholders do not diversify across competitors (or that firms entirely ignore thus-arising shareholder incentives). The baseline prediction of the Antón and others model is that more common ownership leads to less performance sensitive managerial incentives. The reason is that more performance-sensitive incentives induce stronger managerial incentives to exert ‘effort’, which reduces the firm’s cost. Taking product prices as given – as in standard agency models – such a cost reduction would increase margins and profits. However, when firms strategically interact, a firm with lower costs will also optimally produce more output and set lower prices in industry equilibrium. Doing so imposes a negative externality on competitors. The shareholder of a single firm may still favour that, but common owners of competitors internalize such externalities, and therefore are less keen on spending resources to improve governance in any one target firm. As a result, commonly owned firms feature less performance-sensitive managerial incentives, and weaker corporate governance in general, and their costs are higher.

As a result of these higher costs – as opposed to because of higher margins – product prices are higher. Therefore, the view that common owners are relatively ‘lazy’ or low-cost principals that underinvest in stewardship and are ‘excessively deferential’ to managers (as Bebchuk and Hirst argue)Footnote ¹⁵ is not incompatible with anticompetitive effects of common ownership. To the contrary, in the Antón and others model, common ownership is the reason, endogenously, the owners choose to underinvest in stewardship, which in turn harms productivity and increases product prices. Phrased differently, the general insight is that cost-reducing good governance imposes a negative externality on competitors. This is one reason – among others identified elsewhere in the literature -- why common owners underinvest in ‘good governance’, compared to otherwise similar investors who do not also own large stakes in competing firms. As a result, common ownership induces a ‘productive inefficiency’ – an inward shift in supply curves – as opposed to only a redistribution of rents from consumers to producers and a small deadweight loss. This insight is important for the evaluation of policy proposals that would limit the ability of common owners to influence their product market firms but without reallocating such control rights to other investors. In the context of the Antón and others model, everything else equal, such a policy could make the problem worse.Footnote ¹⁶ I will get back to these points in Part 2 of this essay.

The theoretically proposed mechanism is empirically identified by Antón and others using the index inclusion of competitors.Footnote ¹⁷ To illustrate the strategy, assume Delta and United Airlines are already in the S&P 500 index. Southwest gets added to the index. Southwest’s ownership structure changes because of the index inclusion – after all the S&P 500 index trackers now have to buy Southwest – but Delta and United’s ownership structure is unaffected. Yet, Delta and United are ‘treated’ with a dose of common ownership: their pre-existing owners now have greater holdings of Southwest shares as well. This treatment with ‘common ownership’ is followed by reductions in the wealth-performance sensitivity of Delta and United’s top executives’ compensation packages, as predicted by theory.

This mechanism of harm can explain cross-market correlations between common ownership and higher product prices: in markets in which a commonly owned high-cost provider competes with a not-commonly owned low-cost provider, naturally, product prices are lower. In markets in which only commonly owned peers compete, prices are higher. Hence, common ownership correlates positively with product prices, even within firms and across markets. Because a low-cost provider will also produce a relatively greater quantity, the theory also correctly predicts that common ownership is negatively related to market concentration. In fact, it is the only theory to date that can organize the set of empirical results the literature has produced to date.Footnote ¹⁸

The model thus proves that there are no particular informational requirements necessary for common ownership to affect market-level competition via standard governance channels. The mechanism does not even require firm managers to know who their largest shareholders are for the managers to act in the shareholders’ interest: the manager simply takes her incentive contract as given and acts accordingly. The informal conjecture – promoted by prominent commentatorsFootnote ¹⁹ – that for common ownership to have market-level effects, an elaborate corporate governance mechanism would be required simply does not hold up to the scrutiny of mathematical logic. The theory of harm the model proposes, and the empirics identifies, is thus not complicated, but rather follows long-established standard theories and practices of ‘normal’ corporate governance activities.

The Antón and others model also features the organizational complexity that Commissioner Jackson wanted to get clarity about. In the model, it is not the top manager (whose proposed compensation is approved by shareholders) who makes market-level decisions. Instead, it is a market-level pricing specialist, who knows nothing about either common ownership, the top manager’s incentives, or the desire of the firm’s shareholders. Instead, the specialist maximizes profits market by market, taking the firm’s cost as given – as in standard models of industrial organization. This feature is in fact necessary for the theory to correctly predict the wide range of facts the empirical literature has documented. To name one example, if shareholders contracted with the top manager on costs and prices, common ownership would lead to lower costs and higher margins – which is not what is observed empirically.

