Partially Binding Platforms and the Advantages of Being an Extreme Candidate

This paper develops a model in which platforms are partially binding: a candidate who implements a policy different from his platform must pay a cost of betrayal that increases with the size of the discrepancy. I also suppose that voters are uncertain about candidate preferences for policies. If voters believe that a candidate is likely to be extreme, there exists a semi-separating equilibrium: an extreme candidate mimics a moderate candidate with some probability, and with the remaining probability, he announces a platform that is more moderate than a moderate candidate’s platform. Although an extreme candidate will implement a more extreme policy than a moderate candidate in equilibrium, partial pooling ensures that voters prefer an extreme candidate who does not pretend to be moderate over an uncertain candidate announcing a moderate candidate’s platform. As a result, an extreme candidate may have a higher probability of winning than a moderate one.