THE CONTOURS OF CORPORATE MORAL AGENCY

Abstract

This article defends skepticism about the moral agency of corporations, arguing that even if we accept the idea that there exist group moral agents, it makes little sense to suppose that the corporation itself can qualify as such an agent. The discussion considers and rejects arguments from Philip Pettit, Peter French, and Michael Bratman. It concludes that we should not criminally prosecute corporations.

INTRODUCTION

Can massive business corporations—e.g., Amazon, ExxonMobil, or Volkswagen—do wrong? I doubt it. I argue that in cases in which the question of corporate moral agency is most important and most interesting, cases that involve putative corporate responsibility for substantial wrongdoing, it makes no sense to recognize corporate moral agency. Some disagree. For example, Philip Pettit, Peter French, and Michael Bratman each propose a rationale for recognizing the corporation as a group agent.¹ Their arguments at best support recognizing comparatively small groups within the corporation as group agents, but do not support recognizing anything resembling the corporation itself as a group agent, I will maintain. We should, then, be skeptical about corporate moral agency. This skepticism has practical implications. The rejection of corporate moral agency implies that continuing to treat corporations as responsible by blaming them and even subjecting them to legal punishment dilutes and degrades our practices of blame and punishment. My argument against corporate moral agency suggests that we should stop criminally prosecuting corporations.

Proponents of corporate moral agency commonly argue for the possibility that groups of people can have the mental states required for agency, including beliefs, desires, and intentions; they then argue that corporations can be such groups.² Their opponents counter that groups, including corporations, cannot be agents because they lack the mental states required for agency.³ I won’t engage with these arguments or suppose that groups cannot be agents. Sensibly doing so would require settling ongoing and recalcitrant disputes in the philosophy of mind. If the assessment of corporate agency depends on a resolution of these disputes, then we may have a very long wait. So I take a different approach. I concede the possibility of group agency and raise no doubt that some groups can have intentions, beliefs and other mental states, but argue that in typical cases of putative corporate wrongdoing, the corporation itself cannot be such a group.⁴

This essay proceeds as follows. In Sect. II., using the example of the Volkswagen Dieselgate scandal, I present a prima facie case against corporate moral agency, arguing that no agent corresponding to or constituting the corporation itself can be responsible for wrongdoing in Dieselgate. In Sect. III.,  I explore the limits of Philip Pettit’s defense of the idea of corporate moral agency, arguing that whatever agent he identifies cannot be the corporation itself. In Sect. IV.,   I argue against Peter French’s attempt to construct a corporate moral agent on a foundation of a corporate internal decision structure. In Sect.V.,  I argue against Bratman’s attempt to conceive corporate agency as deriving from the hierarchy of reasons connecting organizational leaders to their subordinates, suggesting that his account identifies a diverse plurality of moral agents, none of which can be the corporation itself. In the Sect. VI.,  I conclude that the problems I find in find throughout this essay in locating corporate moral agency are nothing idiosyncratic to the theories I discuss, but that they instead support deep skepticism about the possibility of corporate moral agency and about the idea of punishing the corporation.⁵

A PROBLEM WITH CORPORATE MORAL AGENCY

The Dieselgate scandal illustrates the issues in corporate moral agency. Acting under the direction of corporate management, engineers at Volkswagen developed “defeat devices” for their diesel cars, which were triggered during emissions testing to misrepresent the amount of toxic fumes emitted by Volkswagen diesel engines.⁶ The defeat devices were used to trick government agencies into underestimating the amount of toxic pollutants that Volkswagen cars produced. Arguably Volkswagen defrauded governments around the world and was thereby able to sell cars that had a far worse impact on public health than cars that satisfied legal limits on pollution would have had. Presumably some people died and some people became otherwise ill as a result of the corporation’s actions. Who or what was the group agent responsible for Dieselgate? One might answer that the Volkswagen Corporation itself was responsible.

The thought that Volkswagen itself is a responsible group agent in Dieselgate has its appeal. The engineers and managers that actively perpetrated Dieselgate were no rogues. They were corporate officers and enlisted employees acting in their official capacities even when they did bad things or made bad choices. As long as they act only through the powers granted to them as members of the corporation, one might conclude that their behavior is the group’s behavior, that their wrongs are the corporation’s wrongs, and that the corporation itself is a responsible corporate agent. In their account of corporate action, William Laufer and Alan Strudler embrace this idea:

corporate … action may be found in: (1) agents whose actions and intentions are related to each other in such a way that they assume the characteristics of the corporate form; (2) agents whose status in the organization is such that their actions and intention are those of the organization; and (3) aspects of the organization, such as policies, goals, and practices, that reflect not merely the sum total of individual agents' intentions, but instead attributes and conditions of the corporation that make it possible for these agents to cooperate and collaborate in legally problematic ways.⁷

Yet if it is a fact that a corporate manager’s action just is the action of the corporation, there must be something that makes it so, something that grounds that fact. It cannot be simply that corporate leaders see themselves manifesting the corporate intention, a commercial version of Louis XIV’s declaration, “L’etat c’est moi.” I will conclude that ordinarily nothing makes it so that morally significant managerial action just is the action of the corporation, that responsible moral agency generally does not inhere in the corporation itself, but instead in its leaders. We should be skeptics about corporate moral agency.

