chevron-down Created with Sketch Beta.

The Business Lawyer

Summer 2022 | Volume 77, Issue 3

How Would Directors Make Business Decisions Under a Stakeholder Model?

Robert Miller


  • Under the stakeholder model of corporate governance, directors may confer benefits on corporate constituencies other than shareholders without regard to whether doing so produces benefits for the shareholders even in the long run. Contrary to what advocates of stakeholder theory often say, stakeholder theory does not put all corporate constituencies on a par, letting directors give equal consideration to the interests of all constituencies. Rather, stakeholder theory uniquely disadvantages shareholders, allowing directors to transfer value from shareholders to other constituencies but never from other constituencies to shareholders. More importantly, although critics of the stakeholder model going back to Berle have complained that the model provides directors with no clear standard by which to make business decisions, this criticism grossly understates the problem. In fact, the stakeholder model says nothing at all about which interests of the various constituencies are legitimate interests, much less about how such interests should be balanced against each other. As a result, the model provides no normative criteria of any kind on the basis of which we can intelligibly say that one business decision is any better—or any worse—than any other. Consequently, under stakeholder theory, every possible decision is as good and as bad as every other possible decision. The stakeholder model is thus not just insufficiently determinate but radically indeterminate.

    The question thus becomes whether there are any plausible normative criteria that can be added to the stakeholder model to make it reasonably determinate. Some obvious candidates are Kaldor-Hicks efficiency, hypothetical bargains among the corporate constituencies (both ex ante and ex post), and Delaware doctrines about the apportionment of merger consideration among different classes of shareholders, but it turns out that none of these can supply the normative lacuna in the stakeholder model. The model could be supplemented with a robust normative theory, such as that in Rawls’s A Theory of Justice, Mill’s act utilitarianism, or Aquinas’s natural law theory, but this would require directors to become experts in moral philosophy and so echoes the improbabilities of Plato: until directors become moral philosophers or moral philosophers directors, there shall be no coherent stakeholder governance. The view that decisions made under the stakeholder model are necessarily unprincipled is confirmed from the writings of leading stakeholder advocates who expressly concede that, under a stakeholder model, the decisions of directors will be essentially political—i.e., determined not according to any rational, normative principles but by the varying abilities of different interest groups to pressure or lobby the directors. As an attempt to explain how directors should make business decisions, the stakeholder model is thus hopelessly and fatally flawed.
How Would Directors Make Business Decisions Under a Stakeholder Model?

Jump to:

In Revlon v. MacAndrews & Forbes Holdings, Inc., the Delaware Supreme Court held that, in making ordinary business decisions, directors may take account of the interests of corporate constituencies other than shareholders but only subject to the fundamental limitation that “there are rationally related benefits” accruing to shareholders. That is, the board may confer on a corporate constituency a benefit to which that constituency is not legally entitled, thus imposing a cost on the shareholders, but only if, in an otherwise legitimate exercise of its business judgment, the board believes that the action will in the long term produce a net benefit for the shareholders. A standard example is paying employees severance to which they are not legally entitled. A board may authorize such payments if, for example, it honestly believes that making the payments will help the corporation attract and retain talented and hard-working employees to such an extent that the present value of such long-term benefits exceeds the cost of the payments. As former Chief Justice Strine has put it, “a corporation may take steps, such as giving charitable contributions or paying higher wages, that do not maximize profits currently . . . because such activities are rationalized as producing greater profits over the long-term.” Some academics have attempted to minimize this landmark holding in Revlon or otherwise claim that it does not mean what it plainly says, but the Revlon court was merely repeating the traditional rule in Anglo-American corporate law, and Delaware courts have consistently applied the rule in subsequent cases.

By contrast, under what might be called the strong form of the stakeholder governance model, the board is entitled to consider the interests of all corporate constituencies and make decisions that benefit constituencies other than shareholders even when doing so does not produce net benefits for shareholders in the long term. The board is entitled, in effect, to transfer wealth on a net basis from shareholders to members of other corporate constituencies. This strong form of stakeholder theory thus differs from most ESG advocacy, which typically claims that pursuing ESG goals maximizes value for shareholders, and from weaker forms of stakeholder theory, which merely seek to remind and admonish directors that managing the corporation’s business in the manner of an unreformed Ebenezer Scrooge is unlikely to maximize value for shareholders in the long run.

But if the characteristic claim of the strong form of the stakeholder model is that directors may transfer wealth from shareholders to other corporate constituencies even when there is no long-term benefit to the shareholders, advocates of the strong form of the stakeholder model do not say that directors should do this arbitrarily or randomly or irrationally. They say, rather, that directors should allocate benefits among corporate constituencies fairly or justly, in a manner that is right and good—that is, according to some normative criterion. Critically, however, there is no consensus among stakeholder theorists about what this normative criterion is. Indeed, almost all such theorists have been notoriously vague about the matter, and most have utterly failed to address the issue in any manner whatsoever.

The purpose of this article is to explore, at a conceptual and theoretical level, problems with the stakeholder model that arise from the failure to specify this normative criterion. I begin propaedeutically in Part A by exposing a popular misconception concerning the stakeholder model. That is, although advocates of that model often say (and even more often imply) that the stakeholder model seeks to balance the interests of all corporate constituencies fairly and equally, the reality is otherwise. As I shall show, the stakeholder model uniquely disadvantages shareholders relative to other constituencies. As a result, the normative criterion that stakeholder theory needs but lacks would be used not to adjudicate the competing claims of various constituencies on equal terms but to determine how much shareholders should be disadvantaged in order to benefit which other constituencies and how much. In Part B, I move to my main point and argue that the problem created by the absence of the needed normative criterion is much more severe than most people realize—indeed, that the absence of the criterion leaves business decisions made under a stakeholder model not just unacceptably indeterminate but radically indeterminate. In Part C, I begin considering how the normative lacuna in stakeholder theory could be filled, and I contrast the stakeholder model with the shareholder-wealth maximization model of Delaware law to argue that the lacuna in the stakeholder model cannot be supplied by any criterion based on economic efficiency. In Part D, I consider whether we can fill the lacuna by appealing to hypothetical bargains—e.g., what the shareholders and other constituencies would have agreed to had they negotiated over the relevant issues ex ante—but I conclude that such appeals are ultimately unavailing. In Part E, I ask whether Delaware law contains any other conceptual resources that might help us fill the lacuna in the stakeholder model, and although doctrines relating to how boards ought to apportion merger consideration among various classes of shareholders seem promising, I argue that, on closer examination, these too are unavailing. In Part F, I conclude that the lacuna in the stakeholder model can be filled only by a robust substantive moral theory, but I suggest that there are powerful reasons for thinking that directors could never reach consensus on which such theory to apply and that expecting them to do so is in any event patently unreasonable. In Part G, I turn to how, in the absence of the needed normative criterion, directors would actually make business decisions under a stakeholder model, and I show that, according to both advocates and critics of stakeholder theory, such decisions would be made on the basis of raw political power—i.e., on the basis of how much pressure various constituencies could bring to bear on the directors. In Part H, I make some concluding remarks.