Whereas the empirical analysis in Antón and others identifies a corporate governance mechanism using reduced-form empirical techniques, a second, more recent paper by Azar and RibeiroFootnote ²⁰ offers a structural estimation of a model featuring agency conflicts that also links to product market outcomes, again using the airlines industry as a laboratory. What they find is that the ‘conduct parameter’ that identifies the extent to which managers take shareholders’ portfolio interest into account is significantly positive – but also significantly below the level one would expect without agency conflicts. Hence, the rejection of full internalization by structural models that ignore agency conflicts does not reject the presence of large anticompetitive effects; such a rejection would instead indicate a mis-specified model. What is exciting conceptually is that this is the first time that a structural model not only considers common ownership but also agency problems.

The take-away from these two papers is that the recognition of agency problems in otherwise standard models of competition under common ownership by no means undoes or puts in doubt that there are anticompetitive effects from common ownership of horizontal competitors. Instead, agency problems are a feature that either gives rise to productive inefficiencies and thus higher prices in the first place, or at least is a feature that increases the precision of estimates of how strong the competitive effects of common ownership are. Both aspects become important as we think about policy proposals in Part 2 of this paper.

The structural estimation by Azar and Ribeiro also features and shows the empirical importance of vertical common ownership links. However, the importance of this feature is perhaps best illustrated by describing a reduced-form investigation into this matter.

12.2.2 How Do Vertical Common Ownership Links Affect Estimates of the Effect of Horizontal Common Ownership Links on Product Market Outcomes?

Since the start of the asset management industry’s involvement in the debate on competitive effects of common ownership and its regulation, a central argument of their advocates has been that ignoring vertical common ownership would cause the papers to ‘lack economic logic and factual support from the real world. For instance, why would passive managers want airline prices to be higher given the air travel is a cost to nearly any other business that is owned by the index funds?’Footnote ²¹ Does that argument hold up to scrutiny?

Formal economic theories on partial vertical integration with multiple upstream and downstream firms tend to be analytically intractable, and thus offer little guidance to inform the question. However, it is possible to study the question empirically, and a recent paper has done just that -- and produced a clear answer. Azar and VivesFootnote ²² show that controlling for the extent to which airlines have vertical common ownership links has a negative effect on prices – but that controlling for these effects increases the positive estimate of price effects of horizontal common ownership links. Omitting a variable capturing vertical common ownership links from the horizontal-ownership regressions leads to a bias of the estimates. The finding suggests also that policies that have the effect of reducing horizontal common ownership while strengthening vertical common ownership links may be best suited to deal with the anticompetitive effects of common ownership. Some of the proposals covered in Part 2 have that feature.

Azar and Vives’s contribution not only captures vertical common ownership links, but also offers a methodological alternative over the Azar, Schmalz and Tecu regressions: instead of using a modified Herfindahl-Hirschman-Index of market concentration – which includes potentially endogenous market shares – as the main explanatory variable, Azar and Vives follow Antón and others and others in the more recent literature and use the primitive of the common ownership theory, the profit weights that firm managers are presumed to put on commonly owned firms’ rivals. Azar and Vives is thus also the first academic study that shows with a reduced-form methodology that market shares do not drive the results in Azar, Schmalz and Tecu’s original analysisFootnote ²³ of US airline competition under common ownership, as some have speculated.Footnote ²⁴

In sum, the recognition of vertical common ownership links does not challenge the finding that horizontal common ownership links increase prices – at least in the US airlines industry as the poster child for such studies. Instead, recognition of a role of vertical common ownership links in also affecting firm behaviour strengthens the finding that common ownership of horizontal competitors increases product prices.

12.2.3 Methodological Novelties

A good part of the past five years of the literature on competitive effects of common ownership has been spent debating the relative virtues and merits of reduced-form versus structural approaches – which rely on different assumptions and are suited for different purposes. (A standard view would hold that these approaches are complementary: reduced-form approaches tend to be suited to identify causal links, whereas structural estimations are suitable to make welfare estimations and to analyse counterfactuals.) A thus-far overlooked third set of methods concerns laboratory experiments. Should there be any doubt that the variation in ownership that the empirical studies to data have interpreted to have caused an increase in prices is truly exogenous, these doubts can be examined in a laboratory setting. Hariskos, Königstein, and PapadopoulosFootnote ²⁵ show that exogenously imposed increases in partial cross-ownership increases product prices in the laboratory, and as such, break new ground. Given their study does not feature agency problems, there is no clear distinction between cross-ownership (firms holding direct stakes in competitors) and common ownership (industry outsiders holding stakes in competing firms), but this paper nevertheless opens the door for future work using this methodology as a complement to reduced-form and structural approaches.