Volkswagen, if it is an agent responsible for doing wrong, is a group agent, dependent on people for its existence. In the absence of people, there is no Volkswagen. Who are these people that compose the Volkswagen corporate agent? What are the contours of the Volkswagen corporate agent? These questions are important, because without an answer the corporate moral agent remains a thin abstraction whose status is impossible to assess. Two obvious candidates for Volkswagen membership are its managers and its employees broadly conceived. As best I can tell, among philosophers the most widely adopted view for corporate membership includes the spectrum of corporate employees and is not restricted to managers, a view that I accept for purposes of assessing theories of corporate moral agency.⁸ The alternative robs corporate agency of any content. If the group agent that is the corporation just is the group agent that is corporate management, then the concept of corporate moral agency becomes redundant: after reaching a judgment about management responsibility, it would be pointless and repetitive to then judge the corporation itself as responsible. Consider Dieselgate. No doubt Volkswagen management, for example, had some responsibility for Dieselgate. But if Volkswagen management just is the Volkswagen Corporation, then once one blames management, one adds nothing by blaming the corporation. It is only if the corporate agent is composed of something beyond corporate management that the idea of corporate agency gets distinctive content, that it is at least possible to coherently judge that both the corporation itself and corporate management are responsible for wrong. Out of charity to proponents of the idea of corporate moral agency, then, one must regard them as conceiving of the corporate agent as composed at least of the spectrum of corporate employees, not management alone.⁹ It does not follow, of course, that the corporate moral agents exist. I will argue that they do not.

Later I develop objections against leading arguments for corporate moral agency. Now I want to say more generally why I resist the idea. I maintain that ordinarily the employees of the corporation do not have the right sort of relationship with corporate leaders for it to make sense to say that the actions of the leaders are the actions of the group as a whole. Consider an analogy. Three pirates in times past take control of a ship powered by rowing, enslave its crew, and force them to work the oars. The three pirates compose a collective agent, which may be discontinuous with its pirate membership, let us concede for the sake of argument. Now suppose that the pirates deliberate among themselves, agree to turn the ship southward, and then one of them actually turns the wheel, sending the ship southward, while the crew is busy obliviously scrubbing the ship’s deck. I submit that the whole group of people on board cannot form the basis of the group that turned the ship southward; only the pirates turned the ship southward. The crew, who were the vast majority of people on board, did not participate in the decision to go south, did not authorize the pirates’ action or authorize anyone else to authorize the pirates’ action, and did not unwittingly move their bodies in ways that caused the ship to head south; they were busy cleaning the deck and were otherwise oblivious to the pirates’ action; the crew was therefore no part of an agent responsible for turning the boat southward. Suppose, now, that in addition to members of crew, there were other people on board, locked in the brig. Are they also part of the agent that turned the ship? Is the cadaver on board part of the agent? The parrots and rodents? The only principled reason to find that someone on board helps to compose a group agent is that her will suitably connects with a group’s goals or plans, though it is tricky specifying which connections are suitable. People in the position of this ship’s crew, like the cadaver and the rodents, lack the proper connection, I maintain. I do not wish to suggest that any of the prominent proponents of corporate moral agency would think that the crew or the people in the brig form part of a group from which can be derived a group agent that turned the ship southward. Instead, I suggest that their positions leave them no principled basis for denying that members of the crew and people in the brig form such a group.

A corporation is not a pirate slave ship, but there are similarities. Like the pirate slave ship, large corporations are authoritarian institutions in which typically subordinates are neither consulted nor informed about significant choices.¹⁰ Their work assignments are tiny insular parts of the corporate enterprise. At Volkswagen, they install windshields, clean floors, monitor the supply of parts coming from outside suppliers, serve food in the commissary, and so on. The vast majority of Volkswagen employees sign on to execute these tasks in exchange for income and benefits, but sign on to nothing more than executing these tasks, and have no reason to do anything more. They were not invited to participate in the emissions scheme, which was hidden from them. Why think that a group moral agent, derived from the spectrum of Volkswagen employees, committed Dieselgate misdeeds any more than the whole spectrum of people on board the pirate ship form the basis of a group that steered that ship? I will maintain that no good reason supports that thought.

Before proceeding, I will make two terminological points. First, it is linguistically challenging to discuss the relationship between a firm’s employees and the corporate group agent as proponents of corporate moral agency conceive that relationship. The relationship can be understood as either supervenience or reduction. Consider supervenience. Christian List and Philip Pettit capture the essence of much corporate moral agency theory when they say that the group agent supervenes on the people in the firm.¹¹ Supervenience is a difficult concept. Generally, one set of properties supervenes on another if there cannot be changes in that first set without changes in the second. So the color of a tomato supervenes on its molecular structure; change in color cannot occur without change in molecular structure; color is distinguishable from molecular structure but depends on it. List and Pettit would say the same thing about the corporation: it is distinguishable from corporate members but dependent on them. This kind of dependence differs from composition, as that term is ordinarily used. Just as the color red isn’t composed by any particular molecular structure, the corporation as a group agent isn’t composed of the people who work there, on List and Pettit’s view. As far as I know, we do not have a word that conveniently expresses the supervenience relation across a spectrum of grammatical contexts. It would be linguistically awkward to ask which people are supervened upon by the corporation, a question that I want to ask. I do not wish to beg any questions about whether there is a supervenience relation between the corporation and some of the people who work there, but I do wish to raise questions about which people putatively stand in the supervenience relation. Things are made more linguistically challenging by the fact that not all group agency theorists embrace the concept of supervenience. Consider, then, reductive accounts of group agency. Rather than seeing the group agent as supervening on group members, some philosophers, e.g., Bratman, see a reductive relationship, saying, roughly, that a group can be reduced to its members, or that a group is nothing but its members. There are, then, both reductive and supervenient accounts of group agency. To make matters linguistically easier, I will engage in stipulation: when I ask which people compose the corporation, I should be understood as asking which people stand in either in the supervenience relationship or the reductive relationship to the corporation.¹²