A. Stakeholder Theory Uniquely Disadvantages Shareholders Relative to Other Constituencies

Many advocates of the stakeholder model speak as if the model requires boards to treat all constituencies equally and fairly, in each case considering the interests of all affected constituencies, benefiting now one constituency and now another, all in accordance with some normative criterion. In my view, this picture is misleading in that it implies that shareholders are merely one constituency among others, all of which are in some important sense equal. The reality is that, under the stakeholder model, shareholders are at a fundamental disadvantage relative to all other groups. They are worse off than other corporate constituencies.

The reason is that, although the stakeholder model permits (indeed encourages) boards to transfer wealth from shareholders to other constituencies, the model never permits the board to transfer wealth from other constituencies to shareholders. It must be this way, because the claims of other constituencies such as employees, customers, creditors, and suppliers are contractual in nature. If the board attempted to give such a constituency less than what the corporation owed it under the relevant contract, then that constituency would have a good claim in contract against the corporation. It would be legally entitled to recover from the corporation whatever it had bargained to receive from the corporation under the contract between the parties. Thus, for all constituencies other than shareholders, their contractual claims are a floor under the amount that they may hope to receive from the corporation under the stakeholder model. Consequently, any application of the stakeholder model will involve paying every constituency other than the shareholders what they are legally entitled to receive under a contract with the corporation—and perhaps more. In every case in which a constituency receives more than it is due under its contract with the corporation, the additional value it receives always comes at the expense of shareholders, never at the expense of any other constituency. In other words, applying the stakeholder model results in a transfer of wealth from shareholders to one or more other corporate constituencies. Only in the limiting case when each constituency receives its contractual minimum and no more is wealth not transferred away from shareholders, and in no event is wealth ever transferred from another constituency to shareholders.

To take a vivid example, suppose that the company has entered into a collective bargaining agreement with some of its employees, and, at the time the contract was signed, the agreement was fair (in whatever sense one thinks applicable) to all parties. Later, the company’s business takes off, and its profits greatly exceed forecasts made at the time that the parties entered into the agreement. Stakeholder advocates would likely say that, in such circumstances, the board should pay the employees more than the employees bargained for in the contract, thus leaving less for the shareholders. But if the things had worked out differently, and if the company had grossly underperformed the forecasts made when the parties entered the contract, with the result that its profits were much lower than expected (maybe the company is even experiencing significant losses), then stakeholder advocates would not generally say that the company may ignore its contractual obligations to its employees and pay them less than the collective bargaining agreement requires in order to provide a return to shareholders closer to what everyone reasonably expected at the time the parties entered into the contract. Short of bankruptcy, what the employees are due can be reduced only with their actual consent. Under the stakeholder model, in other words, when the venture in which the various constituencies have an interest works out well, the bargain among the parties may be adjusted to benefit other constituencies at the expense of shareholders, but if the venture works out badly, other constituencies will get the benefit of their bargain and any unexpected losses fall on the shareholders. It should go without saying that no rational commercial party would ever agree to such terms. Indeed, from the point of view of shareholders, the stakeholder model is Bugs Bunny corporate governance: heads I win, tails you lose.

Notice that shareholders are also uniquely disadvantaged relative to constituencies whose claims against the corporation are not contractual in nature but arise by operation of law. For example, stakeholder theorists often say that companies ought to do more for the environment than the environmental laws require, even if this reduces returns to shareholders. But no one thinks that, if a company’s returns are lagging, the board could decide to favor shareholders at the expense of the environment for a while by suspending the company’s compliance with some particularly burdensome environmental laws. Just like constituencies whose claims against the corporation are contractual, constituencies whose claims arise under statutes are absolutely entitled to have those claims satisfied in full and, under a stakeholder model, may perhaps get more besides, and when they do get more, it is always at the expense of the shareholders, not at the expense of any other constituency.

But this result is hardly surprising. Under the traditional view, each corporate constituency has a contractual or statutory right to a limited portion of the corporate assets, and the shareholders are entitled to the residuary. This is why the corporation may distribute cash to shareholders, either as dividends or through share repurchases, only if the corporation is solvent and only if the distribution does not make the corporation insolvent. Any payment to another constituency over and above that to which it is legally entitled reduces the residual value of the corporation and so necessarily comes at the expense of shareholders. Indeed, the traditional view was that, precisely because common shareholders have no contractual claim to any particular distributions but merely an interest in the corporate residuary, they are uniquely vulnerable to being expropriated by faithless managers, and so to protect against this danger the law imposes a fiduciary duty on directors to manage the corporation for the benefit of the shareholders. Remove this fiduciary duty or, what amounts to the same thing, make it run to all corporate constituencies indiscriminately, and exactly the result predicted by the traditional understanding follows: value is transferred away from shareholders to other groups.

The upshot of all this is that stakeholder theory is a one-way ratchet against shareholders. Every constituency other than the shareholders gets its contractual or statutory due, and whatever is leftover—the value that under the shareholder-wealth maximization model belongs to the shareholders—is then distributed among constituencies, with every dollar going to any non-shareholder constituency decreasing dollar-for-dollar the amount going to shareholders. Thus, Professors Blair and Stout, leading stakeholder advocates, expressly say that directors should “decid[e] which members of the corporate coalition receive what portion of the economic surplus resulting from” the corporation’s operations. This does not of itself prove that stakeholderism is bad or wrong, but it does clarify an important point about the nature of a board’s business decisions under a stakeholder model. In making a business decision under a stakeholder model, the question before the board is not how to balance the interests of various corporate constituencies, shareholders among others. The question is, given that all constituencies get their contractual or statutory minimums, how much more (if anything) should be taken from shareholders to benefit which constituencies. This is the question that the normative criterion missing from the stakeholder model has to answer.

