The Causal Mechanisms of Horizontal Shareholding

As I have shown in another paper, repeated empirical studies confirm that, in concentrated markets, higher levels of horizontal shareholding make anticompetitive effects more likely. Nonetheless, some critics argue that we should delay enforcement action until we have more proof on the causal mechanisms by which horizontal shareholders influence firm behavior. In my new paper, The Causal Mechanisms of Horizontal Shareholding, I show that these critiques are mistaken.

The Ample Evidence on Causal Mechanisms

Institutional investors now cast 93% of shareholder votes at S&P 500 firms and are increasingly diversified in ways that give them horizontal shareholdings. The weight that firms put on the profits of other firms can range from 0% (if the firms are totally separately owned) to 100% (if one firm 100% owns the other). Horizontal shareholdings have risen so much that by 2017 the average weight that an S&P 500 firm put on the profits of other firms in their industry was 75%. There are multiple mechanisms by which horizontal shareholders use this power to influence firms.

• 1. Board Elections. A recent economic proof shows that voting by horizontal shareholders on board elections will incline managers to lessen competition, as long as managers care either about their vote share or their odds of re-election. Some argue that shareholder voting on director elections is unlikely to influence corporate behavior because proxy statements do not specify director business strategies and most corporate elections are uncontested. But empirical studies show that decisions to oust corporate managers from their jobs are driven almost as much by industry performance as by individual firm performance. Empirical studies also show that, even in uncontested elections, an increased share of votes withheld from directors significantly increases the odds that those directors will depart the board or lose key committee seats.
• 2. Executive Compensation. As Professors Bebchuk and Fried have observed, firms generally compensate executives using measures (like stock options) that are driven 70% by industry performance and only 30% by individual firm performance. This makes no sense as a matter of firm efficiency. But it makes perfect sense given horizontal shareholding. A recent economic proof shows that increased levels of horizontal shareholding mean that overall shareholder interests are maximized by executive compensation methods that are less sensitive to individual firm performance.
• Some argue that this mechanism is implausible because shareholder voting on compensation is either nonbinding or about high-level terms. But empirical studies show that, even in non-binding votes, higher levels of shareholder dissent on executive compensation lead to lower CEO pay. And high-level terms are what determine the sensitivity of compensation to firm performance. Further, recent empirical work shows that corporations with higher levels of horizontal shareholding actually do adopt compensation methods that make changes in executive wealth less sensitive to their own firm’s performance.
• 3. The Market for Corporate Control. How horizontal shareholders vote in control contests can directly affect whether the corporation pursues a less competitive strategy. Further, managers have incentives to act in ways that please the horizontal shareholders that may be decisive in any future control contest.
• 4. The Stock Market. If managers displeased horizontal shareholders by competing too aggressively, they might sell their stock, which would depress the stock price and the value of executive stock options. 56% of institutional investors say they try to influence corporate managers by selling their shares when displeased.
• 5. The Labor Market. Directors who want additional directorships at other firms and executives who want a promotion to their next job at another firm will want the support of those other firms’ leading shareholders, who are likely to be horizontal shareholders in their existing firm. Empirical studies confirm that increasing the share of votes withheld from a director in one firm’s election reduces the number of directorships that person gets at other firms.
• 6. Direct Communications. Although direct communications are not necessary for any of the above causal mechanisms, they can exacerbate the anticompetitive effects. 63% of institutional investors admitted that they tried to influence corporate managers via direct discussions. The Big Three index fund families require portfolio companies to submit written information on topics that include corporate strategy and executive compensation, and they follow up with thousands of private conversations when necessary.
• 7. Reduced Pressure to Compete. Because competing vigorously is hard work, managers are less likely to do it unless their shareholders are actively pressing them to compete. Horizontal shareholding can make managers less likely to compete simply because it makes those shareholders less willing to exert such pressure.

The Hemphill-Kahan Critique

Professors Hemphill and Kahan argue that the possible causal mechanisms are either empirically untested or generally implausible. They are wrong on both scores.

They claim that mechanisms with consensus effects (i.e., effects profitable for both horizontal and non-horizontal shareholders) have not been empirically tested because the standard MHHI measure of horizontal shareholding goes up when non-horizontal shareholders are less concentrated. But the fact that non-horizontal shareholders have incentives to want their individual firm to undercut noncompetitive pricing does not alter the reality that they profit if horizontal shareholding can simultaneously reduce competition at their firm and its rivals. Further, empirical studies do show the consensus effect of increased market prices.

Hemphill and Kahan also assert that mechanisms with across-the-board effects (i.e., effects on the firm’s general competitive tendency, rather than in targeted markets) have not been empirically tested because the airline study found effects based on differences between routes. But the airline study actually found that 90% of the effect was across-the-board and only 10% varied with specific routes. Further, numerous other studies find effects on general firm competitiveness.

Hemphill and Kahan admit that mechanisms with targeted effects are well tested, but they claim that active targeted mechanisms are implausible. However, because they mistakenly assume away consensus effects, their argument wrongly assumes that active efforts must harm non-horizontal shareholders in ways that create legal risks. Their argument also assumes that horizontal shareholders lack the market-specific knowledge needed for targeted effects, which conflicts with evidence to the contrary, as well as with their own admission that studies have confirmed targeted effects.

The Bebchuk-Cohen-Hirst Critique

Professors Bebchuk, Cohen, and Hirst claim that index funds lack sufficient incentives to exert effort to get firms to increase corporate value, and Bebchuk and Hirst argue that actual effort levels are low.

I show that index funds actually have strong incentives to increase portfolio value by lessening competition because the resulting gains in fees and investment flow are vast, while the incremental effort costs are generally zero or negative. Further, horizontal shareholdings are generally not held by index funds and, even when they are, their shares are voted by fund families that also have active funds. The Bebchuk-Cohen-Hirst argument also mistakenly assumes that horizontal shareholders cannot have anticompetitive effects if their efforts fall short of the level that maximizes firm value, but anticompetitive effects instead depend on their influence relative to other shareholders, who are likely to exert less effort.

Finally, the argument that index funds lack incentives to exert effort to increase corporate valuations conflicts with numerous empirical studies showing that index funds not only engage in extensive efforts, but also are highly effective at changing corporate conduct and performance. The argument also conflicts with two dozen empirical studies showing that common shareholding does influence corporate behavior.

The complete paper is available for download here.