My second terminological point is that although I am concerned to discredit corporate moral agency, I will speak of a corporation engaging in wrongful actions. I do so for convenience. Ultimately, I think it is not the corporation itself but some of the people associated with it who engage in wrongful actions, and that speaking of the corporation itself as responsible is most charitably regarded as using shorthand for identifying the actual wrongdoers. But unraveling this shorthand would be a distraction here.

DERIVING CORPORATE MORAL AGENCY

In defense of the idea of corporate moral agency, Pettit argues from premises about the discursive dilemma, a problem in social choice that arguably confronts corporate decision-makers, to the conclusion that a corporation exists as a group agent discontinuous with and autonomous from the people within the corporation.¹³ The discursive dilemma, on his account, shows that a corporation can form representations and purposes that are not systematically derivable from the independent, individual representations and purposes of its members. We must countenance a group agent, e.g., the corporation, distinguishable from group members, as the bearer of these representations and purposes, as Pettit sees it.

The most radical critique of Pettit denies that the argument from the discursive dilemma establishes the existence of any distinctive group agent.¹⁴ For purposes here, nothing so radical is required. I will assume that Pettit is correct in his argument for the existence of group agents that are discontinuous with their members. The concession facilitates focus on the question of whether Pettit’s group agent should be identified with the corporation itself. In fact, Pettit never says that his argument from the discursive dilemma helps define the contours of any particular group agent. That argument purports to establish the existence of group agents, not their contours or their membership.

If the argument from the discursive dilemma does not fix group membership, what does? Which people, on Pettit’s account, compose Volkswagen? The issue is important. If Pettit’s group agent is nothing but corporate management, then, as my earlier argument implies, on his account the idea of corporate moral agency is redundant: groups of corporate leaders can be responsible agents, not the corporation itself, and it makes no sense to say that both management and the corporation are responsible for some wrongful act. Perhaps, recognizing the issue, Pettit seemingly acknowledges that a corporate group agent must be composed of something more than corporate leaders and decision-makers.¹⁵ He explains that for “other members of the corporation” to be part of the relevant group agent, they must have “committed themselves” to the deeds of corporate leaders:

the other members of the corporation ascribe that authority to them, implicitly or explicitly committing themselves as individuals to rally behind the words of their spokespersons on any relevant issue; they treat those words as expressions of attitude that they have to live up to, on pain of corporate failure, in their actions as corporate members.¹⁶

These ideas about commitment, if successful, would expand corporate membership beyond actual corporate decision-makers. Pettit could thus say that even if an employee is not directly involved in the choice to engage in Dieselgate misdeeds, she might still be tied to the fraud by virtue of her commitment to the authority of Volkswagen executives. And surely all corporate employees commit themselves to the authority of their “superiors,” at least to some degree. Otherwise a corporation would be a chaos.

On Pettit’s view, then, one thing what would makes Volkswagen a responsible moral agent is that Volkswagen employees generally commit themselves to their superior’s authority. That seems unpromising. A Volkswagen employee has no reason to commit herself to anything but the most limited authority of her superiors. A Volkswagen electrician, for example, will accept her foreperson’s order to work on a particular wiring project. She will even accept a superior’s orders to leave Volkswagen premises. But the authority involved is limited to the kind of tasks that an electrician would execute in the ordinary course of an electrician’s employment. She need not accept orders, as part of her job, to serve lunch in the corporate commissary or to clean corporate toilets. She could reasonably complain that she was not hired to do those tasks, even though of course she might get fired for complaining. Neither is it any part of the electrician’s job to even consider granting authority for Volkswagen executives to choose to produce a certain model of car, or to change its supplier of tires. Indeed, corporate employees have no authority to cede on such matters. More generally, it is implausible that a significant portion of Volkswagen employees “commit… themselves as individuals to rally behind the words of their spokespersons on” the acts committed in Dieselgate or indeed the acts of Volkswagen leaders more generally. A commitment, I assume, is typically something resembling a promise, if not simply a promise. Pettit offers no evidence that employees generally make such a promise, and I can conceive of nothing that employees generally do that can be reasonably construed as making such a promise. More important, there is no reason to expect employees to make such promises. An employee gains nothing material by promising to do anything beyond her job narrowly defined, and there is no moral reason for an employee to make such a promise. The best reason for supposing that the vast majority of corporate employees do not commit themselves to the firm is that they have no reason to do so.

