Abstract
A Representative Assembly (RA) consists of representatives electedbycitizens. ACitizenAssembly(CA)consistsofrandomlyselected citizens. I study the normative performance of these two institutions in a game-theoretic model with inattentive voters. The key question is which assembly creates incentives for voters to learn about the decision problem. Because I restrict how the two assemblies affect the probability of being pivotal, I can focus on issues other than the classical paradox of voting. The reason why the RA is inferior to the CA is that it forces the voters to learn about politics instead of policies. Since only policies matter, this is inefficient. The RA may be superior if the voter’s preferences are highly correlated with those of the candidate, but a hybrid CA model (with candidates relegated to mere experts) is not worse and sometimes strictly better. I discuss the role of media in electoral campaigns and a number of other extensions.
Author - Maksymilian Kwiek