The first chapter examines the effect of the composition of federal and state government spending on SO2 air concentrations in the US. The results indicate that a reallocation of spending from RME to PME at the state and local level reduces sulfur dioxide concentrations while the composition of federal spending has no effect. A 10% percent increase in the share of PME spending reduces sulfur dioxide concentrations by the range of 3 to 5% for state and local spending. This is a significant effect since sulfur dioxide concentrations have been falling at an annual average rate of 5% from 1980 to 2008. The results are robust to various sensitivity checks.

The second chapter documents the creation of a US government spending allocation database that provides new data on a set of disaggregated government spending categories covering all the states in the US for the period 1983-2008. The data allows for the comparison of federal versus state and local government spending over time on various spending items. This is achieved by categorizing and aggregating expenditures for over 1,500 federal programs and combining data on state and local government spending. The key challenge in separating federal and state and local government spending is the issue of double counting since part of state and local spending is from the federal government. The dataset presented will aid researchers in separately accounting for both state and local, as well as federal spending in future research.

Finally, the third chapter examines fiscal spending and economic growth in the presence of imperfect markets. Political economy factors tend to induce many governments to spend on private goods (RME) to the detriment of spending on social and public goods (PME). This bias in spending patterns is particularly costly for economic growth when capital markets are imperfect. A theoretical model on government spending and growth is developed and linked quite closely to an empirical model. The empirical results fully corroborate the hypothesis that spending biases in favor of non-social subsidies (RME) reduce the rate of economic over the long run. The empirical findings are exceptionally robust.