B. The Stakeholder Model Radically Underdetermines Business Decisions

Beginning with Berle, critics of stakeholderism have argued that, if directors are licensed to consider the interests of many different corporate constituencies, then directors will be able to justify any decision whatsoever and thus will evade all accountability. According to this argument, after settling on whatever decision they please, the directors need only identify some constituencies benefited by the decision (every decision benefits someone) and then claim that, in the totality of the circumstances, they believed those constituencies ought to be benefited at the expense of the shareholders (again, not at the expense of other constituencies, for all non-shareholder constituencies always get everything they are owed in contract or under a statute). In such cases, the decision in question really will benefit the ostensible beneficiaries, and since stakeholder models allow boards to benefit such constituencies at the expense of shareholders, challenges to the board’s decision will almost certainly fail. Hence, the absence of a clear statement of how, under the stakeholder model, boards are to allocate value among various corporate constituencies is a serious problem for the theory. Berle saw the essential point ninety years ago, arguing that “you cannot abandon emphasis on ‘the view that business corporations exist for the sole purpose of making profits for their stockholders’ until such time as you are prepared to offer a clear and reasonably enforceable scheme of responsibilities to someone else.” This is certainly right, but the problem with stakeholder theory goes much deeper than Berle and even the most severe critics of stakeholder theory have generally realized.

To see why, start with what we might hope that a model of how directors should make business decisions would do. Ideally, such a model would include normative criteria such that, given the same factual information about the issue at hand, all rational inquirers applying the criteria would agree on what action the corporation should take. This is how the shareholder-value maximization model works. That is, subject to disagreement about the facts (which is, of course, often very significant), the shareholder-value maximization model will determine which of the available options the directors should choose: they should choose the option that has the largest positive effect on the present value of the company’s future free cash flows. This is one of the great strengths of that shareholder model: it is complete in the sense that, for any two possible courses of action, the model will determine (modulo empirical uncertainty about facts) which of the two is preferable. But, while such completeness is obviously desirable in a model of board decision-making, it is not absolutely required. After all, for any model whatsoever, empirical uncertainty will inevitably produce a great deal of disagreement as to which option is best; uncertainty arising from other sources, such as indeterminacy in the model itself, is thus not necessarily intolerable. In practice, it suffices if the model is such that, at least in general, real directors of real corporations can effectively use the normative criteria in the model to make real decisions. This means that the criteria must be such that rational persons, given the same factual information, will agree on what is to be done in a sufficiently large percentage of cases to make the model practically useful. Clearly, this percentage cannot be stated with mathematical precision, but there is no reason that it has to be. For instance, a model would be useful in practice if the normative criteria it contained were such that (a) in approximately 80 percent of the cases, given the same factual information, all rational inquirers would agree on which course of action to take in such cases (these might be thought of as the “easy” cases), and (b) in the remaining 20 percent of the cases, given the same factual information, all rational inquirers would agree that two or three certain possible courses of action were superior to all other possible courses of action (these might be thought of as the “hard” cases, where, within certain limits, reasonable persons could disagree as to which of a handful of possible solutions is best).

Now, the problem for stakeholder theory is that, unless supplemented by some additional normative criteria, the principle that, in making a business decision, the board should consider the interests of each of various constituencies is consistent with the result that every possible action the board could take is as good (and as bad) as every other possible action. That is, unless supplemented by some additional normative criteria, the stakeholder model fails to imply that any action is any better or any worse than any other action. While the shareholder-wealth maximization norm uniquely determines which action is best, and while a minimally acceptable model would narrow down the eligible actions within some manageable limits, the stakeholder model never provides any basis for eliminating from consideration any possible course of action. Even if all empirical uncertainty were eliminated such that the directors knew exactly what consequences would flow from every action available to them, the stakeholder model provides no basis at all for thinking that any action available to the directors is any better or any worse than any other action.

The reason for this startling conclusion is that, in providing that the board should consider the interests of each of various constituencies, the stakeholder model says no more than that, in distributing value among certain possible recipients, the board should consider giving at least some value to each of the various possible recipients; it says nothing about how much any particular recipient should receive in any particular circumstances or when one recipient ought to receive less in order that another may receive more. For instance, suppose I tell you to distribute a hundred francs among three musketeers, considering the interests of each musketeer when you do so. Assuming the francs are indivisible, each musketeer must get between zero francs and all hundred, which implies that there are 5,151 possible ways to divide the francs among the musketeers. If I add that you should consider carefully the interests of each musketeer before deciding on a distribution, this says nothing at all in favor of any possible distribution of the francs relative to any other distribution, at least if I do not also tell you which putative interests of the musketeers are legitimate and how to make tradeoffs between interests and across individuals. Without such additional information, each of the 5,151 possible distributions of the one hundred francs among the three musketeers remains equally eligible. To start narrowing things down, you need to know what I have not yet told you—what tends to entitle a musketeer to a franc and how to handle tradeoffs—for instance, how to know when it is good or right that Athos have a franc less in order that Aramis may have a franc more.

The stakeholder model faces exactly the same problem on a far grander scale. The amounts to be distributed are commonly in the hundreds of millions of dollars, and the number of possible recipients in the tens or hundreds of thousands (large corporations have many employees and customers). The number of possible distributions is thus astronomical, and the absence of the crucial normative criteria in the stakeholder model makes each of these distributions equally eligible. In other words, it is not that stakeholder theory fails to provide sufficiently determinate answers as to how a board should make business decisions; it is that, for every business decision, the stakeholder model provides no answers at all. It is thus the exact opposite of what rational commercial parties would typically agree to when, in a complex venture, the profits of the venture must be distributed among various parties who have participated in the venture in different ways. The agreements used by such parties typically provide for elaborate cash waterfalls that determine exactly how much each party is entitled to receive. The stakeholder model is at the very opposite end of the spectrum. It makes every logically possible distribution equally good and equally bad. Worse than insufficiently determinate, it is radically indeterminate.