Even if Volkswagen employees somehow committed themselves to adhere to the choices of corporate leadership, there are limits on the scope of such commitment. Employees would be foolish to commit themselves to clearly criminal activity. If Volkswagen leaders started to murder dissatisfied customers, for example, they could not plausibly say that they were authorized by the commitment of their employees in doing so. Similarly, leaders could not say that they were authorized to engage in the criminal activity of Dieselgate. So Pettit’s account cannot tell us that we should regard Volkswagen as a group agent responsible for Dieselgate. And that result is significant. Dieselgate is no outlier. Corporate wrongdoers are commonly not so brazen as to make their wrongdoing conspicuous. When a large corporation engages in wrongdoing, it is common that the wrongful action is unknowable by the vast majority of its employees. Consider, for example, the Ford Pinto case, in which Ford was criminally prosecuted for homicide because of deaths caused by defective gas tanks in some of its cars.¹⁷ No doubt the vast majority of employees who worked at Ford knew nothing about this defective design and would find it outside the parameters of their job description to seek out information about emission control practices. Or consider the BP Deepwater Horizon oil spill into the Gulf of Mexico.¹⁸ It killed 11 people and did enormous damage to the natural environment of the Gulf. BP acknowledged criminal responsibility. But the firm’s wrong involved overlooking safety standards in drilling procedures, a technical matter about which most BP employees could have known nothing. Cases like the BP oil spill, the Ford Pinto explosions, and Dieselgate are no rarity. To the contrary, it is natural for firms that do wrong to keep things quiet. So Dieselgate is no outlier case, and is relevant to the assessment of Pettit’s account.

One might concur in my skepticism so far about Pettit’s argument for corporate agency in cases like Dieselgate, but object that it has limited relevance for corporate moral agency more generally. Because Dieselgate was done secretly, perhaps the case is not perfectly representative. Perhaps many corporate wrongs do not involve this kind of secret criminality. Imagine, then, a firm that produces weapons for export, and that all employees know that these weapons will be used to harm and suppress a vulnerable ethnic minority in a foreign land, but that their actions violate no criminal law. Or consider Purdue Pharma, a company that was held criminally liable for causing harmful addiction among users of the Oxycontin it produced.¹⁹ Suppose that Purdue Pharma employees were aware, or at least had sufficient reason to be aware, of this illicit cultivation of drug use, and suppose that they should have done something to resist Purdue Pharma’s action: they should have protested that action, or even resigned in protest. Instead they acquiesced, perhaps contributing to the success of Purdue Pharma by doing their jobs. Suppose, then, that the bulk of Purdue Pharma employees were individually blameworthy. One might then say that in cases such as this, because the corporate wrong is open and not secret, employees tacitly commit themselves to the corporation’s actions, thus authorizing the actions and forming part of a responsible group agent. If so, one might think that for many cases of corporate wrongdoing, Pettit is correct in invoking tacit commitment of employees to identify the whole corporation as a responsible moral agent.

This last defense of Pettit distorts the ideas of commitment and authorization. Nothing about your working at a firm suggests your endorsement of its activities or that you authorize its wrongful acts. You may work at the weapons firm and hate what it does, engaging in public protests against its activities, but keep your job there anyway because you desperately need the paycheck. Borrowing terminology from Scott Shapiro, let’s call a worker alienated when she comes to work to execute her contractually defined job tasks in exchange for compensation, but she does not commit herself to any broad goals or policies of the firm.²⁰ Also, let’s elaborate the concept of alienation to include Elizabeth Anderson’s idea that workers are neither consulted nor informed about significant corporate actions. If you are an alienated worker, then perhaps in continuing to work at the weapons firm, you do wrong. Perhaps you should quit. But I do not see that you have done anything more to authorize the firm’s weapons production than many people not part of the firm. Notoriously Oscar Schindler ran munitions plants for the Nazis while employing Jewish workers, thus secretly acting against Nazis to save Jews. The fact that his work was contracted to aid the Nazis, and in fact provided some aid to the Nazis, in no way shows that he was committed to their cause or authorized their criminality. One might nonetheless insist that Schindler was part of a group agent that perpetrated Nazi evils. But nothing in Pettis’s account warrants that position and I can imagine no argument that would warrant that position.

No doubt Volkswagen employees recognize that certain people in the organization have authority to make decisions that bind the organization. That does not require them to support the decisions of such authority. After all, even strangers or outsiders to the firm can recognize the firm leaders have authority to make decisions, but that does not require them to support or endorse these decisions, nor does that recognition involve any commitment to the firm. Volkswagen workers, like total strangers to the firm, were not committed to the particular acts that constitute Dieselgate. Pettit’s account can do nothing to confirm the existence of corporate moral agency in Dieselgate.

Pettit admits that there is more to a corporation than its leaders, that there are “other members” of the corporation. But he needs some normative bond to attach these other members to the corporation. So Pettit offers the idea that the “other members” that compose the firm do so by committing themselves to the authority of the firm. I have argued for skepticism about such commitment. If there is such a phenomenon as corporate moral agency, it requires a normative bond among employees other than commitment.

