This research consists of three distinct projects. The first investigates decisionmaking in the presence of "temptations," options that the person knows he should not choose but finds difficult to resist. While familiar to anyone who has dieted or quit smoking, for example, it is difficult to reconcile the phenomena with the rational actor presumed by standard economics. There are important economic issues linked to such temptations. For example, why do people seem to save too little for retirement? Presumably, this is because they are tempted to consume too much when younger. What is the appropriate role for public policy in such circumstances? This project will develop new mathematical models of how agents make decisions in such settings and, as such, will cast new light on these questions.
The second project also considers more realistic models of decisionmaking, this time by introducing the possibility that there are some courses of action available to the decision-maker that he does not currently recognize. Many times, the solution to a difficult problem is to think of a new option, not to insist on standard ideas. In writing long term contracts, for example, the agents must recognize that they cannot foresee all possible courses of action that might be followed in the future. Yet they need to specify what should be done in specific enough terms to ensure proper performance under the contract. As with the first project, this research will develop new mathematical models for studying such issues.
The third project investigates a very different issue, namely the Electoral College. While it has not happened often in US history, the Electoral College need not yield the same winner as a popular election would. Many commentators have suggested that this must make the Electoral College inferior to popular election and have advocated its abolition. Surprisingly, there has been very little careful investigation of the effects the Electoral College has and its costs and benefits. This research will characterize situations in which the electoral college is more "efficient" than popular election and when the reverse is true. In fact, certain kinds of regional differences can make the Electoral College a very valuable institution. For example, if the main role of the election is to aggregate the information voters have, then it is important to control for correlation among voters. That is, the information held by the 10 million voters in California would not constitute 10 million "data points" since the information is correlated within the state. Hence it would be important to "rescale" this information before aggregating it across states.