C. Kaldor-Hick Efficiency as a Possible Normative Criterion for the Stakeholder Model

The fact that the stakeholder model is radically indeterminate means that, to make the model viable, stakeholder advocates have to produce some generally plausible normative criteria that, when added to the model, would reduce within manageable limits how boards are to distribute value among corporate constituencies. Their extremely prolonged failure to do this—the problem was first identified by Berle almost a century ago—suggests that no such plausible criteria exist. Still, although there is no way to conclusively prove the negative in this case, examining why some of the more obvious candidates fail can illuminate the difficulty of the problem and raise the probability that stakeholder advocates will never be able to supply the needed criteria. Because of the dominance of law-and-economics thinking in corporate law, one obvious candidate is Kaldor-Hicks efficiency. To see why an appeal to Kaldor-Hicks efficiency turns out to be unavailing for the stakeholder model, it helps to begin by comparing the stakeholder model with the Delaware shareholder-wealth maximization model.

As noted above, under Delaware law, directors are permitted to confer benefits on constituencies other than shareholders beyond what such constituencies are legally entitled to receive from the corporation if, as a result, the corporation and thus its shareholders will receive more than offsetting benefits in the long term. More accurately, the present value of the future benefits to the shareholders must exceed the cost incurred by the shareholders in the present from benefiting the other constituency. This means that a decision to benefit another constituency—whether to pay employees severance not legally due, to fund green space open to the public near one of the company’s facilities, or so on—is essentially an investment decision by the board. From a financial point of view, the decision is no different from a decision to invest in a new factory, to purchase certain patents, or to acquire another company. In each case, the question is whether the present value of the resulting expected incremental cashflows exceeds the upfront cost, and the rule of decision in each case is the same: subject to the availability of financing, the corporation should take every project with positive net present value, starting with the project that has the highest net present value per dollar of investment. Under the shareholder-wealth maximization model, business decisions concerning other constituencies are thus standard maximization problems: like other business decisions under that model, these decisions involve maximizing value for shareholders. Moreover, on the assumption that actions by the corporation that would produce costs for unrelated third parties that exceed the benefits to corporate insiders (i.e., all parties with contractual relationships with the corporation, including shareholders) are prohibited by regulation, any decision that maximizes value for the shareholders also maximizes value for society generally.

But business decisions under the stakeholder model do not involve maximizing shareholder value, maximizing value for any other constituency, maximizing value for all constituencies together, or maximizing anything else. They are simply not maximization problems. They are something quite different: they are zero-sum, distributional problems. Under stakeholder theory, the question is how to divide a fixed amount of money among various groups. Economic efficiency provides no guidance at all on such issues. Since every dollar of gain to one group comes at an opportunity cost of exactly one dollar to the other groups, every solution to the problem—that is, every possible distribution of the available money across the various groups—is Kaldor-Hicks equivalent to every other. This may sound like the conclusion reached in the prior section that, under a stakeholder model, every possible business decision is as good and as bad as any other, but in fact the conclusion here is much stronger. The conclusion here is that, even if we supplement the stakeholder model with a normative criterion based on Kaldor-Hicks efficiency—even if we say that the board should distribute value among the various constituencies in the most Kaldor-Hicks efficient manner possible—the stakeholder model remains as radically indeterminate as it was without this criterion. If the stakeholder model can be saved by adding to it some normative criteria that would reduce within manageable limits how boards are to distribute value among corporate constituencies, that criterion will have to be something other than Kaldor-Hicks efficiency.

D. The Results of Hypothetical Bargaining as a Possible Normative Criterion for the Stakeholder Model

Courts sometimes seek to fill gaps in contracts by asking what terms the parties would have agreed to had they bargained over the issue in question ex ante. This is how, for example, Delaware courts apply the implied covenant of good faith and fair dealing. It may seem, therefore, that a board operating under a stakeholder model could resolve problems of how to distribute value among various constituencies by asking what the parties would have agreed to do in such circumstances had they bargained about the matter ex ante. In other words, perhaps the normative lacuna in the stakeholder model can be filled by appeal to some norm about hypothetical consent to hypothetical bargains.

But this idea fails for several reasons. The first is that the resort to hypothetical bargains is really an appeal to the Coase theorem and so takes us back to the Kaldor-Hicks efficiency criterion that we already saw cannot supply the normative lacuna in the stakeholder model. That is, the Coase theorem states that, if transaction costs are zero, then, regardless of how the law initially allocates rights and duties, parties will bargain to an efficient solution, i.e., will transfer each right to the party who values it most highly and each duty to the party who can fulfill it most cheaply. When courts ask what terms the parties to the contract would have agreed to had they bargained over the matter ex ante, they are really asking what terms the parties would have agreed to if transaction costs had been zero (the reason parties do not agree on every possible issue ex ante is that transaction costs are not zero but positive and after a certain point further negotiation produces greater costs than benefits). When they speculate about what terms the parties would have agreed to, courts do so by determining which allocation of rights and duties would have been efficient—which party would have valued a right most highly, which party could have performed a duty (such as avoiding a cost or bearing a risk) mostly cheaply—and they then interpolate terms into the contract accordingly on the assumption that the parties would have wanted to maximize the joint surplus produced by the transaction. The upshot of this is that, when courts resort to hypothetical bargains and ask what contracting parties would have agreed to, they are really asking how rights and duties can be efficiently allocated between the parties. If the goal of the inquiry is to fill a gap in a contract, this is a perfectly sensible thing to do. If the goal is to resolve problems under the stakeholder model, however, then it is an entirely pointless endeavor because, as we already saw, the problem under a stakeholder model involves distributing value among potential recipients, and every possible distribution is as efficient as any other. Hence, since an appeal to a hypothetical bargain is an indirect appeal to an efficiency criterion, and since this indirect appeal to an efficiency criterion is no more able to resolve the indeterminacy problem under a stakeholder model than a direct appeal to the same efficiency criterion could do, an appeal to a hypothetical bargain of the kind used in the law of contracts thus does nothing to fill the normative lacuna in the stakeholder model.

