T his paper examines the impact of the form of government--presidential or parliamentary--on fiscal outcomes in democratic systems. Based on data for democracies in 98 countries between 1970 and 2002, it shows that the gross domestic product ratio of the central government budget balance is higher in presidential than in parliamentary democracies. It also shows that this impact is not due to the fact that presidential systems are not subject to the "costs of coalition" that allegedly afflict parliamentary democracies: the coalition and status of the government are of no consequence for budget balances in either presidential or parliamentary systems. Presidential systems matter for budget balances because they generate relatively high incentives for governments to keep budgets under control. They do so because in presidential systems, unlike in parliamentary systems, voters are by design able to identify and punish those responsible for economic policies. Presidents, however, vary in their capacity to affect budget policies. This paper demonstrates that presidential systems in which presidents are constitutionally able to dominate the budget process or to effectively veto legislation tend to have higher budget balances than those in which the budget process is dominated by the legislature or the president is unable to exercise existing veto powers. D oes the form of democratic government matter for economic outcomes? Specifically, does it matter for economic performance whether a country has a presidential or a parliamentary constitution? In this paper I show that it does, at least when it comes to fiscal outcomes: on balance, budget deficits are smaller in presidential than in parliamentary democracies. The reason, I suggest, has to do with the way presidential constitutions define the relationship between the voters and the government; specifically with the fact that electoral identifiability--the ability of voters to identify and punish those responsible for economic policies--is by design high in presidential systems, thus generating incentives for the president--the head of the government--to keep budgets under control. Moreover, I show that, given their incentives to control budget deficits, presidents vary in their capacity to do so.Most of the comparative studies of presidentialism have focused on the impact that a system in which the executive and the legislature are independent from one another might have on the survival of democracy. There were empirical and theoretical reasons for this concern. A cursory look around the world shows that there is only one long-living democracy that is also presidential--the United States. At the same time, Latin America, the region of the world where presidential institutions have dominated since the 19th century, is also the region with the highest level of regime instability, understood here as shifts between dictatorship