Theories of Corporate Agency

French, L&P, and the scholars working with their accounts have focused almost entirely on the agency of corporate agents — primarily on the function underlying their intentionality (the gen­eration of commitments by various mechanisms) and more rarely on the function underlying the actions that flow from those commitments.

To see their mechanisms in context — and to distinguish between the corporate agent itself and the larger context in which it operates — we need a way to identify the agent, the whole that bears the capacity of agency. The existence of such an agent is assumed, or even entailed, by FLiP’s accounts. Barring a pretty radical metaphysical dualism, capacities like intentionality, rationality, and agency cannot be free-floating, so there must be a bearer, and here it seems clear that it will have something to do with “the members” (whoever they are, exactly). But the cor­porate agency literature has remained generally silent about the res that bears the capacity of agency — about the basic metaphysical questions of its existence, unity, limits, identity, and per­sistence conditions, among other things. It is well beyond the scope of this chapter to address those issues here (but see Hess 2018c), but we can at least look for the res — for the reasonably unified, cohesive, and well-defined entity that provides an adequate subvenient base for the

intentionality that FLiP have claimed for it. Having identified it, we can then ask whether FLiP’s theories account for its behavior.

8.1.1 French­

French argues that “corporations” ³ are capable of“intentionality” — and hence agency — because they can make decisions and act on them.

(1) an organizational or responsibility flow chart that delineates stations and levels within the corporate power structure and (2) corporate decision recognition rule(s) (usually embedded in something called “corporation policy”). The CID Structure is the personnel organization for the exercise of the corporation’s power with respect to is ventures.

(French 1979: 212, also 1984: 41-42, 1995: 25)

CID Structures establish the rules about how corporate decisions should be made with respect to established corporate ends, and thus set the standard for what counts as a corporate deci­sion and a corporate action. This implies that decisions and actions that come about by other mechanisms are not corporate decisions or actions — an implication which French affirms (French 1984: 53, 1995: 31). In doing so, he acknowledges that the whole does act on other bases, and that his account simply doesn’t address such actions.

Of course no corporation ever follows its own official rules perfectly, and French acknow­ledges the occasional need for “less formal decision lines... [which] although they may not have the baptismal blessing of the organization’s directors,. function in important ways in decision-making” (1984: 49). He acknowledges that “in practice, corporate policies are rather more flexible than my earlier comments may have suggested. Often they are no more than partial plans that will be revised, modified, extended, and reconsidered as the corporation confronts unanticipated situations and opportunities” (1995: 33). It is thus possible for the official CID Structure to be amended around the edges, and French even considers the pos­sibility of a “phantom, i.e. unpublished, network” that would run counter to the published network; the phantom would be the real structure — apparently replacing the published CID Structure rather than supplementing or amending it — as long as “everyone who matters in the corporate decision structure is well aware of [it]” (1984: 51).

Despite this slight softening, however, French’s CID Structures are still quite limited, encompassing only the publicly identified and acknowledged policies and procedures that shape corporate behavior — the ones that are “epistemically transparent” (1995: 25—26) and “relatively easy to detect, even from outside the corporation” (1995: 33). These elements are “stable and transparent. and not couched in the kind of language that would be appropriate to indi­vidual purposes” (1995: 30). It is essential to his later claims about the superior moral potential of corporations (1995: chapter 14) that the policies and procedures of the CID Structure be obvious to outsiders (and, apparently, “couched in language”). Further, these policies

seem to be inviolate. Certainly the basic ones are — indeed, they must be. In that respect they are unlike policies adopted by individual humans. You could adopt a policy of honesty but you may occasionally violate that policy by lying. Yet, when you lie, it is still you lying. If employees act in ways that violate [the adopted] corporate policy [of honesty], their acts are no longer corporate.

(1995: 32, see also 1984: 58-59)

French’s CID Structures thus include only those policies and practices that enable corporations to do what they’re supposed to do, pursuing acknowledged ends by acknowledged means.

This theoretical commitment has significant practical and normative implications. Seemingly “corporate” behaviors — collective action by members using corporate resources and relationships from their positions within the corporation — which deviate from acknowledged policies are not captured by the account, and are in fact actively rejected as not truly “corporate.” The awk­ward result is that the members can violate official corporate policy, but the corporation itself literally cannot.

