Incomplete Contracts and the Theory of the Firm: What Have We Learned over the Past 25 Years?
GAICIL Insights
Who owns the model when nobody could have written down what the model would become? A survey by Philippe Aghion and Richard Holden, published in the Journal of Economic Perspectives, offers the deepest available answer without ever mentioning AI. It takes stock of 25 years of research on incomplete contracts and the theory of the firm, revisiting the answer Sanford Grossman and Oliver Hart developed in their 1986 paper on the costs and benefits of ownership.1 The central issue: economic actors are boundedly rational and cannot anticipate all contingencies. Some states of nature or actions cannot be verified by third parties and cannot be written into enforceable contracts.
When contracts are incomplete, any contract negotiated in advance must leave some discretion over the use of assets. The owner of the firm is the party holding residual rights of control. The optimal allocation of property rights, or governance structure, minimizes efficiency losses. That yields a theory of ownership and vertical integration and a theory of the firm. Ownership is what fills the silence of the contract.
The authors explain with a simple numerical example and show how the approach analyzes firms' internal organization, financial decisions, the costs and benefits of privatization and the organization of international trade between arm's-length and intrafirm transactions. The theory spurred an empirical literature on firm boundaries. Grossman and Hart predict underinvestment by parties without ownership. Evidence from UK firms shows more vertically integrated firms display higher R&D by final producers and lower R&D by supplying industries. Other evidence shows ownership providing incentives for care of assets that are difficult to specify in contracts.
The approach is a cornerstone of organizational economics. Ownership matters because contracts cannot specify everything, so control rights shape incentives to invest. Parties with control invest more in relationship-specific assets; parties without control may underinvest. The party left out of control holds back its best effort.
Now the AI angle. Artificial intelligence turns this theory into a live question. Who should own data, models and the outputs they generate? Contracts governing AI development and deployment are inevitably incomplete, because AI systems' uses and capabilities are hard to anticipate. Allocation of residual control rights over AI assets will affect incentives to invest in them. Every licensing deal for a foundation model is an incomplete contract with high stakes.
The integration-R&D evidence maps onto AI value chains. Integration affects which parties invest in innovation. AI firms decide whether to integrate data, computing infrastructure, model development and applications, or rely on partners. The theory implies these choices affect investment incentives at each stage. Integration may strengthen some parties' incentives while weakening others'. A model developer that buys its data supplier may gain control and lose the supplier's drive to improve.
Internal organization connects too. Delegating authority within firms, or to AI systems, involves allocating control under incomplete contracts. The framework helps explain when delegation improves outcomes and when centralized control is preferable.
Policy is implicated. Regulations affecting ownership of data and AI models alter control rights and therefore investment incentives. Rules on data ownership, interoperability or model access should be judged by their effects on investment in AI assets, beyond their immediate fairness. Reassigning control rights by statute is reassigning investment incentives.
Breadth is the framework's strength. Privatization, international trade, internal organization: in each, control rights shape outcomes when contracts are incomplete. AI, spanning all these domains, will be shaped by the same considerations.
A hard question for AI regulators drafting data and model ownership rules: the first question to ask is who will invest less once the rule takes effect. In a world of incomplete contracts, that answer may be the most consequential line in the regulation.
Introduction
Sanford Grossman and Oliver Hart used the theory of incomplete contracts to develop answers to the question “What is a firm, and what determines its boundaries?” in their path-breaking paper on “The Costs and Benefits of Ownership: A Theory of Vertical and Lateral Integration” (Journal of Political Economy, 1986, vol. 94, no. 4). Perhaps the central issue is that economic actors are only boundedly rational and cannot anticipate all possible contingencies. It might well be that certain states of nature or actions cannot be verified by third parties after they arise, like certain qualities of a good to be traded in the future, and thus cannot be written into an enforceable contract. When contracts are incomplete, and consequently not all uses of an asset can be specified in advance, any contract negotiated in advance must leave some discretion over the use of the assets; and the “owner” of the firm is the party to whom the residual rights of control have been allocated at the contracting stage. The optimal allocation of property rights—or governance structure—is one that minimizes efficiency losses. This produces a theory of ownership and vertical integration as well as a theory of the firm. First we spell out Grossman and Hart's argument using a simple numerical example. Then we show how the incomplete contracts approach can be used to analyze the firms' internal organization; the firms' financial decisions; the costs and benefits from privatization; and the organization of international trade between inter- and intrafirm trade. We discuss several criticisms of the incomplete contracts/property rights methodology and review recent developments of the incomplete contracts approach.
Takeaway
“In this essay, we have focused attention on theoretical debates around the incomplete contracts/property rights approach, but we would be remiss not to add that this theory has also spurred an empirical literature on fi rms’ boundaries and organization. For example, Grossman and Hart (1986) predict underinvestment by parties that do not have ownership (and hence residual control rights). Acemoglu, Aghion, Griffi th, and Zilibotti (2010) use U.K. fi rm-level panel data and input-output information about U.K. fi rms to show that more vertically integrated fi rms—fi rms that produce a higher fraction of their inputs in-house—also display higher R&D investment by fi nal producers and lower R&D investment by supplying industries. In another example of how ownership can address incomplete contracts, in the trucking industry, one ownership form is for the driver to own the truck directly, while another form is for a trucking company to own the truck and to hire the driver as an employee. Baker and Hubbard (2004) fi nd that driver ownership provides an incentive to take care of the truck but means that the dispatcher will underinvest in planning how the truck should be used (in particular, in fi nding return trips to utilize capacity). It would be tricky to draw up an enforceable contract on either of these points: taking excellent care of the truck, or using the capacity of a truck as effi ciently as possible.1
Thus, Baker and Hubbard fi nd that driver ownership is more prevalent in long-haul settings, where truck maintenance is more important and where fi nding a return back-haul load is more likely. They also fi nd that a technological shock—the introduction of onboard computers which can help monitor how well the truck is taken care of—leads to a decrease in driver ownership.”
Notes
- Philippe Aghion & Richard Holden, Incomplete Contracts and the Theory of the Firm: What Have We Learned over the Past 25 Years?, 25 J. Econ. Persp. 181 (2011), https://doi.org/10.1257/jep.25.2.181. Takeaway source: Philippe Aghion, Richard Holden, Incomplete Contracts and the Theory of the Firm: What Have We Learned over the Past 25 Years?, Conclusion. ↩
Cite as: Philippe Aghion & Richard Holden, Incomplete Contracts and the Theory of the Firm: What Have We Learned over the Past 25 Years?, 25 J. Econ. Persp. 181 (2011), https://doi.org/10.1257/jep.25.2.181
Header image: Park Ave Building interior from loft.jpg by Mbrickn, CC BY 4.0, via Wikimedia Commons.