FRENCH ON THE STRUCTURE OF CORPORATE MORAL AGENCY

Writing twenty years before Pettit, French argued that the corporation is an agent that can act autonomously and bear responsibility for doing wrong.²¹ Like Pettit, French maintains that the corporation has beliefs, intentions, and desires that cannot be derived from group members. Unlike Pettit, French does not employ the discursive dilemma or social choice theory more generally to establish the autonomous agency of the corporation, nor does he see employee commitment as a way to specify the contours of corporate moral agency. In explaining such agency, French relies on putative facts about how a corporation functions. The corporation acts through its employees, he explains, and does so when corporate employees act in ways that conform to a corporate internal decision structure (CID), which consists of an organizational flow chart that specifies roles for the people working at a firm, and a statement of fundamental corporate policies. The idea of a CID thus forms the core of the French’s account. The idea that the structure of the corporation is essential to the agency of the corporation takes corporation agency theory in a direction that differs from Pettit’s reliance on the notion of commitment to fix the contours of corporate agency. The problem with French’s appeal to a CID, I will argue, is that he never explains its connection to people in the firm. His CID lacks normative force; French cannot explain the normative bond that holds organizational members together in a way that forms a group agent.

Despite the attention scholars devote to his CID, its nature remains elusive. One might regard it as a document, writings on paper or the digital equivalent. Suppose, then, that corporations have CID’s constituted by organizational flow charts and statements of policy that contain sufficient detail to allow identification of authorized employee conduct. It seems doubtful that the existence of a CID will suffice to ground corporate responsibility. Drawing up a document is easy. Doing so does not make the people to whom the document refers part of a responsible corporate entity, however. Suppose that I draw up a CID. It includes an organizational flow chart that specifies the roles of members of our organization: Boris Johnson, you, and me, even though we don’t know one another. I write down in an organizational chart the policies of this corporation, which give me discretion to commit a crime. Then I go out and commit the crime of painting graffiti on Pasadena City Hall. Did our group engage in this criminal act? Arguably not. This CID is a sham. A mere document, drafted in the absence of an appropriate normative connection to the people who purportedly compose the organization, cannot ground a claim of organizational responsibility. A CID by itself cannot explain corporate agency or solve the problems of identifying corporate responsibility. A CID matters only if it has normative authority.

Why, then, think that a CID is practically relevant rather than an empty formality? French explains:

When an action performed by someone in the employ of a corporation is an implementation of its corporate policy, and accords with its procedural rules, then it is proper to describe the act as done for corporate reasons or for corporate purposes, to advance corporate plans, and so as an intentional action of the corporation.²²

French seemingly invokes here the employment relationship: acts done by “employees” in conformity with a CID are acts done by the corporation itself. It seems doubtful, however, that the employment relationship suffices. Suppose that Gustav is an arborist on the payroll of Volkswagen whose job it is to trim trees. Then trimming trees, done in conformity with the firm’s CID, is a Volkswagen corporate act. Now suppose that Volkswagen decides to outsource tree trimming. It fires Gustav. Then Gustav starts a tree trimming firm, and Volkswagen contracts with his firm to trim its trees. Tree trimming is no longer a Volkswagen corporate act, if we take seriously French’s reliance on the idea of employment. Gustav does the same act of tree trimming as an employee and as a contractor, and all tree trimming occurs under direction of the firm. It seems arbitrary to say that tree trimming is a corporate act while Gustav is an employee but not while he is a contractor. I am not sure how French would respond. He might say that employment is an element of a sufficient but not a necessary condition for corporate agency, and that Gustav’s act, because of the involved contract, is an act of Volkswagen corporate agency. Or French might say that Gustav’s acts switch between being acts of corporate agency and being acts outside the corporation, depending on whether he works as an employee or a contractor. His position has an unsettling arbitrariness.

The arbitrariness in French’s position is a symptom of a problem. The position seems arbitrary because it lacks an account of corporate membership that can explain why the fact of membership warrants the inference that corporate members compose a responsible moral agent. French identifies corporate membership in terms of employment, but never explains why being a corporate employee, when coupled with being listed in a CID, leads to even partially composing a responsible agent. Employees do not swear an oath of allegiance to the firm. What make them part of the corporate moral agent? Of course, one can be part of a group without swearing allegiance to anyone. An infant is part of a family though she cannot swear allegiance, for example. But it hardly follows that when the other members of the family commit a crime, the infant should be regarded as partly composing the agent of that crime, or as someone upon whom the agent supervenes. We need more than a CID coupled with the fact of the employment relationship to explain how a responsible agent can emerge at the corporate level. Bratman offers a promising attempt to identify this additional element.

BRATMAN ON THE STRUCTURE OF CORPORATE MORAL AGENCY

Like French, Bratman does not rely on ideas of commitment or consent to identify the contours of a corporate agent, but instead invokes corporate structure to explain those contours . He offers a way to bind employees to that structure by leveraging the idea of a reason for action to specify the contours of a corporate agent.²³ While French finds this structure in the corporate CID, Bratman finds the structure in the hierarchy of reasons connecting organizational leaders to their subordinates.

On Bratman’s account, there are two very different kinds of reasons for action that a corporate member might have, depending on whether the person is in the kernel or the penumbra of the corporation. People in the kernel are those that set the rules of the organization and act for reasons stemming from these rules; they “share a policy in favor of the group’s conformity to” the rules put forward by the people in the kernel. People outside the kernel are in the penumbra. They intend to conform to the rules put forward by people in the kernel, even if they do not favor or feel committed to these rules. People in the penumbra need not accept the reasons for action that reflect corporate policy as members of the kernel do, but they will nonetheless accept “kernel-induced reasons.” For example, a penumbral member’s choice to execute her job role will be induced by instructions coming from the people in the kernel, which provide that a member must execute her job role in order to get her paycheck.

