Political Institutions and Information AggregationPublic Deposited
In this dissertation, I analyse the effectiveness of various political institutions in aggregating private information. In the first two chapters, I consider diversity of information among the voters about given electoral alternatives and examine the conditions under which plurality rule voting aggregates such information. In the third chapter, I consider information private to electoral candidates and study the strategic provision of such information to the voters through political campaigns. In Chapter 1, I analyse elections where the utility of the voters from an alternative (A or B) depends on their preference type and the realization of an uncertain state (1 or 2), about which they receive a private signal. I show that information is aggregated in equilibrium for any voting rule short of unanimity if and only if the set of voters who favour A in state 1 includes the set of voters that favour A in state 2 (preference monotonicity). If there are any two groups of voters such that a change in state induces their rankings to change in opposite ways (preference reversal), there exist equilibria with outcomes different from those under full information. Preference reversal is generic if preference type is multidimensional, suggesting that elections may often fail to produce "correct" outcomes. In Chapter 2 which is joint work with Timothy Feddersen and Wolfgang Pesendorfer, we consider a more general model where a state is simply a vector of proportion of each signal in the population. We show that there exists a symmetric strategy profile that aggregates information if and only if the set of states in which each alternative is preferred by the majority is convex, provided that the majority is rarely indifferent between the two. This condition imposes a strong restriction on the set of environments in which information can be aggregated: aggregation may fail due to single-peaked preference over states or preference reversal. In Chapter 3, I study political campaigns as public debates where one candidate's willingness to misinform the voter is checked by the other's ability to inform. The "good" candidate wants to reveal relevant information to the voter, while the "bad" candidate wants to avoid such revelation. However, each candidate can discover the quality of the rival only through costly research. Therefore, debates take the form of a hide and seek game under incomplete information. Correct selection is assured only when the good candidate is sufficiently rare, and the probability of the good candidate winning the election may actually go down with an increase in the expected candidate quality.