12.2.4 Endogeneity of Portfolio Choice in the Presence of Strategic Interactions between Product Market Competitors

The key challenge for empirical economists working on the question whether firm ownership affects firm behaviour is to assess whether observed correlations in the data likely have a causal interpretation – or if they are just that: correlations. The emphasis of industrial organization economists, as it comes to scrutinizing the correlation between common ownership and higher prices, has been to focus on the potential endogeneity of market shares in polluting such estimates. I have explained above how this concern has been addressed by recent research, both with structural methods that explicitly model that endogeneity, and with reduced-form methods that avoid the problem simply by using measures of common ownership that don’t contain market shares in the first place.

However, there is another source of endogeneity that is much less well understood – but plays a crucial role in understanding the likely effect and thus desirability of the various policy proposals: it is the endogenous choice of portfolios by investors. If investor portfolios affect both firm behaviour and market outcomes, including firm profits, asset prices and the asset market equilibrium are affected by investor portfolio choice. Of course, in turn, portfolio choice is also affected by asset prices. A model capturing these mutual dependencies did not exist in the economics literature to date. In other words, we don’t know how investors choose portfolios when firms strategically interact and the ownership structure affects firm behaviour.

The empirical literature deals with this lack of theoretical understanding by finding quasi-exogenous ‘shocks’ to the ownership of a particular set of firms that is unlikely to be related to the product market dynamics in question. However, policy proposals can’t avail themselves of such techniques to predict what the likely effect of the proposals on asset prices and asset market equilibria is going to be. Such predictions will have to be made based on theoretical considerations, as emphasized in Part 2 of this paper. As a foundation, I therefore assess where we stand in our formal understanding of this complex system.

In recent years, more than half a dozen papers have attempted to address this question.Footnote ²⁶ In particular, each paper shows that strategic considerations determine whether a particular portfolio allocation is an equilibrium or not. However, there are severe limitations. The literature does not prove but appears to suggest that anything beyond duopoly settings (or otherwise extremely specific parameter) are analytically intractable. In other words, the progress in this dimension is far from offering practically relevant predictions on how new policies may change the asset market equilibrium. This negative result is useful – because it tells us that we may have to accept that we are unlikely to learn much more in this dimension in the near future, and there is no point in waiting for ‘more research’ before policy decisions are made. Instead, the uncertainty must be taken as a given when trading off the cost of waiting, and potentially continued harm from lessened competition and governance against the uncertain costs (or benefits!) that some of the proposed policies may bring. I will get back to this tradeoff in Part 2.

12.2.5 The Gradual Lifting of Data Limitations

Limitations on high-quality and comprehensive ownership data have been a bottleneck for the literature ever since its inception. Lifting those limitations is important for at least two reasons.

For one, studies that claim evidence of absence of competitive effects of common ownership (reviewed by Azar, Schmalz, and TecuFootnote ²⁷) make these statements based on analyses that lack data on some of the most important and powerful investors. Such studies are run using 13F filings filed by passive institutional investors alone – but omit other blockholders, whether they are activist investors (who must file SEC 13Ds rather than 13Fs) or individual blockholders or large insiders (who file 13Gs). Such research thus attempts to explain the competitive behaviour of Facebook, Google, and Amazon, without considering the economic interests and control rights of Mark Zuckerberg, Larry Page and Sergey Brin, and Jeff Bezos. Thus, much of the variation in the measures of common ownership is missed, resulting in biased estimates.Footnote ²⁸ The original Azar, Schmalz, and Tecu paperFootnote ²⁹ hand collected ownership from the various SEC filings and alternative sources for that industry – however, for studies beyond a specific industry that approach is impractical. The paper on top manager incentives under common ownership by Antón and others makes qualitative progress in that dimension, by combining ownership information from 13F filings with information on 13D and 13G blockholders, while controlling in their regressions for direct effects of institutional ownership on managerial incentives. The results of this improvement in data quality are that the effects of common ownership on incentives are more than two times larger than when only 13F institutions are taken into consideration.Footnote ³⁰ This result proves the point that accurate ownership data is a key ingredient in better empirical research on the question.

