Oliver Hart’s economic theory of the firm

On Monday, two economists, Bengt Holmstrom of MIT and Oliver Hart of Harvard, were awarded the Nobel Prize in economics, for their contributions to the area that economists describe as ‘contract theory.’ I suspect that this Nobel will be particularly difficult for non-economists to fully appreciate. Usually, macroeconomics and financial economics are the fields of econ that make the front page of the Wall Street Journal, and capture the public’s attention. Plus, contract theory can often be abstract and technical—using game theory and probability to address questions that most people would not think to ask. So I wanted to write a brief post about Oliver Hart—as I’ve been a student of his and been deeply influenced by him—and his theory of the firm, and then talk about some recent high-profile news events that we might apply his theory to. Of all economic theories of the firm that I have encountered, Hart’s gives the most coherent and persuasive economic theory of ownership, control, and the boundaries of firms. But it seems like there are lots of transfers of control that are hard to rationalize with this theory, and, I suspect, any model of financial and economic rationality. So I’ll wrap up this post by asking how much we can expect of economic theory in explaining the world.

***

First, what do we mean by ‘theory of the firm’? Basically, a theory of the firm is a theory of why certain economic activities take place within the command hierarchy of a firm, rather than via arms-length market transactions. When our cars break down, most of us take our cars to local mechanics, pay them for a single service, and then drive home. But most firms have internal IT departments to fix computers and resolve other IT problems. Why should this be the case? Why couldn’t companies just hire computer-repair services from an external IT firm—or external IT contractors—piecemeal? Why do firms ‘own’ so many diverse functions in-house—IT, marketing, human resources, R&D, in-house, instead of outsourcing and buying services piecemeal, just as we do for most goods and services in ordinary life, for groceries and haircuts? Economic theory is not satisfied with the claim that it’s cheaper for firms to do these things in-house, rather than go to the market, as that just punts on the question of why it’s cheaper—why should it be more efficient?

So, a comprehensive theory of the firm should explain the full scope of firm ownership decisions—why it owns this tool rather than renting it or requiring its employees to provide the tool, why it has an in-house IT division rather than outsourcing, etc. On a more concrete level it should be able to explain concrete changes in the firm boundaries and asset ownership—why did firm X acquire firm Y instead of just having a contracting relationship?

Oliver Hart’s major theory of the firm revolves around three major concepts: (1) ownership as residual control rights, (2) non-contractible contingencies/ ‘incomplete contracts’, and (3) non-verifiable relationship-specific investments. Once we define what we mean by these concepts, it will be straightforward to understand his theory of the firm.

First, Hart defines ownership as residual control rights. Given that I can contract to use an asset that you own in any number of ways, all that your ownership really means is that you control the use of the asset in any situation or circumstance that we have not contracted around—hence ownership means residual control rights. Ownership is just who has a say where contracts pass over in silence.

Second, Hart assumes—and there’s a lot of deep, theoretical debate in contract theory over this assumption—that there are certain contingencies (that is, possibly future events) that it is impossible to write contracts around, whether because it is impossible to foresee them or impossible to legally specify them, or simply too costly to try. In other words, contracts can’t specify everything that must be done in every situation that matters to two different parties. As a result, ownership matters—in those non-contracted-for situations, the party that has ownership rights (residual control rights) gets its way. An example of a contingency that it might be hard for firms to contract around might be subtle changes in consumer tastes or the company’s marketing strategy: A car-maker contracting with a supplier that makes its auto bodies will want the auto-body maker to adapt each year’s bodies to that year’s fashions and tastes. But it might be hard to specify in a contract how fashions and tastes might change, how we would formally identify those changes, and how much the auto-body maker can be expected to make costly retailorings of its manufacturing process to satisfy those demands.

Now with these first two parts of the theory, alert readers might think we already have a theory of the firm: There are certain future contingencies that we cannot contract around, and so economic parties will demand ownership—residual control rights—to protect their interests in these situations. Concretely: The car manufacturer will buy the auto-body maker to directly control the production of the auto-bodies, based off of its knowledge of changing consumer tastes and demands. But this doesn’t totally satisfy economic theorists. After all, two contracting parties could just agree that, if one of these non-contractible contingencies takes place, they can renegotiate at that point. The car manufacturer and auto-body maker could just write a new contract each year, once they’ve gotten new information about consumer tastes that year. It turns out that, with only the first two components of our theory, it would actually be more efficient to do things that way—renegotiate at each contingency. So, to use the preferred economic language, this doesn’t quite explain why ownership matters ex ante—the parties could just renegotiate/re-contract ex post.

So Hart introduces a third, critical component to his theory: In the real world, parties often have to make unobservable relationship-specific investments in complementary assets—that is, investments that only pay off fully if the two parties continue their economic relationship. For example, the auto-body maker might invest in retooling its machines so that it is especially good at making bodies that are specifically tailored to the manufacturer’s chassises; just as the manufacturer might make chassises that work particularly well with the auto-body maker’s bodies. Relationship-specific investments can be good, efficient, and important, but expose parties that make relationship-specific investments to “hold up” risk. In a supplier-customer relationship without relationship-specific investments, hold-up isn’t a problem: If a supplier produces a commodity product, and its usual customer tries to ‘hold it up’ by demanding generous price concessions, the producer of the commodity product can just go sell to another buyer in a competitive market. But once parties have made relationship-specific investments, they no longer have this luxury: Given that other auto manufacturers might use different chassises (or just be located further away), the auto-body maker cannot easily sell to them.

