Three Models of Elections as Public Signals.

摘要

I present three game-theoretic models that treat elections solely as public signals of the incumbent regime's strength before a strategic interaction between the incumbent leader and an audience.;The first model asks why incumbent leaders would choose to hold an in this framework. I argue that elections are held (1) when the incumbent has private information, making the decision to hold a costly signal of strength and (2) when the incumbent can use the information generated by the election to manage political interactions more effectively, and (3) when there are multiple equilibria in the post-election game and the audience coordinates on playing an unfavorable equilibrium for leaders that do not hold elections. This framework can apply to both noncompetitive and competitive elections, and I argue that the third motivation for elections is required for equilibria where competitive elections occur and electoral rules followed.;The second model takes the election as a given and gives the incumbent the ability to distort the electoral outcome with hidden actions (i.e., fraud) and take actions to make this distortion more visible (i.e., election monitoring). Even though the model considers a noncompetitive election, the incumbent has incentives to cheat to look stronger, though this is ineffective in equilibrium as the audience is strategic and discounts the election result accordingly. As a result, the incumbent invites election monitoring to tie their hands and reduce the expectations of how much fraud is committed.;The third model places the election before a bargaining game between the incumbent and an opposition group. As suggested by the more abstract first model, the incumbent may hold an election to gather information and bargain more effectively. I then consider whether elections make conflict more or less likely, finding that they prevent violence in violence-prone cases but may increase violence in less violence-prone cases. However, under somewhat general assumptions elections decrease violence on average.