It was great to wake up this morning and find that Oliver Hart and Bengt Holmström were awarded the Nobel Prize in Economics for 2016. Their research is the bread and butter of modern economic theory and hence is taught in all first-year PhD microeconomics courses and more applied versions are taught to MBAs in electives on organizational economics.
Let me begin with the work of Bengt Holmström. The prize announcement begins with his work on the principal-agent model with moral hazard: An agent privately chooses an action that impacts the welfare of a principal. The principal observes noisy signals of the action and rewards the agent as a function of the signals to align incentives. Holmström asks: Which variables should be included by the principal is her performance measure and which should be omitted? In the “sufficient statistic” result, he shows that variables should be included if and only if they contain information about the action. Adding more signals into a performance measure would add superfluous noise into payments which the a risk-averse agent would have to be compensated for. On the other hand, subtracting informative signals from a performance would eliminate useful information for aligning incentives.
This result poses a puzzle which Bengt turns to in later work: Real-world contracts are rarely as complex as the informativeness principle would suggest. Why is that the case? In joint work with Paul Milgrom, Holmström introduced the multi-task principal agent model. The main innovation was to allow the agent to perform multiple tasks and to substitute from one to the other. Holmström and Milgrom show that in certain circumstances it is better not to may pay responsive to performance. Suppose someone working in a fast food restaurant can look after the kitchen or sell burgers. Burger sales are measurable but time spent looking after the kitchen is not. Then making pay depend on burger sales can backfire as the agent substitutes away from looking after the kitchen. Better to have low-powered incentives which are relatively flat in burger sales.
Bengt has made at least two other seminal contributions to moral hazard models. In his work on moral hazard in teams he shows that it might be impossible achieve total value maximizing outcomes when joint output is measurable but individual output is not. In his career concerns model, he shows that an agent trying to prove he is high ability to a market might work too hard at the start of his career and then tail off at the end. All these papers are workhorses of applied theory. They show Holmström’s flair of coming up with models that serve as vehicles for others to make interesting contributions to understanding incentives.
I want to end my appreciation of Bengt Holmström by pointing out that several of these papers were written when he was an Assistant Professor in the MEDS Department at Kellogg. Roger Myerson has already won a Nobel Prize for the work he did when he was in MEDS.
Oliver Hart took contract theory in a different direction by emphasizing the role of property rights. In the principal-agent model, the principal might be an employer and the agent an employee. Or the principal might be one firm and the agent an independent subcontracter. In other words, the model cannot address the question of when trade should take place within a firm or across two firms. Building on some informal ideas of Oliver Williamson, Oliver Hart with Sandy Grossman and John Moore used the idea that contracts are incomplete to offer a unified theory of the optimal allocation of property rights. The key idea is that ownership of an asset confers residual rights of control so you can use it for production should a relationship break down. Suppose a buyer and a seller are trading a widget. They can both invest ex ante to increase the value of trade but because contracts are incomplete they must haggle over the price ex post. This means both are subject to hold-up: the benefit of any costly investment is shared with the other trading party. Hence, both will underinvest. This underinvestment is mitigated by the fact that if a player owns an asset he can use it to trade with others so at least he can capture some value from his investment. So, if one player’s investment is particularly important for value creation he should be own all the assets and employ the other – so we have an integrated firm. If both players’ investments are important, then they should both own assets and then we have trade across two independent firms.
Debt and equity confer decision rights in different ways. So, Oliver Hart’s way of looking at control rights has proved to be very fruitful in corporate finance. But there is a lot to be done. In particular, without a theory of why contracts are incomplete there is a tension between the lessons of mechanism design and the ideas in Grossman-Hart-Moore. This tension was pointed out by Maskin and Tirole. Eric Maskin was my PhD advisor at Harvard and Oliver arrived at Harvard just as I graduated. At that point, my friend, co-author and advisor Tomas Sjöström, who was sitting at the podium during the announcement as he is on the Nobel Committee, was an assistant professor at Harvard and I would visit to work with him. We became interested in Oliver’s ideas and also knew of the Maskin-Tirole critique. So, Tomas and I wrote a few papers studying optimal decision rights when agents can collude or renegotiate inefficient outcomes without falling afoul of the Maskin-Tirole critique. I continue to work on these questions still. I would never have worked on those papers without Oliver’s seminal insights to build on. So, I am particularly happy personally with this prize.