The second reason it is important to lift these data limitations is that it is difficult for scholars, policymakers, and lawmakers to know who owns Corporate America, and hence what precisely is the problem that needs fixing, and how big it is.Footnote ³¹ To answer that question, and to implement of any one of the policy proposals discussed in what follows, we need high-quality ownership data. Regulators have the power to supply it, but have not to date; only one academic paper exists that makes such a resource available, as I now discuss.

12.3.1 A Framework to Discuss Policy Proposals

The conceptual problem that causes anticompetitive effects of common ownership is when within-industry diversification is achieved at the same level at which corporate control is exercised. This insight is easily forgotten but has a long lineage in US policy. The 1934 Senate Securities (‘Pecora’) Report proposes that Congress must ‘prevent the diversion of these trusts from their normal channels of diversified investment to the abnormal avenues of control of industry’; the Investment Company Act bill opines that ‘the national public interest … is adversely affected … when investment companies [have] great size [and] excessive influence on the national economy’Footnote ³². Accordingly, the Investment Company Act of 1940 limits the fraction of any issuer’s shares a fund can hold to 10% of the outstanding stock.Footnote ³³ It was obvious even to the ‘father’ of index investing that this rule was meant to apply not to funds but to the level at which corporate control is exercised – in practice the fund family – and that the existing rule does not technically cover that spirit: ‘But when and if our index fund gets to 10%, all we have to do is start a second one and that would be in technical compliance. There should be limits’.Footnote ³⁴ By my own calculations based on scraped SEC filings, Vanguard already in 2020 held more than 10% on average in the S&P 500’s companies.

A clear understanding of what the problem is also makes clear the basic directions that all effective solutions to the competition problem must take: solutions to the competition problem need to separate the levels at which diversification is achieved and at which corporate control is exercised. This can logically happen in two principal ways: either asset managers diversify across competitors but then have to leave influencing portfolio firms to other investors, or asset managers have to limit themselves to diversifying across industries, and leave diversification within industries (by diversifying across asset managers) to the ultimate asset owners.

A first insight that arises from recognizing a distinction between the levels of asset manager and asset owner as in the above framework is that full diversification can be achieved at the household level even if no asset manager or fund holds more than one firm in any one industry. Therefore, there is not necessarily a tradeoff between household diversification and product market competition, as the early theoretical contributions to this literatureFootnote ³⁵ and some policy proposalsFootnote ³⁶ may have suggested. I will get back to this point in my evaluation of proposals below.

The second insight to keep in mind is that it is not possible to independently address the antitrust and governance problems generated by the rise of concentrated institutional ownership. For example, any intervention in governance that affects the firm’s cost structure also has competitive consequences, as the Antón and others model shows. In that model, any strengthening or weakening of the governance rights of institutional investors will induce a strengthening or weakening of the implementation of competitive or anticompetitive incentives implied by their portfolio. As such, desirable proposals should not narrowly address governance mechanisms that some suspect causes common ownership to increase profit margins (there is no evidence that common ownership increases profit margins). Instead, proposals should also address the lack of incentives for good governance practices that can lead to a lessening of productive efficiency and thus increased product prices, reduced output, and other harms.

A tertiary goal of any proposal is, or should be, to minimize disruptions to asset markets and the asset management industry, subject to attaining the first two goals.

With light of this framework, how should we evaluate the policy proposals made by various legal scholars, considering the research that has accumulated in the past five years? I first summarize the most prominent contributions’ main arguments and then proceed to analysing their likely effects.