Notice that, when they appeal to hypothetical bargains and the Coase theorem to fill gaps in contracts, courts implicitly ask how parties would maximize the joint surplus arising from the transaction between them, but courts never go further and ask how that joint surplus should be divided between the parties. In the actual contract, the joint surplus is divided by means of the price term. When the court supplements the terms actually agreed to by certain hypothetical terms (i.e., efficient terms the parties would presumably have agreed to), courts never revisit the price term, asking how the parties might have adjusted the price had they also agreed to the hypothetical terms the court is supplying. This is so even though, had the parties negotiated about the matter in question, they may well have adjusted the price term as well. The reason that courts do not inquire about changes to the price term, however, is clear: the price term merely divides value between the parties, and every division of value will be as efficient as every other, and so efficiency provides no basis for any adjustment of the price term. Any adjustment of the price term is a wash between the parties, and so any efforts that the court might expend adjusting the price term would constitute a net loss to society.

But since negotiations over price are analogous to problems of distributing value under a stakeholder model, if we hope to find in hypothetical bargaining a criterion to fill the normative lacuna in the stakeholder model, negotiations over price would seem to be the place to look. What we find, however, is not encouraging. Negotiations about price terms are commonly determined by such factors as the negotiating acumen of the parties, external pressures they are under to reach an agreement within a certain timeframe, the degree of market power each party holds, and so on. Now, directors certainly could conduct inquiries along these lines: they could ask what distributions of value corporate constituencies would have agreed to in given circumstances had they negotiated about such things ex ante, with the results determined by factors of the kinds mentioned. Of course, there would be many problems with such an approach. For one thing, the evidence on the basis of which the directors could conduct such inquiries would often be quite scant (indeed, the relevant information is often the kind of information that parties work hard to keep secret from their contractual counterparties). As a result, directors would be tempted to substitute for hypotheses about what the parties would have agreed to, the directors’ own notions about what the parties should have agreed to, thus covertly smuggling into the inquiry the very normative criteria the inquiry was meant to supply. For another thing, even when the evidence was as plentiful as might reasonably be hoped and the results of such inquiries would be reasonably certain, there is no clear reason why bargains reached on the bases of factors like negotiating acumen and bargaining power are entitled to any normative respect.

Beyond all this, however, there is a fundamental problem with attempting to supply the normative lacuna in the stakeholder model by appealing to what the various corporate constituencies would have agreed to ex ante. In particular, we do not need to ask how the various corporate constituencies would have agreed ex ante to divide the joint surplus created by their relationship because we already know how they actually agreed ex ante to divide that surplus. As noted above, the various constituencies other than shareholders—employees, customers, creditors, suppliers—are in pre-existing contractual relationships with the corporation. If we want to know what these parties would have agreed to ex ante about dividing the joint surplus created by their relationship, we already know the answer full well: they would have agreed to precisely those terms they actually agreed to. Thus, consider again the example of the board deciding whether the corporation should offer a severance payment to an employee beyond what is legally required by any agreement between the corporation and the employee. That the company might terminate the employee is not some unforeseeable circumstance that the parties never considered; on the contrary, it was an eventuality obvious to everyone from the beginning and was thus already provided for in the agreement between the parties. The employee and the corporation have already actually bargained over what severance payments the employee would receive in the relevant circumstances. Since whatever bargain the parties struck on that issue is already reflected in the agreement between them, it is fatuous to ask what the parties would have agreed to regarding severance because we already know what they actually agreed to regarding severance in their actual agreement. Hence, if we try to make decisions under the stakeholder model on the basis of what the parties would have agreed to ex ante, then other constituencies should get exactly what they actually bargained for and nothing more. This would make the stakeholder model equivalent to the shareholder-wealth maximization model or perhaps even less generous to other constituencies than that model, for under the shareholder-wealth maximization model the board may (and should) pay the employees more than is contractually required if doing so produces net benefits for the shareholders in the long term.

Now, this argument relates to hypothetical bargains made ex ante, that is, made at the time the relationship among the constituencies and the corporation began, for that is the time at which the actual bargain between the parties was struck. We can also consider a hypothetical bargain negotiated between the parties ex post, that is, at the time the board has to make its decision about distributing benefits among various constituencies. What would the parties agree to if they were negotiating then? The answer to that question, however, is that since the non-shareholder constituencies have no legal right to receive anything beyond what they are entitled to receive under their contracts with the corporation, the shareholders would be inclined to give the other constituencies their contractual minimums and nothing more. All the bargaining power would be on the side of the shareholders. There is an exception, of course, which is that the shareholders, being rational profit-maximizers, would agree to transfer value to another constituency if the shareholders were also made better off by the transaction. To continue the example from above, if paying the employee severance beyond what was contractually required created a benefit for the shareholders with value in excess of the cost of the severance payment (which it well could, if we consider the corporation’s interests in recruiting and retaining high-quality employees), then the shareholders would of course agree to make such payments. But if this is how the board should make decisions under the stakeholder model, then that model just collapses back into the shareholder-wealth maximization model.

Perhaps we can make progress if we add an additional counterfactual premise to the inquiry. That is, we can ask what the parties would agree to if they were bargaining ex post and the consent of the other constituencies were required before any distributions from the corporation could be made. In other words, we can invest the other constituencies with counterfactual bargaining power they do not really have and ask what would result from negotiations among the parties under such conditions. Of course, we are now envisioning a hypothetical negotiation in which, before the corporation could make any distribution to any constituency, every constituency would have to consent. We would now be asking, in other words, what bargain would be struck among all the corporate constituencies if each of them held a veto over any distributions from the corporation (other than those required by contract). Just how many constituencies there would be is unclear, but for large corporations it would have to number a dozen or more, even if we treated, say, all employees as being a single class even though the interests of employees are far from homogeneous. Worse, no party would have an incentive to agree to a distribution in which it received nothing, and so every party would likely demand at least some value in every distribution. Assuming it is impossible for the parties to make binding side agreements, it seems clear what would generally happen: as in the notorious Polish Sejm with its liberum veto, strategic behavior would set in, negotiations would fail, and no agreement would be reached. If stakeholder theory requires boards to make the decisions that would likely result from such hypothetical negotiations, boards would be paralyzed, for the result of such negotiations would likely be a stalemate. Moreover, even if problems of strategic behavior could be overcome, any bargain reached would still reflect nothing more than the relative negotiating acumen and bargaining power of the parties (or, more accurately with respect to the non-shareholder constituencies, the counterfactual bargaining power with which we are investing such constituencies), and, as noted above, it is difficult to see why a hypothetical bargain based on such things should be regarded as having any normative valence. It would be an odd normative theory that implied that the most cunning negotiator has a just claim on the largest slice of the pie.