Setting that difficulty aside, what does this account tell us about the agent itself — about the thing that has this CID Structure and bears the intentionality that it facilitates? French does not address the matter, but there must be some kind of material bearer of the capacity — a subvenient base adequate to the supervenient properties French has claimed.⁴ Given the breadth of the CID Structure — which includes the official “stations and levels within the corporate power structure... for the exercise of the corporation’s power with respect to is ventures” (1979: 212) — it seems plausible that the base will include at least most of the apparent mem­bership of the corporation, though probably not all of it. It will have to, in order to meet the requirements for supervenience, where every change in the upper-level property (“planning that X” or “acting on plan X”) has a correlating change in a lower-level property within the base (presumably, some configuration of the membership).

This is a good start, but leaves at least four “loose ends” regarding the metaphysics. First, does the agent include just the human members, or does it include things like artifacts (production machinery, communication technologies), recorded data (memos, programs), or infrastructure (buildings, property)? All of this becomes more significant in an age of increasing automation, and French does acknowledge the possibility of a fully automated corporation with no human members (1995: 35). The answer to this question will have implications for continuity, identity, and location, all of which will be relevant to action and (thus) to responsibility.⁵

Second, more importantly, does the agent defined by the CID Structure include all of the apparent members? French’s CID Structure will certainly capture a broad swath of the apparent membership, but it is unclear whether this will include (e.g.) support staff and custodial staff.

Third, is the agent limited to the apparent membership, or does it include external contributors like consultants, attorneys, and subcontractors? If membership is a function of a contract, or influence, or a continuing relationship then such apparently external players will be included; in terms of influence, in fact, they would be included while many intuitively “internal” contributors would not. It’s important that the subvenient base be sufficient for the supervenient capacity, but also desirable that it not extend infinitely (as on theories of global supervenience), and it well may if we’re tracking causal chains of influence.

Fourth and finally, whatever is included in the agent, what unifies all those pieces into a single entity? What holds them together, welding them into an entity that is robust enough to bear sophisticated capacities like intentionality, rationality, and action?

French’s account puts us in a place to ask such questions about the agent, but (in itself) lacks the resources to answer them.

8.1.2 List and Pettit

French’s account generally dominated the literature until L&P’s Group Agency: The Possibility, Design, and Status of Group Agents (2011).

A group agent is “jointly intentional” (possesses the capacity of intentionality) when its members “each intend that they together act so as to form and enact a single system of belief and desire, at least within a clearly defined scope... [and] all of this is a matter of common awareness” (34). L&P differ from collectivists like Margaret Gilbert and Seumas Miller in acknowledging that “some, or even many, group members may not participate in any joint intention at all. Generally, all we require for a group agent to count as ‘jointly intentional’ is that the group’s performance involve joint intentions on the part of some of its members” (35). The membership is thus not defined by this joint intention; instead, the relevant group consists of those members who do share the joint intention, those who “authorize” the group agent by “accepting its claim to speak and act for the group,” and those who act “in full awareness for the pursuit of the group’s ends” (35).

Even within the sub-class of group agents captured by this definition, however, L&P’s account of group agency is explicitly limited to groups that (1) form exclusively binary commitments,

(2) on the basis of communication, or more narrowly still, on the basis of various forms of voting (formal and informal). The former limits the account to groups with “on-off” commitments to propositions rather than “attitudes that come in degrees of strength”⁶; the latter limits the account to groups that form these attitudes — and thus to attitudes formed — via “communica­tion,” which includes deliberation, voting, and dictatorship (in which one member’s attitudes are simply imposed on the group) (37). L&P present a variety of scenarios in which members con­sider a proposition (X), and then “aggregate” their individual attitudes towards X in various ways to yield a group attitude towards X. These group attitudes “supervene holistically” on the attitudes of the members (or at least, presumably, on the attitudes of the members who participated in the process, since they are the only ones who expressed attitudes towards X).⁷ L&P acknowledge the logical possibility of group attitudes that “are not dependent on individual attitudes on the relevant propositions,” but reject this as “not very realistic”; they thus — again — restrict themselves to “cases where the members’ attitudes play this role” (66). Despite their title and the way their work has been received, then, L&P have not attempted to address the broad phenomenon of group agency at all. They have instead presented a highly developed account of one very specialized — and some­what exotic — mechanism for the enactment of group intentionality, and explicitly so.