I will not defend skepticism about whether the people in the Volkswagen kernel may compose an agent that commits the acts constituting Dieselgate. My doubt instead concerns whether people in the penumbra should be regarded as at least partly composing that group. Corporate employees in the penumbra, on Bratman’s account, act on kernel-induced reasons. But many people clearly outside the organization act on kernel-induced reasons, for example, Gustav the outsourced arborist who comes on to corporate grounds to trim trees. Let us suppose that Gustav must get permission from corporate security to come on to corporate grounds, that he must speak with a finance clerk in order to arrange compensation, etc. These interactions give Gustav reasons for action that ultimately derive from people in the kernel. Corporate outsiders can thus act on kernel-induced reasons just as much as most those full-time employees who come every day to earn their paycheck and have no larger ambitions for Volkswagen. Why identify the contours of the corporate agent as including penumbral corporate employees but excluding corporate outsiders like Gustav? Bratman has an answer.

By itself, the notion of a kernel-induced reason seems too thin to distinguish anything like a corporate agent. So Bratman supplements the notion. He suggests that organizations have a complex set of rules governing the roles of individuals and their interactions, and that an individual should be identified a member or participant of an organization when she acts on kernel-induced reason while being cooperative regarding these rules. Further, he contends that an employee’s interactions with others “would induce rational pressure … either to acknowledge relevant authoritative demands or drop out of the shared policy,” and that these interactions distinguish penumbral employees from corporate outsiders like Gustav. In describing people who are at least “minimally cooperative” with the rules of the firm, Bratman aims to include the spectrum of workers in a firm, including those who do not endorse or approve of the firm’s general policies and aims, but are alienated. I soon argue that he does not achieve this aim. Before criticizing Bratman’s account, however, it will prove useful examining an insight it contains.

In identifying the contours of an organization and thus the participants in an organization, I think that Bratman captures the idea that managers—presumably people in the kernel—at least partially compose a group agent: managers can be plausibly viewed as importantly and distinctively cooperative with respect to a complex of social rules within an organization, and much of significance follows from that fact. A marketing manager, for example, would ordinarily be enmeshed in corporate practices in ways that our contract-based arborist is not. She does not simply execute tasks—she helps decide what should be done. Doing her job requires planning, negotiation, and other interaction with people from corporate finance, compliance, human resources. She must negotiate her budget with outsiders, learn about new product development, ensure that her subordinates comply with company rules and the law, and do all this while respecting the institutionally-established roles for the people with whom she interacts. In interacting with her colleagues, she has a say about the rules of the firm. Moreover, she may act with the aim of pursuing institutional aims that transcend her personal aims and interests; she may act for the sake of the institution, not merely herself. She thus does more than acknowledge authority. By shaping firm policies through her interactions and negotiations with others, she is part of authority. The structure of her relations with others in the firm confirms the existence of multi-party or group agency. Or so Bratman might plausibly argue.

Let us suppose, then, that Bratman’s appeal to “minimally cooperative” participants in a corporation leads us to identify a group agent. The question remains: what are the contours of that group agent? Who are members that compose the group? I suggest that the group agent Bratman’s account would identify does not correspond to the corporation itself, at least in cases in which a corporation is a massive enterprise with many alienated workers, but that it instead corresponds to some group of managers and people they enlist to knowingly participate in their projects. Indeed, in a massive corporation, there is no reason to think that only a single group agent would emerge. Instead, there might well be multiple group agents corresponding to the multiple loci of power within the firm, for example, an accounting group agent, a research group agent, and so on. At Volkswagen during Dieselgate, no doubt some group of managers and enlisted engineers might be counted as a group agent responsible for Dieselgate, because they did not simply execute assigned tasks but helped to decide what should be done by planning, negotiating, and so on. In virtue of their collaborative nature, however, this group agent excludes alienated workers, and therefore cannot correspond to the corporation as a whole. Of course, Bratman himself aims to include alienated workers as part of the firm, saying that they form part of his “coordinated union.” That aim is dubious. A worker who installs seats in a car under production, for example, need not have significant contact with people outside her position on the factory line. She need not talk with people in corporate finance, compliance, human resources, or facilities beyond, perhaps, making simple requests for assistance. Doing more is not part of her job. Nor is it part of the job of most blue collar workers, people who serve food in the cafeteria, wash the floors, replace electrical bulbs, etc. Managers may negotiate corporate practices but blue collar workers more simply execute the nondiscretionary tasks assigned to them. The blue collar workers at Volkswagen elude identification on Bratman’s account of group agency. Most people who work at Volkswagen—the blue collar and front line workers—do not form part of a corporate agent. Volkswagen itself, composed of the whole set of Volkswagen employees, cannot be identified as a responsible corporate agent.