E. The Delaware Standard on Apportioning Merger Consideration as a Possible Normative Criterion for the Stakeholder Model

If neither efficiency nor hypothetical bargains can supply the lacuna in stakeholder models, perhaps other conceptual resources drawn from Delaware law can do so. For, there is at least one kind of situation in which directors face decisions that are highly analogous to decisions under a stakeholder model: that is, cases in which an acquirer has agreed to pay in a merger a fixed total consideration to acquire the equity of the target, and the target board has to decide how to apportion this merger consideration among two or more classes of shareholders—e.g., the common shareholders and the holders of a class of preferred shares. As with decisions under a stakeholder model, such decisions are zero-sum distributional problems, for every dollar allocated to one class of shareholders is a dollar less for another class of shareholders. Now, Delaware courts have several times confronted cases in which a class of shareholders has claimed that the board’s apportionment of the merger consideration was unfair. Such cases, therefore, are a promising place to look for wisdom on how a board should make such distributional decisions.

Unsurprisingly, decisions apportioning merger consideration are notoriously difficult for boards to make, for, in exact analogy with decisions under stakeholder theory, every possible apportionment of the merger consideration is as efficient as every other, and every dollar apportioned to one class of shareholders is one dollar less apportioned to another. Consistent with broad and deep trends in Delaware law, a natural move to make in such cases is to shift from substance to procedure. That is, the intuition is that an apportionment of merger consideration is fair if such an apportionment would result from fair procedures leading to the approval of the apportionment. At the extreme would be cases in which each class of shareholders had a right to approve the merger and did so in an uncoerced vote after full disclosure. Such a situation would be analogous to the situation imagined above in which each corporate constituency had a veto over any proposed distribution. Lesser measures would include appointing for each class a committee of independent directors, backed by its own advisors, to negotiate on behalf of the class. Still, such procedures would only amount to facilitating bargaining among the classes of shareholders, and, as we just saw, bargaining (whether real or hypothetical) among the various corporate constituencies does not yield a useful normative criterion to fill the lacuna in the stakeholder model.

As it happens, however, Delaware courts have not generally required such procedures in merger allocation cases, and, in any event, such procedures would at most change the applicable standard of review applied by Delaware courts in considering the board’s decision. There must still be an applicable standard of conduct governing the decisions of directors deciding how to apportion merger consideration across various classes of shareholders. The Delaware courts have reached this issue, it seems, only reluctantly, for the doctrine that emerges from the cases emphasizes that preferred shareholders must receive whatever their contractual rights embodied in the certificate of designation require in the kind of transaction at issue, after which the board’s duty is to maximize value for the common. These principles, the courts believe, will suffice to resolve most cases. Only in the unusual situation in which the certificate is silent on the rights of preferred in the context of a merger, that is, only when “there is no objective contractual basis for treatment of the preferred” in the circumstances of the case, would the board have to make a substantive decision about how to apportion the merger consideration. In those cases, we are told, “the board must act as a gap-filling agency and do its best to fairly reconcile the competing interests of the common and preferred.”

To be sure, this is a bathetic result. The Delaware courts telling directors that they should be fair to all parties is reminiscent of Judge Learned Hand’s famously—but intentionally—fatuous admonition to Justice Holmes: “Do justice!” Nevertheless, like Judge Hand’s remark, the non-result of the Delaware merger–apportionment cases in fact proves very illuminating, and this is so for two reasons. First, the fact that the extraordinarily talented Delaware judiciary, after struggling with the problem of apportioning merger consideration in several cases over the course of more than twenty years, could do no better than to say that directors should apportion merger consideration fairly suggests very strongly that nothing better can be said in such cases. Now, as I noted above, the fact that advocates of the stakeholder model have failed to supply a normative criterion to fill the lacuna in the model does not necessarily show that no satisfactory criterion exists, and since, in this case, as in most cases, it is impossible to prove a negative, there will always remain a possibility that someone will articulate a convincing criterion to complete the stakeholder model. But as the years go by (and we must remember that this problem with stakeholderism was pointed out by Berle nearly a century ago now) and as instances like the Delaware merger–apportionment cases mount (very competent people prove unable to articulate a useful standard), eventually we have to conclude that no satisfactory criterion will ever be articulated. At some point, you have to stop searching for El Dorado.

Second, recall Justice Holmes’s response to Judge Hand. Doing justice, he said, was not his job; his job was “to play the game according to the rules.” Justice Holmes, of course, was contrasting two kinds of norms, substantive moral notions, and legal rules that are supposedly more definite and more certain in their application. I have argued thus far that various kinds of norms commonly used in corporate law—efficiency, the Coase theorem, hypothetical bargains, and so on—all fail to supply the lacuna in the stakeholder model. What is particularly illuminating about the merger apportionment cases is that, when faced with a conceptually similar problem, the Delaware courts resorted to a robust moral notion—fairness—a synonym for “justice” as used in the Hand–Holmes exchange—precisely because the usual legal norms ( Justice Holmes’s rules of the game) will not suffice. In other words, the Delaware merger–apportionment cases suggest exactly the distinction Justice Holmes made in his reply to Judge Hand. If this is right, the implication is that only a substantive moral theory can fill the normative lacuna in the stakeholder model. I thus turn to the possibility of supplementing the stakeholder model with such a theory.

F. A Substantive Moral Theory as a Possible Normative Criterion for the Stakeholder Model

In my view, the two most striking results of the inquiry so far are the following points: the stakeholder model leaves business decisions radically indeterminate, and supplementing the model with some fairly strong assumptions—an efficiency criterion, the Coase theorem, the results of hypothetical bargains of various kinds—do literally nothing to reduce the indeterminacy of the model. This suggests that, to make the model yield determinate answers, the additional assumptions will have to be very strong indeed. The natural move at this point is to supplement the stakeholder model with a robust moral theory such as utilitarianism, some form of Kantian deontology, or a virtue-theoretic system in the Aristotelian–Thomistic tradition—in other words, a fully articulated system of moral philosophy.

