In this paper we empirically analyze how different forms of democracy affect redistributive
spending programs, the size of government and political participation. Specifically, we
compare a representative democracy with direct democracy, i.e., town-meeting, using a very
large data set on Swedish local government for the period 1919-1950. Due to the Swedish
Local Government Act, we can implement two different design-based strategies: a
regression-discontinuity design and a nonparametric instrumental variables approach. Our
results indicate that going from a direct democracy to a representative system dramatically
increases political participation, redistributive spending, and the size of government. The
estimated effects on public spending to the poor (poverty relief, child welfare and basic
public education) are on the order of 35-70 percent while the effect on political participation
is between 150-200 percent. We argue that these results most likely reflect that direct
democracy is more prone to capture by (rich) local elites than representative democracy. We
present further evidence that supports the theoretical framework developed by Acemoglu and
Robinson (2006, 2008) based on the persistence of de facto political power.
2010. 1-59 p.