Economics

We analyze the e¤ectiveness of some commonly discussed anti-corruption reforms on political corruption, using a theoretical model of competition between two candidates in a probabilistic voting setup. Candidates, who may di¤er both in their ability to produce the public good, and popularity with voters, propose a tax rate and a public good level. The budget constraint implies that taxes collected must equal the sum of funds used in public good production plus funds stolen by the elected politician. We identify the conditions under which constitutional constraints on policies, higher penalties for corruption, and higher wages for elected politicians increase (or decrease) voters'welfare. We discuss how the asymmetric information and the rigidity of constitutions reduce the e¤ectiveness of the reforms, and how distributional e¤ects of reforms may reduce the voters'support for a welfare-improving reform. Finally, we argue that e¤ective reforms may not be proposed by both corrupt and honest politicians.

This paper proposes a mechanism for the regulation of duopolies: a revenue contests among the firms. Under the mechanism, the firm with the lower revenue is to pay a penalty to the firm with the higher revenue proportional to the difference between their revenues. In a homogenous good Cournot duopoly with convex cost and demand functions, the mechanism implements the optimal outcome when the firms have symmetric costs. When one firm is more efficient, the mechanism leads to increased social surplus under a large set of parameters.

We study the effects of stochastic (probabilistic) voting on equilibrium locations, equilibrium vote shares and comparative statics in a setup with three heterogenous candidates and a single-dimensional issue space. Comparing the equilibria with and without stochastic voting, we find that under an appropriate level of uncertainty about voter behavior, the model has a pure strategy Nash Equilibrium (PSNE) that is free from several non-plausible features of the PSNE under deterministic voting. The results are robust to extensions to asymmetric density and plurality maximization.

Prior studies of strategic voting in multi-party elections potentially overestimate the extent of it by counting erroneously votes cast under different motivations as strategic votes. We propose a method that corrects some of this overestimation by distinguishing between strategic voting (voting for a candidate other than the most preferred one to reduce the likelihood of an election victory by a third candidate that is disliked even more) and the votes cast under the "bandwagon effect" (voting for the expected winner instead of the most preferred party to conform to the majority or to be on the winning side). Our method follows from the observation that a vote cannot be strategic unless the voter believes that it will affect the outcome of the election with a non-zero probability, while a vote cast under the bandwagon effect requires no such belief. Employing survey data that include the respondent"s assessment of the importance of his vote, we illustrate this method by estimating the extent of strategic voting in the 2005 UK general election. The estimated extent of strategic voting (4.22%) is strictly less than self-reported strategic voting (6.94%), but the discrepancy cannot be attributed in a statistically significant way to the bandwagon effect, suggesting that motivations other than those identified in the literature may be at work.

We study the effects of stochastic (probabilistic) voting on equilibrium locations, equilibrium vote shares and comparative statics in a setup with three heterogenous candidates and a single-dimensional issue space. Comparing the equilibria with and without stochastic voting, we find that under an appropriate level of uncertainty about voter behavior, the model has a pure strategy Nash Equilibrium (PSNE) that is free from several non-plausible features of the PSNE under deterministic voting. The results are robust to extensions to asymmetric density and plurality maximization.