The Contract Theory Nobel (and Some Grad Student Advice)
Yesterday, the Royal Swedish Academy of Sciences announced that the 2016 Nobel Memorial Prize in Economic Sciences was given to Oliver Hart and Bengt Holmström, respectively of Harvard and MIT, for their work on contract theory.
As far as I am concerned, it was about time those two got the Nobel. But then again, I have been a big fan of Holmström and Hart for a long time–I went into graduate school wanting to work on applications of contract theory, and I did just that by writing a dissertation titled, rather un-enticingly, Three Essays on Agrarian Contracts.
In 2004, I spent eight months in Madagascar to collect primary survey data for my dissertation. Because I arrived during cyclone season, my field site was not accessible for some time, and so I decided to go through the entire syllabus for a course in applied contract theory that was then taught at the European University Institute by Pascal Courty. In the process, I read a number of things by Hart and Holmström, all of which were very enlightening.
Most of the media stories I read or heard regarding this year’s Nobel talked about CEO compensation, but the two Holmström contributions that have been the most useful in my own work have been Holmström and Milgrom (1987), which identifies the condition under which a linear contract is optimal, and Holmström (1982), which looks at moral hazard in teams.
On linear contracts, suppose a principal contracts with an agent for the production of some output [math]q[/math], which is subject to a certain amount of uncertainty [math]\epsilon > 0[/math], so that the agent produces [math]q[/math] but the principal can only see [math]q + \epsilon[/math]. The idea is for the principal to design the incentive scheme that will induce the agent to provide her best effort. Many real-life incentive contracts–such as the sharecropping contracts I was studying in the first two essays in my dissertation–are linear. That is, the agent is paid [math]a + b(q + \epsilon )[/math], a remuneration which is linear in (observed) output, and which consists of a fixed component [math]a[/math] and a variable component [math]b(q + \epsilon )[/math]; the latter is what people mean by “incentive payment.” There are tons of potential contract shapes, and there is nothing that says that a linear contract is de facto optimal. Holmström and Milgrom (1987) showed the (rather restrictive) conditions under which a linear scheme is the best possible contract shape. To be fair, my hunch has always been twofold: either people settle on linear contracts because the transactions costs of going for a “better” nonlinear contract outweigh the benefits, or… what we observe to be linear contracts in fact include many small clauses that we do not see which serve to nonlinearize the contract, and thus bring it closer to the optimal (nonlinear) contract.
On moral hazard in teams, Holmström (1982) considers the case where a principal has a team of agents working under him, all jointly producing an output whose quantity is also subject to uncertainty. Here, Holmström showed that one of the agents has to be made supervisor of everyone else being made residual claimants on the principal’s profit in order for supervision to work in the principal’s best interest. (This has always struck me as being close in spirit to the work of Alchian and Demsetz in their 1972 AER article titled “Production, Information Costs, and Economic Organization.”)
Oliver Hart’s work I am less familiar with. Though I have read some of it, looking back at my research, I cannot find a citation to his work. The two things by him I remember best reading are (i) the book that came out of his Clarendon Lecture, titled Firms, Contracts, and Financial Structure, and (ii) Hart and Moore (1990), on incomplete contracts.
Gratuitous Bit of Grad Student Advice of the Week: Reflecting on all this brought to mind one important lesson I learned during grad school. When writing your dissertation, it really pays to invest in reading the broader technical literature surrounding your topic. In my case, my topic was sharecropping and contract farming, so I could have chosen to only read papers about those two institutions. Instead, I felt that to credibly talk about and do applied contract theory, it was necessary to know the major results of contract theory, both theoretical and applied. At the end of the day, I feel as though that strategy helped me on the job market, when I could talk to people doing labor, health, etc. about problems in their field rather than only talk about development. Grad school is the last time in your life you will have tons of free time to read as broadly as you like–use that time wisely to read the major papers in the theoretical literature that underpins your application(s) as well as applications of said theory in other fields.