Now Bratman might retort that blue collar workers have sufficient contact with the larger organization to be counted as constituents of the corporate agent. A regularly employed seat- installer at Volkswagen may converse with the human resources department about changes to his benefit package, may see that maintenance is called when there is a spill on the factory floor, etc. But these kinds of contacts are simply requests for assistance or clarification. They no more express “cooperation” with a firm’s social rules than does a prisoner’s query about when he will be released express cooperation with prison policy. Most problematically for Bratman, the kinds of contacts I identify for our alienated worker—e.g., the seat installer—do not suffice to distinguish Volkswagen employees from outsiders to the corporation. Gustav, our outside contractor engaged to trim trees, will have the same sort of contact with parts of the corporation as the blue collar employee. Gustav may have to speak with security in order to get on to corporate grounds, speak with another office about getting compensated, and speak yet with a third office when he has trouble finding electrical outlets for connecting his equipment. Those contacts do not make him a member of the Volkswagen Corporation, but they seem just as significant as the contacts that Bratman might invoke to show that the seat installer is a member of the Volkswagen corporation. Here I echo Anderson’s observation: for most of its workers the business corporation is authoritarian institution in which typically subordinates are neither consulted nor informed about significant actions. With respect to their relation to authority in a corporation workers and outside contractors are on par. They stand outside the agency of the firm. They do contractually assigned tasks, and move on. Bratman plainly intends that his proposal models a whole organizational agent, that the people in the kernel and the penumbra compose the corporation, in terms of managers and their collaborators alone. But his plan misfires.

One might think that by excluding Volkswagen employees as members of the Volkswagen corporate agent, I have been too hard on Volkswagen, at least with respect to its relation to employees. The corporate governance structure at Volkswagen reflects the law of Mitbestimmung, the German commitment to worker co-determination, which purports to give voice to workers in controlling the firm.²⁴ If Mitbestimmung gives Volkswagen workers a right to meaningful control over their firm, a voice in the structure of their workplace and the governance of the firm, perhaps that limits their alienation, and perhaps they therefore partially compose a Volkswagen corporate agent. Large empirical issues lurk here. It is not so clear that Mitbestimmung brings relevant voice to employees. Indeed, some argue that Mitbestimmung leads unions to identify with their employers, placing competitiveness of the firm above the well-being of its employees.²⁵ Perhaps as a result, the popularity of Mitbestimmung among workers is declining.²⁶ The empirical issues that regard whether Volkswagen employees indeed had the right to meaningful control during Dieselgate cannot be resolved here. I assume that as things stand now, corporate democracy in large publicly traded corporations, at least in the United States, is not a regular phenomenon. Workers are alienated, not because of their uncooperative disposition, but because they have no meaningful opportunity to exercise voice in corporate or workplace matters. The existence of a minority of corporate democracies would not undercut the claim that worker alienation more generally conflicts with the idea of a corporate agent composing the full spectrum of firm employees.

GOODBYE CORPORATE MORAL AGENCY

Not just any group of people can reasonably be construed as composing a particular corporate moral agent or as being the people on whom a corporate moral agent supervenes [for ease of expression, again, I will refer to both supervention (or supervening) and composition (or composing) as “composition” (or “composing”)]. Obviously, for example, the group of people across the world whose initials are “LMN” cannot plausibly be identified as the group composing Volkswagen, because presumably the vast majority of members of the LMN group have no significant relationship to one another or to the production of Volkswagen vehicles. So the principle that people compose a corporate agent if their initials are “LMN” fails. I have in this essay surveyed a number of other principles specifying when people compose an agent, including the principle that people make a commitment to the authority of a firm’s leadership, the principle that people are identified in an organizational flow chart, and the principle that people are either in the kernel or the penumbra of a particular kind of hierarchy. These principles, like the LMN principle, fail in identifying a corporate moral agent. That suggests some reason to be skeptical about corporate agency: If, even using some of the best theories of corporate moral agency, one must fail to credibly identify anything as a corporate moral agent, then one has reason to be skeptical that corporate moral agents exist.

Still, one might think it rash to doubt corporate moral agency. I have surveyed only a few principles for identifying corporate moral agents. There are many more possibilities. Perhaps some principle not yet discussed will be more successful. I doubt it. There is a pattern of failure in the principles we have so far encountered, and good reason to think that the pattern would unavoidably continue in all principles purporting to identify corporate moral agents. As Anderson suggests, for the vast majority of its workers the business corporation is an authoritarian institution in which subordinates are neither consulted nor informed about significant actions. Workers execute the nondiscretionary tasks assigned to them and have no say in organizational decision-making, no awareness of the events in wrongful corporate decision-making, and make no commitment to broad purposes of the firm as articulated by corporate management. I argued that the corporation must be conceived as composed of the vast majority of its employees, including blue collar workers and other front line employees. Thus conceived, the corporation is no agent but instead a collection of people who happen to take their orders and get their paychecks from the same source. If a group of people compose a corporate moral agent, something must unite them into agency. The facts that most workers are alienated in the ways that Anderson suggests shows no relevant unity among corporate workers. The metaphysical conclusion is clear: there are no corporate moral agents.

Criminal culpability requires moral desert, and in the absence of moral agency, there is no moral desert and therefore no corporate crime. If we wish to locate responsible corporate agents within the business realm, we do best by searching within the firm for individuals and groups that act as agents: deliberating, collaborating, planning, and acting on morally significant corporate matters. If the business corporation involves group agents, they are composed of leaders and the people whose cooperation they enlist within the corporation, not the corporation itself. Corporations themselves are unfit to be criminally prosecuted.

In denying corporate moral agency, I argue against commonsense. People blame corporations regularly, thus recognizing as part of commonsense the idea that corporations are moral agents. I contend that commonsense is wrong. I do not, of course, stand alone in arguing for the conclusion that commonsense sometimes errs in its ontology. Some metaphysicians—ontological nihilists, fictionalists, or eliminativists—argue that ordinary objects like tables and chairs do not exist, but that at most microparticles exists.²⁷ Although the idea that commonsense ontology is defective is not new, in the case of corporate moral agency nihilism, the defects in commonsense ontology have practical implications should that trouble us in ways absent when we consider ordinary object nihilism.