The obvious advantage of such an approach is that, whichever moral theory we may choose, adding it to the stakeholder model will definitely achieve the desired level of determinacy. For instance, we could adopt a form of act-utilitarianism and say that, in making a business decision, the board should distribute value among constituencies in the way that maximizes utility among all the members of all the constituencies. Since the members of different constituencies almost certainly derive different amounts of utility from the same amount of wealth (e.g., because of the declining marginal utility wealth or other factors), adopting the utilitarian criterion would allow boards to make decisions under a stakeholder model in a sensible fashion. One decision would be better than another if and only if it produced more utility among the stakeholders in the aggregate. Of course, problems inherent in utilitarian thinking generally would appear in the stakeholder context as well (e.g., it is not clear that inter-subjective comparisons of utility are actually meaningful, we can only guess at other people’s utility functions, it is unclear whether we should be maximizing total utility or average utility among stakeholders, etc.), and there would be some special problems in applying an act-utilitarian moral theory in the stakeholder context (e.g., if the board should maximize average utility, is it average utility on a per-constituency basis or do we do look through the constituencies to the average utility of the individuals composing them? If the latter, in computing average utility, do all individuals of all constituencies get weighted equally, or do individuals in some constituencies count more than those in other constituencies?). Still, such problems should be manageable. At least this would likely be the case if utilitarianism itself is a viable moral theory.

Furthermore, something similar could be said of most other moral theories. Rawls’s A Theory of Justice, for instance, could be adapted to the stakeholder context, and it too would provide criteria whereby stakeholder decisions could be rationally evaluated as good or bad, better or worse. Presumably, a version of Rawls’s difference principle would apply to the board’s decisions, and so a decision would be permissible only if the least well-off constituency was not made better off by any other decision. Similarly, we could adopt an Aristotelian–Thomistic system and require that directors make, in each set of circumstances, the decision that best advances the common good of society as understood in such system. Basically, every moral system seriously entertained in contemporary philosophy could be adapted to supply stakeholder theory with normative criteria by which boards could rationally evaluate the various options available to them in making a business decision.

But therein lies the problem. Directors are not moral philosophers, and most do not have sophisticated views as to exactly which system of moral philosophy is best. Indeed, many very capable directors are probably unaware that there are multiple systems of moral philosophy. Even if we could assemble, say, eleven individuals who were all well-versed in moral philosophy and who would otherwise compose a competent board of directors for a public company, it is most unlikely that they would all agree about which moral theory ought to be applied in making business decisions. And, of course, the differences between moral theories would often matter quite a bit. If some directors were utilitarians, others Rawlsian deontologists, and yet others virtue theorists, they would tend to disagree systematically on a host of issues and so often be unable to reach agreement on how to make stakeholder decisions. In short, we have the corporate version of the dilemma in Plato’s Republic: until philosophers become directors or directors become philosophers, the indeterminacy of the stakeholder model cannot be cured by a resort to a substantive moral philosophy.

G. Decisions Under the Stakeholder Model as Political Decisions

Thus far the argument has shown that the stakeholder model is radically indeterminate unless supplemented by some normative criteria, and that, albeit for different reasons, neither economic nor moral criteria can supply the lacuna in the model. This suggests that, under a stakeholder model, directors would make business decisions in an entirely unprincipled manner. This is a surprising and brutal conclusion, but I find confirmation for it in one of the rare points of agreement between the leading advocates of stakeholder theory and their severest critics. Thus, Dodd thought that what managers should do would ultimately be guided by public opinion:

Such a development of business ethics which goes beyond the requirements of law and beyond the dictates of enlightened self-interest … can happen only if managers of [public] corporations have a degree of legal freedom to act upon such an attitude without waiting for the unanimous consent of the stockholders… . If we think of [the corporation] as an institution which differs in the nature of things from the individuals who compose it, we may then readily conceive of it as a person, which, like other persons engaged in business, is affected not only by the laws which regulate business but by the attitude of public and business opinion as to the social obligations of business.

Professors Blair and Stout are even more explicit that directors acting under a stakeholder model are essentially political actors responding to political pressures. They write that corporate directors

enjoy considerable discretion in deciding which members of the corporate coalition receive what portion of the economic surplus resulting from team production. Although the board must meet the minimum demands of each team member to keep the coalition together, beyond that threshold any number of possible allocations among groups is possible. Thus, the returns to any particular corporate stakeholder from participating in the corporation will be determined not only by market forces but by political forces.

As example of what they mean, Professors Blair and Stout suggest that the rise of institutional investors and the decline of labor unions in the last third of the twentieth century explain why boards became more likely during that period to allocate value to shareholders rather than employees: shareholders simply had more clout, more raw power against directors, than did employees.

But in recognizing that stakeholderist decisions would become political decisions, the advocates of stakeholderism, ironically but significantly, find agreement among their critics. Thus, Manne thought that trends in corporate social responsibility “follow the social and economic concerns of the broader public,” and Friedman concluded that “the doctrine of ‘social responsibility’ taken seriously would extend the scope of the political mechanism to every human activity.” Berle, too, understood this perfectly all the way back in 1932. Speaking of the stakeholder model, he said,

The only thing that can come out of it, in any long view, is the massing of group after group to assert their private claims by force or threat—to take what each can get, just as corporate managements do. The laborer is invited to organize and strike, the security holder is invited either to jettison his corporate securities and demand relief from the state, or to decline to save money at all under a system which grants to someone else power to take his savings at will. The consumer or patron is left nowhere, unless he learns the dubious art of boycott. This is an invitation not to law or orderly government, but to a process of economic civil war.

What Blair and Stout describe as politics, Berle describes as war, which makes perfect sense if we accept Clausewitz’s famous dictum that war is merely the continuation of politics by other means. One’s response to this conclusion will depend on one’s antecedent views on political decision-making. In my view, however, only someone with an absurdly optimistic view of politics would want to expand the process of political decision-making to corporate boardrooms.

