Contract theory

• 1. Introduction

  ›

  Contract theory begins with a deceptively simple question: how do you get someone to do what you want, when you cannot watch everything they do? Kenneth Arrow was the first economist to treat this question formally, in the 1960s. Half a century later, the field had grown sophisticated enough that Oliver Hart and Bengt Holmström shared the Nobel Memorial Prize in Economic Sciences in 2016 for their work on it. Their citations ranged across topics as distant as CEO pay and the privatization of public services.

  At its core, contract theory sits at the crossroads of economics, law, and the study of incentives. It asks how economic actors construct agreements when one party knows something the other does not. That gap in knowledge - information asymmetry - is the engine driving nearly every model in the field. Hart and Holmström approached that engine from different angles. Holmström focused on the tension between incentives and risk. Hart trained his attention on the unpredictability of the future and the inevitable holes it punches in any written contract.

  The questions contract theory raises are not abstract puzzles. They show up in health insurance markets, in public procurement, in job applications, and in the design of a manager's pay package. What gets promised, what gets hidden, and what gets left unwritten turns out to matter enormously.

• 2. Coase and the Founding Idea

  ›

  Ronald Coase, who would receive the Nobel Prize in 1991, planted the seed of contract theory in 1937 with an article called "The Nature of the Firm." His observation was precise: the longer a contract for the supply of goods or services runs, the harder it is to foresee the future, and the less appropriate it becomes for the buyer to specify exactly what the other party must do. That single insight contained two big claims. First, Coase was already thinking about economic behaviour as a network of contracts. Second, he was suggesting that when contracts grow thin and incomplete, firms tend to step in where markets cannot.

  Armen Alchian and Harold Demsetz pushed back against Coase's framing of the firm as a substitute for the market. In their reading, both the firm and the market are contracts, with no fundamental difference between them. What makes a firm distinctive, they argued, is team production. And the central problem of team production is measuring each individual worker's contribution - which opens the door directly to moral hazard.

  Michael Jensen and William Meckling carried that logic further, defining a business as a nexus of contractual relationships between individuals. On their account, the firm is a legal fiction whose main function is to serve as the connecting point for those contracts. James Mirrlees and Holmström then gave formal mathematical shape to the moral hazard problem within a principal-agent framework, drawing on game theory as a tool. Eugene Fama extended this static picture into dynamic contract theory, introducing questions of principal commitment and the way an agent's reputation plays out over the life of a long-term contract.

• 3. Complete and Incomplete Contracts

  ›

  Complete contract theory imagines that the parties to an agreement can foresee every future scenario and write rules for each one. Under this view, a firm and a market are both just contracts; there is no essential difference between the two forms. Principals and agents can develop optimal schemes for sharing risk and transferring revenue, reaching the best outcome available under their constraints. This framework is equivalent to what economists call principal-agent theory.

  Reality, of course, does not cooperate. A fully complete contract would have to specify the legal consequences of every possible state of the world. Writing such a document is either impossibly complex, prohibitively expensive, or both. Oliver Hart and his co-authors built the theory of incomplete contracts precisely around this fact. Their leading application is the Grossman-Hart-Moore property rights approach to the theory of the firm, laid out in Hart's 1995 book. The central insight is that when parties cannot write every contingency into a contract - particularly when relationship-specific investments are involved - who owns the relevant assets becomes the decisive question for how incentives are distributed.

  Because real contracts will always leave gaps, the law steps in with default rules that fill what the parties did not address. Eric Brousseau and Jean-Michel Glachant argued that contract theory should fold in not just incentive theory and incomplete contract theory, but also the new institutional transaction costs framework, treating all three as parts of a unified whole. During the last twenty years, much of the field's energy has gone into dynamic contracts, with early contributors including Edward Green, Stephen Spear, and Sanjay Srivastava.

• 4. Moral Hazard

  ›

  Moral hazard refers to the gap between what an employee does and what an employer can actually see: whether they work, how hard they work, and how carefully they do so. In the language of contract theory, the principal cannot observe or verify the agent's action. That unobservable behaviour is called a "hidden action."