The arguments I offer against corporate moral agency are largely conceptual. One might suppose, however, that conceptual issues should be set aside as being merely “scholastic,” in the pejorative sense of that term; that unless we hold the corporation itself responsible, we lose something of important practical value. Here I consider two ways in which that practical value might be exemplified: in terms of the possibility of imposing adequate fines and in terms of the expressive value of corporate punishment; and I argue that we best respect these values without recognizing corporate moral agency.

One might think that there is a distinct value, connected to money, that gets lost if we stop prosecuting corporations. When corporations are successfully prosecuted for their crimes, the fines can be enormous. Billions were extracted from Volkswagen coffers to pay for Dieselgate, for example. Convicted individuals and groups within the corporation can only pay a comparative pittance for their misdeeds. If we can prosecute only individuals and groups within the corporation for criminal activity, we may seem to lose access to corporate coffers. Such loss seems a pity. The funds can be used to compensate victims and to demonstrate recognition of the magnitude of harm done in cases like Dieselgate. This practical problem has a straightforward solution that does not require corporate criminal punishment, I believe. We may embrace the principle that when criminal activity arises from corporate managers and others acting within their roles in the corporation, payment for their wrongdoing may come from corporate coffers. The consequence of embracing such a principle is that the costs of rectifying wrongs gets largely borne by Volkswagen shareholders, who may seem comparatively innocent. Why make these shareholders pay for harm done by managers? The answer, I think, is that when shareholders invest in a corporation, they assume some risk—perhaps limited to the value of their investments—of paying for the any wrongs committed by managers acting within the scope of their employment. One might doubt the fairness of imposing such risk on comparatively innocent shareholders. To fully answer that doubt would require an essay devoted to the topic. For purposes here, I contend that it is more plausible to say that it is fair to impose the relevant risk on shareholders than it is plausible to accept the metaphysical morass of regarding the corporation as itself a responsible agent.

Fines aside, one might think that denying corporate agency and responsibility to undermines value in our practices of punishment. That value that can be articulated in terms of the expressive value of punishment. Suppose, then, something widely held: that an essential feature of criminal punishment is its function expressing moral condemnation of criminal wrongdoers. Suppose, also, that because of the complex structure of a business corporation, when criminal activity emerges from a corporation such as Volkswagen, we may know that some individuals or groups within the corporation engage in wrongful conduct, even though we cannot with particularity identify each such individual or group. If we focus only on people we can readily identify for prosecution when we have reason to believe that others have been involved in crime, some guilty people will evade punishment, which may seem to frustrate a purpose of criminal law, blunting law’s condemnatory function and understating the magnitude of wrong done to the victims of crime. Holding the corporation responsible may seem to salvage the condemnatory function of criminal law, “a way of holding responsible individuals [or groups] within the organization whose identities we cannot determine”.²⁸ Blaming the corporation as a whole would seems to allow us to spread the blame beyond guilty people or groups that we can readily identify; the corporation would serve as a surrogate for anonymous wrongdoers within the corporation, one might argue”. But I disagree.

Prosecuting corporations cannot be regarded as “a way of holding responsible individuals [or groups] within the organization whose identities we cannot determine”. Taking matters literally, in handing out a guilty verdict against the corporation itself, we do not convey a message that merely unidentified groups or individuals within the corporation merit censure. Instead we convey the idea that the corporation itself—whether conceived as an entity that supervenes on it workers, or conceived as the set of workers themselves—deserves censure, an idea that says nothing in particular about unidentified groups or individuals within the corporation. It would be a twisted message that has us blame the undeserving corporate entity, or undeserving people within a corporation, for a wrong committed by only a small minority of people in the corporation. Such indiscriminate targeting of the undeserving undercuts rather than supports the legitimate condemnatory function of law, undermining the credibility of punishment by establishing that our punishment practices are not connected to the desert of the accused, robbing us of something practically valuable. Alternatively, one might suppose that the message one conveys in punishing the corporation itself should not be taken literally, but rather should be regarded as shorthand or code for condemning those deserving individuals or groups within the corporation that we cannot identify. The problem with the shorthand interpretation is that there is no reason to think that it represents a message that gets communicated. Treating a verdict against the corporation as shorthand would rely on the idea of sending a message that is not explicit but implicit, and would leave the audience with wide latitude on how to interpret the message. The audience would be free, for example, to interpret the message along the lines suggested by the theories of Bratman, French, or Pettit, as directed at the corporation itself, and not as directed at some subset of individuals or groups within the organization. That freedom muddles and undermines the condemnatory message, and thus should be avoided. To avoid the muddle, prosecutors would have to change the condemnatory message from implicit to explicit, explaining that the censure in their message is aimed at hard-to-identify individuals or groups within the corporation, not at the corporation itself. Of course, doing so would be an acknowledgment that it is not the corporation itself that gets prosecuted, which strikes me as a sensible thing to do. It would, however, confirm the pointlessness of prosecuting and convicting corporations rather than responsible groups or individuals within corporations.

As I argued earlier, prosecuting the corporation itself makes no sense conceptually. It turns out that as a practical matter, we lose nothing by denying corporate moral agency and not prosecuting corporations.