That leaves one important question unanswered. If, as I have argued, the stakeholder model leaves business decisions radically indeterminate, how is it that there is so much agreement among public company directors, institutional investors, proxy advisors, politicians, members of the corporate bar, management consultants, and corporate law scholars on how corporations ought to treat stakeholders? After all, there seems to be broad and deep agreement that companies should reduce their carbon emissions, increase diversity (especially racial and gender diversity) at all levels of the corporation (but especially at the most senior levels), promote economic equality, combat systematic racism, and so on. If the stakeholder model is indeterminate, how can there be so much agreement on such a particularized agenda? The answer is straightforward. For, as its advocates and critics agree, business decisions made under the stakeholder model are the product not of rational deliberation based on some set of normative criteria internal to the model but are rather the outcome of political and other non-rational forces operating on directors. Hence, the reason for the observed broad agreement must be that, at the current time, a sizeable majority of the individuals in the socio-economic class from which public company directors, partners at elite law firms, senior officers at institutional investors and proxy advisory firms, politicians, management consultants, and academics are drawn overwhelming favor one particular political agenda—i.e., a largely progressive political agenda that emphasizes issues such as climate change, environmental concerns, racial and gender diversity, economic equality, systematic racism, and so on. There is strong empirical evidence that this is so. In other words, it is not that a belief in stakeholderism, because of norms internal to stakeholderism, leads to support for the current stakeholderist agenda; it is that support for such an agenda leads to stakeholderism, because, under the stakeholder model, directors will be free to, and will be encouraged to, follow such an agenda. In any case, however, it is only the broad support among the right class of people for a particular political agenda that produces broad agreement about which stakeholderist decisions are good and which bad. The stakeholder model itself is perfectly vacuous.

Moreover, precisely because the model is an empty vessel, any political agenda whatsoever can be poured into it. That is, under different social conditions, the political and moral content of the stakeholder movement would be different. Henry Ford might be thought of as the grandfather of stakeholderism, and stakeholder theorists often mention his concern for employees and customers. But about the time Ford was losing Dodge v. Ford Motor Co., he also began publishing a weekly column entitled “The International Jew: The World’s Problem.” There is nothing in the stakeholder model that is incompatible with Ford’s odious racist and anti-Semitic views because there is nothing in the stakeholder model at all. It can become the tool of any political agenda, whether the arguments and reasons underlying that agenda be good or bad, sound or unsound, reasonable or unreasonable. To be adopted by boards operating under a stakeholder model, a political agenda need only be popular among the kind of people involved in corporate governance.

Indeed, the political agenda involved need not even be well-supported in society more broadly. That is, stakeholderism provides a way of advancing a political agenda that has lost in the democratic process. Take climate change, a key issue for stakeholderists and ESG advocates more broadly. It is clear beyond dispute that reductions in carbon levels large enough to have a significant effect on the global climate would require action not by one company or even one country but by substantially all large companies in substantially all industrialized countries; if some or even only a good majority of businesses emitting high levels of carbon reduce their emissions, the net result would likely be trivial, for others can and likely will free ride on their efforts. The obvious solution is regulation requiring all companies to reduce their emissions very substantially, the solution that has been used successfully to control other forms of pollution for the last fifty years and the solution climate change activists actually favor. But such regulation does not have broad support among the public, and such regulations have not been implemented in any developed country (whatever regulations have been imposed fall far short of what most climate change activists believe necessary). Stakeholderism provides a possible answer to this problem: it may be possible to convince corporations to do voluntarily what they are not required to do legally.

H. Conclusion

I began by noting the objection commonly made by critics of the stakeholder model that a board’s decisions under such a model would be standardless. I amplified that criticism, arguing that it is significantly understated and that, in fact, under a stakeholder model, no decision permitted by the model is any better or any worse than any other. I expressed this conclusion by saying that the stakeholder model leaves business decisions radically indeterminate. I then considered various additional normative assumptions advocates of stakeholder theory might make in order to reduce the indeterminacy of the stakeholder model within manageable limits, and I found all of them wanting. An efficiency criterion, the Coase theorem, various kinds of hypothetical bargaining, and Delaware doctrines about the apportionment of merger consideration all fail to reduce the indeterminacy of the stakeholder model. The only apparent way to reduce that indeterminacy is by assuming that directors should make business decisions in accordance with some particular moral theory. The problem with that solution is that directors are not moral philosophers and we cannot reasonably expect that they will become moral philosophers, and until they become moral philosophers—and moral philosophers all of the same stripe—they cannot resort to moral philosophy to make business decisions together in an effective manner. The upshot is that, under the stakeholder model, decisions directors might make would be entirely unprincipled. That surprising conclusion, I argued, is actually confirmed by a rare point of agreement between stakeholder advocates and their critics, both of whom conclude that, under a stakeholder model, business decisions would become political decisions, the product of whatever pressures, threats, or inducements various constituencies could bring to bear on directors to decide to allocate value to them rather than their rivals. Any apparent consensus about how corporations ought to behave under a stakeholder model flows not from any normative criteria internal to that model but from the fact that individuals involved in corporate governance tend to come from the same socioeconomic class and thus tend to share the same political and social values.

To reiterate, however, the key point is that the stakeholder model of corporate governance lacks any normative criteria by which business decisions may be evaluated as better or worse, right or wrong. Moreover, there seems to be no prospect of supplementing the model with plausible criteria in order to allow it to produce the needed evaluations. Berle pointed out this problem nearly a hundred years ago, and still the advocates of the stakeholder model have yet to provide a plausible solution. As a result, the model remains an empty vessel for whatever political or moral agenda may command sufficient support to be implemented, whether that agenda is wise and good or foolish and wicked. Even if the agenda being advanced is wise and good, however, that does not make stakeholderism a good form of corporate governance. Indeed, if by corporate governance we mean a model of how business decisions should be made and corporate affairs managed, then stakeholderism is not a good form of corporate governance because stakeholderism is not a form of corporate governance at all. It is not a standard or a norm, but the absence of all standards and all norms.

I thank for helpful discussions Stephen M. Bainbridge, Lucian Bebchuk, Henry N. Butler, Charles M. Elson, Richard A. Epstein, Lisa Fairfax, Thomas P. Gallanis, Benjamin Johnson, Charles Korsmo, J. Travis Laster, Saul Levmore, John O. McGinnis, Jennifer L. Miller, Seth Oranburg, Jonathan Povilonis, Leo E. Strine, Jr., and Ashley Walter. I also thank Peter Sotos for his excellent work as my research assistant.