12.3.2 Summary of Extant Policy Proposals

The first policy proposal in response to Azar, Schmalz, and Tecu’s paperFootnote ³⁷ being circulated was contained in Elhauge’s ‘Horizontal Shareholding’Footnote ³⁸. The opinion advanced therein is that existing law provides sufficient power to adjudicate competition problems created by common ownership of horizontal competitors, and that this power should be used to that effect. Whereas Elhauge discussed other applicable laws as well, including the Hart-Scott-Rodino Act, the main argument is that Clayton Act Section 7 already prohibits ‘any acquisition of assets that [has] the effect … of substantially lessening competition’.Footnote ³⁹ The crux is that the Act does not prohibit a particular type of conduct of horizontal shareholders of competitors (or ‘common owners’), but rather that it prohibits the asset acquisition itself, to the extent the resulting holding has anticompetitive effects. Also, no intent is required. The so-called passive investment exemption does not apply, as long as investors vote their shares.Footnote ⁴⁰ As a result, a consequential enforcement of the law would result in a reduction of common ownership links between horizontal competitors, to the extent that they are likely to have anticompetitive effects. This idealized result from enforcement of existing law has some similarities to the intended effects of several other proposals that followed.

One of them is Posner and othersFootnote ⁴¹, who agree with Elhauge’s legal assessment, but point to practical and administrative problems with Elhauge’s proposal to enforce anticompetitive common ownership links as Section 7 violations. In particular, Posner and others note, correctly, that the extent to which one institution’s holdings has anticompetitive effects depends on the holdings of other institutions. Therefore, ‘institutions obeying the law at one moment could become liable simply because other institutions changed their holdings and thereby made an industry less competitive. Institutional investors would need to determine other institutions’ ownership shares plus an appropriate definition of hundreds or even thousands of industries to comply with the Clayton Act. Thus, a large institutional investor acting unilaterally in the current environment cannot ensure it is not violating the Clayton Act. That is a difficult position for institutional investors, who require clarity about where they can legally invest’.Footnote ⁴² Further, there could – and perhaps would – be lawsuits for each transaction in each oligopolistic industry, which seems unwieldy.Footnote ⁴³ To avoid such mayhem, Posner and others propose a safe haven for institutional investors, which would exempt institutional investors from antitrust scrutiny as long as they do not hold more than 1% of the shares of more than a single effective firm in an oligopoly or are index funds that are unable to make discretionary trades and are entirely passive not only with respect to their portfolio choice but also their governance activities, including voting.Footnote ⁴⁴ Hence, investors have to choose if they want to hold large stakes in competitors or influence portfolio firms. Posner and others anticipate that large institutional investors will find this safe haven sufficiently attractive to divest from all but one firm in each oligopolistic industry, and concentrate their holdings in one target firm per industry, which would have the benefit, in Posner and others’ estimation, of improving investors’ incentives to engage in good governance in their chosen target firms. This idea has appeal as it both promises to address both the competition and the governance challenge laid out above. It would also reduce the cost of operating an index fund, due to saving on governance costs.

Rock and RubinfeldFootnote ⁴⁵ disagree both with Elhauge’s and Posner and others’ reading of Clayton Act Section 7 and with the logic behind Posner and others’ proposal of a safe haven at 1% ownership limits; they propose a safe haven for institutional investors that hold less than 15% of the issuer’s stock, don’t have board representation (whereas it is left unexplained what that means for an institution that votes 15% of the shares and thus very likely is pivotal in director elections), and only engage in ‘normal’ governance activities, including the setting of executive compensation (which itself is likely to be a problematic mechanism as per Antón and others).Footnote ⁴⁶ Rock and Rubinfeld’s reasoning is based on the assumption that passivity in corporate governance matters would protect investors from suits under Clayton Act Section 7 (which contradicts Elhauge’s and Posner and others’ legal assessment). Based on that legal opinion, Rock and Rubinfeld don’t believe investors would choose to reduce within-industry diversification in response to PSW’s safe harbour. Nor do Rock and Rubinfeld believe investors would simply not react to the PSW proposal – which would be indicated if investors truly believed, like Rock and Rubinfeld, that there is no substantive antitrust violation to be concerned about. Instead, Rock and Rubinfeld believe investors would choose to become entirely passive to protect themselves from Clayton Act 7 violations and avoid changing their business model of holding horizontal competitors. They believe this would lead to a worsening of corporate governance standards; they do not discuss why no other investor would respond to the power vacuum. Notwithstanding Rock and Rubinfeld’s criticism of the logic of PSW’s proposal, they propose a qualitatively similar rule, namely a safe harbour but with a 15% instead of 1% limit, implying that four asset managers (or individuals) could jointly control 60% of the voting shares of all firms, in any, every, and all industries or 7 asset managers could control 100% of all shares -- and yet be exempt from antitrust scrutiny. They do not explain why this policy would have the effect of protecting consumers and competition, in addition to protecting investors from antitrust suits under present law.

