Evaluating the Fiscal Policy Effects in Brazil Using Agnostic Identification
20 Pages Posted: 29 Oct 2008
Date Written: September 13, 2008
This article investigates the effects of fiscal policy shock in the Brazilian economy using quarterly data during the period between January/1995 and December/2007. We follow the agnostic procedure suggested by Monford and Uhlig (2005) to verify separately the impact of the unexpected positive impulse of current government spending and the net public revenues on some variables of the economy. The main advantages of this method regard it allows to isolate the fiscal impulse from the movements that comes from business cycle and the management of monetary policy. We find that in response of an expansionary shock of public expenditures the private consumption increases surely. It can indicate that there is some kind of crowding out effect with a reduction of private investment because the GDP contemporaneous decreases with 77.1 percent probability. The GDP reacts negatively with 56.6 percent probability immediately after a positive shock of the public net revenues. But in long run the probability of this response to be positive rises strongly. With 76.1 percent probability the private consumption decreases after this shock. Finally, another distinctive feature of the agnostic identification used in this paper pertains to the assessment the business cycle and monetary shock. With a 70.0 percent probability the real GDP decreases immediately after a contractionary monetary shock on the SELIC rate and this effect is negative and very persistent. Further, the most likely path of the price index (IPCA) indicates a drop of 0.4 percent in this variable during the first five months after a monetary shock. Considering the business cycle, government spending is not countercyclical in a view that during an economic boom the endogenous response of expenditure of government is positive.
Keywords: Fiscal policy, Vector autoregressions, Bayesian methods, Agnostic identification and Fiscal impulse
JEL Classification: C32, E60, H20
Suggested Citation: Suggested Citation