  When the agent is risk-neutral and there are no limits on financial transfers, hidden action turns out not to matter very much. The principal can hand the realized output directly to the agent, require a fixed upfront payment in return, and the agent becomes a residual claimant with every reason to maximize total surplus. The contract achieves what economists call the "first-best" outcome. Risk aversion changes that calculation. A risk-averse agent cannot bear the full volatility of output, so the contract must trade off incentives against insurance. Neither goal can be fully satisfied at once, which is why outcomes under risk aversion are described as "second-best."

  A different wrinkle appears when the agent is risk-neutral but cannot afford the fixed upfront payment. In that case, the principal must leave what the literature calls a "limited liability rent" - the agent earns more than their outside option simply because they lack the wealth to post a bond. The moral hazard model with risk aversion was pioneered by Steven Shavell, Sanford Grossman, Oliver Hart, and others during the 1970s and 1980s. William Rogerson later extended it to repeated settings, while Holmström and Paul Milgrom tackled the case where agents handle multiple tasks simultaneously. A study on potential solutions found that adding moral sensitivity to the principal-agent model increases its descriptiveness and pedagogical usefulness by inducing employees to work at an effort level proportional to the wage they receive.

• 5. Adverse Selection and Signalling

  ›

  Adverse selection enters the picture before a contract is even signed. The problem is that the principal does not know a crucial characteristic of the agent at the contracting stage - what theorists call the agent's "type." Health insurance is a textbook illustration: people who expect to get sick are more likely to buy coverage, so insurers face a pool of customers skewed toward higher risk than the general population. Public procurement offers another case: a government agency hiring a private firm does not know that firm's true costs.

  The principal's standard response is to offer a menu of contracts and design it so each type of agent voluntarily selects the option intended for them - an "incentive-compatible" menu. To pull this off, the principal must leave an information rent with better-informed agents. They earn more than they would if no contract were written at all, simply because their private knowledge is worth something. The field predicts that, in most cases, trade will be lower than under full information - a downward distortion - except when the agent is of the very best type, a property the literature calls "no distortion at the top." Roger Myerson, Eric Maskin, and others pioneered adverse selection theory in the 1980s. George Akerlof described a related phenomenon in the used-car market, where sellers know more about quality than buyers do.

  Signalling flips the dynamic. Rather than waiting for the principal to design a menu, the better-informed party takes action first, choosing how and whether to present information. Michael Spence formalized this in 1973 through his job-market signalling model. In Spence's framework, job applicants signal their skills to potential employers, reducing the chance that a less qualified candidate crowds out a more capable one. Leland and Pyle applied similar logic to IPOs in 1977, arguing that companies going public can reduce adverse selection by sending clear signals to investors before the offering.

• 6. Incentive Design in Practice

  ›

  Designing incentives within a contract means choosing how to link rewards to performance. Contract theory distinguishes two main approaches. Absolute performance-related rewards tie pay directly to how an individual performs against some fixed standard. Relative performance-related rewards rank employees against one another, paying them according to their position in that ranking.

  Absolute rewards are the more widely used method in practice, recognized across economics as a baseline incentive mechanism. Their appeal is that they give each employee a direct line between their own effort and their own pay. Two drawbacks temper that appeal, however. The first is the incentive to cheat: when pay depends on measured output, some people will manipulate the measurement rather than improve the underlying performance. The second is sensitivity to external conditions, whether a recession or a sudden boom, that shift measured output independently of any worker's effort.

  Group-based rewards try to split the difference by aggregating output across teams and distributing the reward collectively. The problem is free-riding: a member who contributes little receives the same group bonus as one who works hard. Contract theorists argue that competitive structures - where better performance against colleagues yields higher rewards - can address this by making individual contributions more legible within the group. A distinct but related problem arises when the agent's task is to compute the value of something that belongs to the principal, such as an assessor pricing the principal's property. Designing a contract that makes the agent report the true value, rather than one that serves the agent's own interests, is its own branch of the principal-agent literature.