If corporations are to maximize shareholder welfare, managers need to discover what shareholders value; political theory shows how difficult this can be.
Editor’s note: To mark the 50-year anniversary of Milton Friedman’s influential NYT piece on the social responsibility of business, we are launching a series of articles on the shareholder-stakeholder debate. Read previous installments here.
One of the virtues of wealth maximization is its simplicity. Managers do not need to know anything about the preferences of investors when making decisions, they simply adopt all projects with a positive net present value.
In contrast, maximizing shareholder welfare would require managers to know the preferences of shareholders. For example, are they willing to give up 10 percent of profit in order to reduce greenhouse gas emissions? And it would require a way to adjudicate differences of opinion. What does Walmart do if some shareholders want to stop selling guns while others want to continue gun sales?
The comments in this series often mention the complexity of decision-making if managers were charged to maximize shareholder welfare (or stakeholder welfare), but the problem is even more complex than sometimes is recognized. Here, I would like to flag some challenges that have been identified by political economy research and offer thoughts on how they might be addressed.
To maximize shareholder welfare, managers somehow would have to aggregate the preferences of individual shareholders into a “social welfare function” that can be maximized. One aggregation procedure, for example, would be to determine preferences by voting, allowing each share one vote, and let the majority rule.
1. The first challenge is Arrow’s Impossibility Theorem, proved by Nobel economist Kenneth Arrow in 1951, which shows that it is impossible to aggregate preferences in a way that does not violate at least one of a small number of basic decision principles. These principles include: one person’s preferences do not always determine the outcome (non-dictatorship), if every person prefers option A to option B then option A is selected, and preferences over irrelevant alternatives do not affect the outcomes.
The theorem is difficult to explain in plain English but is recognized by theorists as the starting point for analysis of collective decisions. Its scope is far-reaching. It is not that we haven’t yet figured out a coherent way to aggregate preferences; it is that a coherent solution does not exist: every collective decision process must violate at least one of the basic principles. Charging managers to maximize shareholder welfare would ask them to solve a problem that has no solution.
2. A second important insight pertains to voting, which likely would be part of any process of maximizing shareholder wealth. The Gibbard-Satterthwaite Theorem, proven in the 1970s by a philosopher and an economist, shows that almost all voting systems are subject to strategic voting. That is voting in which people do not select their top choice. A simple example, familiar to most readers, is a three-candidate race in which only two candidates are viable. A voter who prefers the third candidate might not vote for that candidate but rather strategically choose one of the top two. The importance of the theorem is that companies cannot rely on voting to accurately reveal preferences.
3. Another concern is the power of agenda control in determining the outcome of elections. The person who decides what is to be voted on, and in what order, has tremendous power to influence the outcome. One of the key theorems in this regard is that if the issues are two-dimensional or more and voters disagree over the options, by arranging the order in which options are decided (taking a vote on A vs. B, followed by the winner vs. C), the person with agenda control can induce any possible outcome. To prevent this sort of manipulation, legislatures impose extensive procedural constraints on their decision processes. It would be necessary to restrict managers’ control of the voting agenda as well.
4. Finally, there is “capture theory,” the problem of interest groups’ influence, which can be traced to an article by Nobel economist George Stigler in 1971 that emphasized the vulnerability of democratic systems to influence by wealthy and/or organized groups. In the public sector, capture occurs when elected officials and their appointees are susceptible to lobbying, revolving doors, and campaign contributions. In a corporate environment where directors and managers were charged to make political tradeoffs, similar channels of influence could well emerge. This problem is mitigated if voters are fully informed, but many investors are now passive owners, with little incentive to collect information, and few are confident that proxy advisors can cover for the information deficit. We already have some evidence that organized groups such as unions and public pensions are able to use shareholder proposals to advance narrow objectives.
Despite these theoretical challenges, countries are able to function more-or-less effectively with democratic systems (although, many would say they skew toward “less” these days). Political theory points to some possible paths around these challenges.
Even though theory suggests that voting outcomes are unreliable indicators of voter preferences, there is one special exception. May’s Theorem, proved by mathematician Kenneth May, shows that voting accurately represents preferences and avoids the Arrow problem when there are only two choices and the winner is decided by majority rule. One path out, then, may be voting on issues for which there are only two essential options, usually called “referendums” in the political context. The obvious limitation is that many issues are not binary in nature—one can imagine a host of climate abatement strategies or an election with more than two candidates—and forcing them into a binary choice would allow the agenda controller to induce an outcome. Nevertheless, there may be some issues that naturally resolve themselves into two broad options.
To solve the challenge of manipulation by agenda control, the agenda can be constrained in advance. Perhaps the simplest way would be to require a specific action in the corporate charter, such as Patagonia’s commitment to contributing one percent on net revenue to organizations promoting environmental conservation and sustainability. A more flexible approach would be to require yes-or-no shareholder votes on certain issues. States and cities require voters to approve actions such as issuance of debt, authorization of a casino, and school district budgets. One could imagine a progressive company’s charter requiring a shareholder vote before investing in a country listed as a human rights violator by international organizations, or to get shareholder approval before testing products on animals.
What does this mean for the bottom line: is shareholder or stakeholder capitalism feasible or not? Steve Kaplan’s argument in this series is compelling: the current system of shareholder value maximization has served us well, and should not be replaced without a strong sense that there is a superior alternative. At the same time, Hart and Zingales are right that there ought to be space for investors to create and operate companies that pursue social goals if that is what they prefer. The lesson I would take away is that it may be possible to implement stakeholder capitalism, but we would need to know in more detail how the collective decisions would be made, and how the known challenges to collective welfare maximization would be addressed.