Lund’sFootnote ⁴⁷ proposal is similar to Posner and others’ in that she proposes to restrict the voting rights of index investors, but differs in that she does not propose a size threshold below which indexers would be allowed to engage in governance. Also, she does address index funds specifically, rather than common owners more generally, which also can – and often does -- include non-indexer investors. Posner and others by contrast aim to remove anticompetitive incentives arising from all types of common owners, but exempt index funds that don’t engage in governance activities. The idea behind Posner and others’ proposal to target common ownership by all types of investors and not just index funds is that common owners can be individuals, conglomerates such as Berkshire Hathaway, and in some cases even activist hedge funds. Posner and others’ idea behind exempting index funds is that the threat to sell shares can itself be used to influence firms; index funds that cannot make discretionary trades do not have that lever at their disposal. Lund’s argument to target index funds per se, by contrast goes as follows. Lund points out a variety of reasons why asset managers that predominantly offer ‘passive’ investment products have reduced incentives to engage in governance with the aim of maximizing individual firms’ value. Those reasons include lacking financial incentives, free-rider problems, and cost pressure. For these reasons, passive funds may encumber votes and thus prevent other investors with stronger incentives to govern and improve firm performance, but not use their power to that effect themselves. Further, ‘even if a fund does choose to intervene, it will rationally adhere to a low-cost, one-size-fits-all approach to governance that is unlikely to be in the company’s best interest’. She thus proposes that lawmakers consider restricting passive funds from voting at shareholder meetings altogether, thus leaving governance to other investors – namely, those that have incentives to be better informed and discipline management.

I have myself never made or endorsed a policy proposal, except for advocating for the collection and provision of high-quality ownership data that would allow for more and higher-quality research into the problem.Footnote ⁴⁸ To date, no such efforts have been made to my knowledge by the federal agencies to provide such a public good. Amel-Zadeh and others have since scraped and parsed the SEC’s EDGAR system for ownership records and make the resulting data base freely available for academic research; it is the first and thus far only directly sourced, complete dataset on the ownership of US firms that is usable in academic research.Footnote ⁴⁹

12.3.3 An Analysis of Extant Policy Proposals

Elhauge’s proposal to enforce existing law may first appear to be the least disruptive, by virtue of not changing the law but merely enforcing it. Further, as his proposal attacks the anticompetitive incentives implied by the holdings rather than specific governance channels, one would expect there to be no detrimental ‘chilling’ effect on investors’ governance activities from intensified enforcement, which satisfies the second goal to some extent. However, upon inspection, the proposal may be not much less disruptive to asset markets and the asset management industry than the alternative proposals to change rules or laws: after all, a large fraction of extant institutional owners’ portfolios would be affected by litigation. Abstracting away from this practical concern, the main substantive question mark with the proposal may be whether relevant judges would agree with Elhauge’s reading of the Act and the burden of proof Elhauge deems necessary to proof a lessening of competition. To the extent a common ownership case under Section 7 would be difficult to win in practice, the main risk of Elhauge’s proposed policy would be underenforcement, and thus an insufficient strengthening of competition – the first objective in my proposed framework. Hence, Elhauge’s proposal would be least disruptive mainly to the extent it fails to satisfy the primary objective, which is restoring vigorous competition.

Regarding the second goal, one would expect improvements in governance insofar as ownership structure would change from heightened enforcement, and investors with greater incentives and ability to manage agency problems would become relatively more powerful owners. The result would be increases in productive efficiency. As such, Elhauge’s proposal seems a modest push in the right direction, in both dimensions. It has that feature in common with Posner and others’ proposal.

Posner and others’ concern about liability being caused by other investors’ changing portfolio weights can be illustrated as follows: suppose two tech firms, Mazebook and Poodle, were ‘competitors’ in some meaningful sense, and suppose one individual controlled each firm via dual class share structure. Further, suppose diversified institutional investors hold voting shares in both firms but have no influence due to the presence of the controlling founders. As there is no influence by common owners, there are no anticompetitive effects from common ownership. Now, suppose one of the individuals or both sell their shares to many small investors, who don’t exercise the voting rights. Suddenly, the institutional common owners are the largest investors, and their anticompetitive incentives begin to materialize, for no fault of the investors, but simply because of the absence of the formerly present controlling founders. This is the motivation to consider safe havens for investors.