So again, what does this account tell us about the agent itself — about the thing that has these voting procedures and bears the intentionality that it facilitates? There must be some kind of material bearer of the capacity — a subvenient base that is adequate to the “holistically superven­ient group attitudes” L&P have claimed, and (presumably) for the enactment thereof.

For one thing, L&P’s assumed agent will presumably involve “the membership,” and unlike French, L&P have carefully defined the group: within the intuitive “whole” of an organiza­tion, the membership of the group agent will include only those members who are jointly committed to the shared enactment of group agency, act “in full awareness for the pursuit of the groups ends,” and/or have authorized the group agent to “speak on their behalf.” This is actually a rather high bar, and it requires that each member have far more conscious, inten­tional engagement with the broader structure, function, and guidance of the group than is often acknowledged. This is a much smaller group within the organization than French’s, and seems unlikely to extend very far beyond management in most cases.

This raises a difficulty, as it is not clear that the membership L&P identify would form an adequate subvenient base for the full agency they have claimed. Agency requires both inten­tionality and action — an agent constitutionally incapable of acting on its beliefs is a bit of an oxymoron — and while this group would probably provide an adequate subvenient base for the group intentionality L&P have claimed, it’s not clear that it provides an adequate subvenient base for enactment. To form an adequate subvenient base for action, the membership must include the people involved in implementing the group actions that follow from the group intentionality, as well as the people involved in the voting processes that generate the group intentional states, but it is not clear that the people in this (much) larger group would meet the requirements of acting “in full awareness” or authorizing the group agent to “speak on their behalf.” So many people in large organizations are just “doing their jobs” without the kind of engagement or attention that L&P require for membership. This is an empirical question, of course, and the answer will vary from group to group, but it seems unlikely that the line workers at (say) Exxon are acting in full awareness of Exxon’s agency or have authorized Exxon to speak on their behalf.⁸

For another thing, L&P explicitly acknowledge the legitimacy of mechanisms other than their voting procedures (unlike French, who denies the legitimacy of mechanisms beyond the “epistemically transparent” practices and policies of his CID Structure). This is unsurprising, given their explicit commitment to functionalism (21), but again raises questions. While L&P accept the possibility of alternate mechanisms, it is unclear how they would accommodate multiple mechanisms, and it seems likely that their assumed agent would have them. Given a group that generates some commitments by voting and meets all of their other criteria (thus falling within the scope of their account), how would they incorporate commitments generated by those other mechanisms within the same agent? French’s proceduralist approach avoids this difficulty — any commitments, patterns, or behaviors arising via other mechanisms simply don’t count, no matter how reliable or effective they may be — but L&P already rejected this constraint. L&P’s assumed agent thus seems likely to include a multiplicity of mechanisms, but their account gives us no way to address the resulting complexity.

Finally, L&P’s account leaves many of the same metaphysical loose ends as French’s: it doesn’t tell us whether “the agent” is limited to the members or includes artifacts, or how to understand the relationship between the group agent and the people who are members of the organization — and perhaps necessary for enactment — but not members of the group agent. This is a more pressing concern here, as the scope of L&P’s agent (membership) is far more restricted than French’s. Certainly support and custodial staff are unlikely to clear the high bar established here, and it seems unlikely that line workers or people in entry level positions will qualify (though it does seem to exclude apparently external players like consultants). And finally, again, there’s no discussion of what unifies these pieces into a single entity — of what (all) holds them together, welding them into an entity that is robust enough to bear sophisticated capacities like intentionality, rationality, and action.

In each case, then, we have developed accounts of corporate agency that focus exclusively on a single mechanism and pass silently over the matter of the agent — the metaphysically robust res that bears the sophisticated capacities claimed for it. This has led critics to conclude, in effect, that there is no agent — that corporate agency involves some kind of “ghostly,” “mysterious,” “incor- poreal,”“bizarre”, and “hovering entity.”⁹ Given the collective silence about the agent itself, this is not an unreasonable interpretation, but nor is it a particularly charitable one. Again, FLiP’s claims of corporate agency entail the existence of a corporate agent, and their regular, general references to members give a general sense of where to look for it. The critics are correct, though, that nei­ther FLiP nor the theorists working with their accounts have told us much about it.

8.2