This project examines the total impact of democracy on behavior and argues that the total impact may consist of two effects: an instrumental effect and a direct effect. The instrumental effect refers to the fact that democracy may lead societies to choose different policies. The direct effect refers to the fact that the response of people to a given policy may depend upon whether or not it was chosen democratically. The idea that democracy may have a direct effect on behavior can be traced back to Tocqueville's Democracy in America. This project uses the methods of experimental economics to gain some insight into the relative importance of these two effects.
The difficulty of measuring the direct effect of democracy lies in the fact that democratic societies differ from undemocratic ones in many and complex ways. This makes it inadvisable to simply compare democratic and undemocratic societies. It is possible, however, to examine whether experimental subjects respond to policies that are democratically selected in a different way than they do to the same policies imposed undemocratically. It is also possible to test for alternative mechanisms that may underlie any observed direct effects of democracy.
The research proposed here is expected to have two different types of broader impact. First, evidence on whether and how democratic institutions affect the types of policies adopted and the responses to these policies can aid in the design of efficient institutions. Second, the results may cast doubt on the validity of the policy recommendations derived from some empirical studies. Much applied work in social sciences seeks to identify the treatment effect of policies, institutions, or products. Since people usually choose their policies, institutions, and purchases, it is usually necessary to account for difference between groups to measure the "true" treatment effect. Based on such estimates, policy recommendations may be made that involve assigning the treatment without choice. If the treatment effect differs based on whether it is chosen or imposed such policy recommendations may be unwarranted.