In terms of evaluation, Posner and others’ proposal is more attractive in some dimensions than they describe it to be, because of the first insight from my proposed framework: whereas Posner and others argue the loss of diversification benefits due to restricting portfolios to only one firm per oligopolistic industry was minimal, I have argued that the loss of theoretically possible diversification to households is zero. There is no loss of diversification to the ultimate investors.Footnote ⁵⁰ The households would be made poorer in their identity as shareholders (whose portfolio firms’ profits would drop because of firms operating in more competitive markets), but richer in their identity as consumers. Given the facts about the US wealth and consumption distribution, for the vast majority of the population – and certainly for the population as a whole – the latter effect dominates.Footnote ⁵¹

Whereas the asset market disruptions of such a change and other practicalities are not discussed in detail, the appeal of this solution is that not only would it remove anticompetitive incentives from holding competitors, but it would also strengthen corporate governance, both by increasing the stake investors hold in firms and thus reducing free riding by other investors, and by removing disincentives to engage due to the externality on other firms that drives the Antón and others model. Hence, the proposal addresses both goals I have declared to be desirable above. Concerning the third goal, it certainly would be disruptive, but it is not clear whether more or less disruptive than a barrage of suits to the industry under existing law! Moreover, some of the feared disruptions – such as a dramatic reduction of the profitability of portfolio firms, would be a feature not a bug in this proposal, to the extent these profits are ‘excessive’ as they are illegally derived from anticompetitive incentives from common ownership, and a reflection of the social harm of reduced competition.

As a further reflection, a key element of the proposal is not to allow investors to qualify as ‘passive’ if they merely not communicate with top management: there are many other channels by which investors can influence governance and competitive outcomes. Indeed, not engaging with companies while encumbering voting rights is a mechanism that can itself cause a lessening of productive efficiency and thus cause higher product prices. Investors should have to choose between any corporate governance interventions at all – including voting – or not holding large stakes in competitors. Stated this way, the Posner and others’ proposal is not dissimilar to Lund’s proposal, whose idea mainly differs in that she proposes to outright limit index funds’ voting rights, rather than giving investors the freedom to self-assess their likely antitrust liability and choose whether they want to engage in governance or become purely passive in both meanings of the word.Footnote ⁵²

Rock and Rubinfeld’s proposal, by contrast, appears to have the effect of protecting institutional investors from antitrust liability, even if they do in fact cause a lessening of competition, up to a limit at which it is virtually assured that there would be a lessening of competition by common ownership: I am not aware of economic theories that would predict that if 100% of a firm’s shareholder base identically overlaps with the shareholder base of all competitors, that one could expect no lessening of competition. Yet, this would be explicitly covered and exempt from scrutiny under Rock and Rubinfeld’s proposal. The proposal therefore is likely to be ineffective at addressing any anticompetitive effects of common ownership, but in fact removes concerns about enforcement some institutional investors may have that engage in strategies that are likely to have anticompetitive effects.

Rock and Rubinfeld’s proposal further appears to do worse than either Elhauge’s or Posner and others’ in the second dimension. The reason may be that Rock and Rubinfeld rely on an erroneous understanding of the economic theory at the core of the common ownership problem. Rock and Rubinfeld appear to believe that active involvement in corporate governance is necessary for common ownership to bring about a lessening of competition. Their proposal holds that not-too-large investors should be exempt from antitrust scrutiny as long as they pursue only ‘normal governance activities’, including setting executive compensation, which they expect would not change competitive outcomes, and hence not fall under what Clayton Act Section 7 is meant to capture. Yet, Antón and others are the last paper in a long line of economic literature that has shown that compensation can cause anticompetitive outcomes.

Recall that Rock and Rubinfeld’s proposal would protect corporate control rights of institutions that have questionable incentives to use them for the benefit of society. The incentives are doubtful be it because asset managers face agency problems of their own (as in Bebchuk and Hirst), because governance improvements in one firm harm other firms (as in Antón and others), because they don’t sufficiently care about individual firms (as in Lund), or because of excessive competition to reduce fees in the market for asset management product, and thus limit the resources devoted to governance.

As such, the Rock and Rubinfeld proposal achieves either of the two aims I outlined in the framework to evaluate the policy proposals. I conclude that whereas this proposal appears suitable to deal with the SEC’s concern to help investors deal with the ‘investor protection challenge of the 21st century’,Footnote ⁵³ the proposal seems to have the effect of weakening existing consumer protections afforded by the Clayton Act.