So now we can put all three elements together: In some economic relationships, efficiency requires the parties to make relationship-specific investments. But each party will be wary of making such relationship-specific investments if they fear that, in future non-contractible contingencies, the other party will try to hold them up. So, the parties don’t fully invest in the relationship-specific investments and complementary assets ex ante, which is inefficient. But when complementary assets are owned together—by one economic agent with all residual control rights—this fear is not relevant, and the owner can fully invest in making the assets compatible, yielding all the efficiency. 

To make it concrete: When the auto body maker and the car manufacturer are housed separately, each will be wary of investing in tailoring their machines and processes to make compatible chassises and auto bodies, since such co-dependence could expose each of them to hold-up. By integrating under common ownership, prospective hold-up problems disappear and the owning firm can fully invest in making the assets compatible.  (This abstract example I’ve developed is based loosely off of the famous case of Fisher Body and General Motors.)

The short—but technical—way of summarizing this is as follows: Assets are commonly owned in firms in order to mitigate hold-up problems in non-contractible contingencies that would prevent separate contracting parties from efficiently investing in relationship-specific complementary assets.

In practice, there are various forces that push against integration, that can make integration costly even when there might be the potential for relationship-specific investments in complementary assets. Managers may not have the omniscience to run ever-more complex firms. Incentive problems may become worse and worse as more divisions and functions within a firm become ever more removed from the firm’s bottom-line. But the Oliver Hart’s theory is about the basic underlying economic forces that pull assets together, under common ownership, inside firms, and why, at the most basic theoretical level, integrated firms make sense and are so pervasive.

***

Oliver Hart’s theory of the firm is the most coherent and persuasive I have encountered. But it is very much an economic theory of the firm—it explains what firm boundaries make sense and why, given people who are behaving rationally toward financial ends. Is such an economic theory complete? Readers might object that many of the acquisitions that we’ve read about in recent years are hard to account for with this theory. When CEOs issue press releases explaining their M&A activity decisions they rarely reference non-contractible contingencies and unobservable relationship-specific investments.

For example,can this theory account for why Twitter should be acquired by anybody, or by SalesForce in particular? Or why Yahoo was acquired by Verizon? In the case of Twitter, the idea—as far as one can discern—is that because Twitter has not Grown (where Growing is defined as growing as fast as Facebook), it needs a Strategy and should find a buyer. The Twitter go-shop process does not seem to be motivated by expected synergies, relationship-specific investments, actions that Twitter cannot take as a standalone company, or anything else that economic reasoning would tell us invites an acquisition per se. It seems to be driven by human narrative reasoning: Twitter’s stock price has gone down and so we need to Do Something Bold, such as selling to another company so that we can have narrative resolution on the Story of Twitter.

The business press is replete with examples of companies making acquisition decisions that have no coherent economic logic, but which are very easily explained by managers’ vanity, failure to understand certain financial identities, fixation on ‘growth’ even when it is inorganic and thus not meaningful, narrative reasoning, and etc. It seems hard to avoid the conclusion that a large fraction of M&A activity and thus changes in firm scope and control is explained by, in two words, human folly. But economic journals like to publish models of rational economic behavior and empirical tests of those theories. This means that an empirical paper that finds some theoretically-motivated relationship—say, between observability of relationship-specific investments and integration—will be published and make it into the literature and the official understanding of the firm that econ and business-school PhDs learn and internalize. And the large fraction of the residual in that regression that is explained by simple human folly won’t make it into the official theory. One is unlikely to publish a paper whose thesis is, “Managers often grope along in darkness and ignorance, as evidenced by their own statements, and a significant portion of M&A activity is explained by this fact alone.”

None of this is a criticism of Hart’s theory per se—just an explanation of why there is a disconnect between the sophisticated theories that academics win Nobel Prizes for and the kinds of things we read in the business press every day about M&A. But I think that Hart’s theory is extremely useful for two different things:

  1. First, it is useful as a normative theory (rather than a purely descriptive one), of how agents should behave. This should be useful for consultants and managers who actually want to do what is in the interest of a company’s stakeholders. Hart’s theory shows why integrated ownership can add value for everybody. If it is impossible to articulate why the goals of an acquisition cannot be achieve through arms-length contracting, or to express the logic of the acquisition in terms of important relationship-specific investments in complementary assets, then the acquisition should not be undertaken.
  2. Second, it should descriptively explain firm boundaries and integration decisions in the special circumstances in which we can expect the economic rationality assumption to hold. These assumptions would hold when either (a.) agents are smart enough to reason through the optimal solution to their contracting problems and are actually motivated by financial goals, or (b.) competitive selection pressures are strong enough to eliminate suboptimal firms and allow optimal contracting relationships to be imitated and replicated, even when the human beings involved in these contracts aren’t smart enough to know what’s going on.

The selection argument operates over a fairly long horizon. As such, Hart’s theory of the firm perhaps best explains firm integration decisions that run so deep that we take them for granted and don’t even notice them. In other words, perhaps it is best at explaining the basic patterns of how business units are held together—which assets are co-owned and what activities tend to be paired together—but not necessarily why corporations and conglomerates buy and sell those business units. And that could be its true genius, that it explains the arrangements we take for granted—the co-owned assets that never, ever get separated and sold off—and not the ownership arrangements that are spun off, traded, and acquired, and thus make the front page of the WSJ.