The Macroeconomic Effects of Tax Competition: the Brazilian "Fiscal War'' and Public Goods Underprovision

 

Social welfare and economic development can be hindered by an under-provision of public goods. This paper studies the role of fiscal wars between state governments in lowering overall welfare, public goods provision, and income levels. To do so I build a multi-region general equilibrium model with endogenous state taxes, local public expenditures, and firm location choices. I then estimate the model to match a novel dataset encompassing state-level tax cuts in Brazil. Under my most conservative estimates, I find that moving towards a centralized system of VAT taxes would increase public goods provision by 21-27 percent, GDP per capita by 1.5-4 percent, and aggregate welfare by 1-4 percent. Tax centralization would, however, likely generate winners and losers and can be regressive.

The costs of running a minority government

 

Executive branch representatives must garner support from elected legislative officials to govern. Building a legislative majority is an important stepping stone in most executive-branch mandates. This majority-building process may, however, impose significant costs upon society. Using a regression discontinuity design, I show that municipalities in Brazil whose mayors hold few seats in the municipal chamber experience substantially more turnover in their bureaucracy. RD estimates demonstrate that non-tenured civil servants are hired (+46.7 percent) and fired (+37.5 percent) at substantially higher rates under minority mayors. This turnover is not confined to high-ranking government positions but extends to roles filled by both skilled and unskilled public servants. Using teacher and school principal surveys, I show that these new hires are generally inexperienced workers who fall short in indicators of job performance. Ultimately, municipalities that elect a mayor with limited legislative support experience a significant drop in standardized test scores (-0.048 to -0.073 std. dev.).