Lund’s proposal to limit the voting rights of institutions at first seems like a recipe for a corporate governance catastrophe.Footnote ⁵⁴ By main own calculations, in typical US publicly traded firms, the largest shareholder who does not also own similarly large stakes in competitors (and who should therefore be captured by a version of her proposal for the purposes of addressing the antitrust concern, whether a passive fund or not) tends to rank above the top 50 and tends to hold less than 1% of the cash flow rights. Giving this investor a disproportionate share of the control rights would dramatically misalign cash flow and control rights, and thus lead to the definition of a corporate governance problem.

However, this thinking overlooks the asset market’s equilibrium response. Whereas an analytic solution and precise prediction for this response remains elusive to the researchers working on the question, it is clear that some qualified investors interested would be attracted to purchase the shares of a firm in which control or significant influence can be bought by acquiring less than 1% of the outstanding stock. Indeed, the implementation of the proposal might trigger a revival of activist shareholders who in recent years have increasingly concentrated on campaigns that are agreeable to the big common owners.Footnote ⁵⁵

In an ironic twist, the outcome of this endogenous reallocation of cash flow and control rights may lead to a similar outcome as the one Posner and others’ proposal aims for. In the re-allocated equilibrium, there exist fully diversified investors, but they don’t engage in governance; the only investors that engage in governance are those that have concentrated stakes in the target firm. The same benefits in terms of jointly improving competition and governance would result.

Both proposals also have in common that index funds could implement the proposals without any disruption to their business model, other than stopping their costly governance activities, leading to a further cost reduction for ultimate household investors, as well as a reduction in the concentration of corporate control.

The above discussion aims to organize the proposals along the principal directions that matter. The proposals differ in many other details, but which can be iterated and improved upon. For example, Lund’s proposal to prevent index funds from voting does not capture other common owners and thus would be under-inclusive compared to Posner and others’, but this feature could be adjusted to avoid such under-inclusion. A discussion of such details is beyond the scope of this paper, so as to focus attention to whether the principal direction of travel appears suitable to achieve the main goals of the policies.

In sum, the proposals that would actually address the issue – Elhauge’s, Posner and others’, and Lund’s – differ substantially in their methods, but are essentially similar in that they all aim for the same, desirable outcome: a separation of diversification and influence over portfolio firms. My contribution is not to judge which of those means are most appropriate, politically feasible, or realistic, but to enable readers to see the commonalities and the cost and benefits from an economic perspective.

There is no doubt that each one of these proposals would lead to a substantial reorganization of asset management and asset markets, which would be disruptive, which many appear to view as per se undesirable. On the other hand, the counterfactual to implementing a version of these proposals is to allow for no less fundamental changes to occur, if incrementally day by day. Namely, it would amount to letting an unprecedented concentration of control over industry to grow ad infinitum and remain virtually unchecked, which would be at odds not only with economic theories of competitive markets but American political ideals as well.

It hence appears increasingly likely that, earlier or later, depending on the political climate, something will be done. ‘Recognising this potential issue further down the line, BlackRock has taken an active step in allaying concerns by offering asset owners in 40% of its $4.8trn equity index fundsFootnote ⁵⁶ the opportunity to vote directly with companies, instead of the firm partaking itself’.Footnote ⁵⁷ In other words, BlackRock implemented a version of the Posner and others/Lund proposal so as to forestall regulation. This appears to be substantively desirable at first glance. However, ‘While this goes some way in addressing the issue in wider indexing, the problem will continue in the ETF space where there is less transparency on who owns the ETFs due to the fact they trade on the secondary market’.Footnote ⁵⁸ In other words, it remains unclear whether this move also obscures the picture of who controls Corporate America, thus disabling further research on the matter, and reemphasizing my call for more complete and correct ownership data to be made available. On second thought, it is also substantively undesirable, first because the move fails to separate the levels at which corporate control is exercised and the level at which diversification is achieved. And if the practice achieved such separation by ultimate shareholders not voting their shares, we would go back to the old Bearle and Means problem of rationally apathic shareholders who fail to control management. Therefore, a solution in which blockholders exist but in which these blockholders don’t hold competitors and have strong incentives to monitor management is more desirable along all key dimensions.