The macroeconomic effects of fiscal policy shock in Ethiopia: evidence from a bayesian var approach

Abstract

This paper investigated the macroeconomic effects of fiscal policy shocks in Ethiopia using a Bayesian Vector Auto Regression model. We examined the dynamic responses of output, inflation, interest rate and exchange rate to fiscal policy shocks employing quarterly data from 2000/01Q1 to 2015/16Q4. The empirical evidence suggests that government spending shock had a positive impact on output and inflation but effect was too small. Initially the interest rate responded negatively to government spending shocks and was positive with small effect and the nominal exchange rate showed deterioration. In addition, government revenue shocks had positive effect real GDP and exchange rate and then they responded negatively. The inflation response to the net tax was medium and negative whereas its effect on interest rate was positive, and persistent. Furthermore, positive shocks to recurrent expenditure had a persistent positive impact on real output. Recurrent expenditure appeared to be responsible for inflationary pressure. Interest rate picked up slightly as a result of recurrent spending shocks in the short run. The response of exchange rate recurrent expenditure was small and remained negative. In contrast, capital expenditure was found to have an insignificant effect on output. The reasons could the administrative lag and contractual bottleneck that are sometimes involved executing capital projects and that appeared to be responsible for inflationary pressure. In the short term, the interest rate responded negatively and the estimated impact on exchange rate was insignificant. Following indirect tax revenue shocks rise in output and inflation was very persistent. Regarding the effects of indirect taxes on interest rate and exchange rate our results show a clear and negative impact whereas direct taxes were found to affect output and inflation very little and were insignificant. Initially, interest and exchange rates responded positively to direct shock later interest rate and exchange rate become insignificant and negative respectively. The results support the idea of a ‘crowding-in’ effect and when we take into account the feedback from government debt, the results suggest that the effects fiscal shocks on the majority of macro variables is too small except for the real GDP for government revenue shock. Therefore, empirical evidence shows that it important to consider government debt dynamics in the model.

Key words: Ethiopia, Fiscal policy shock, public debt, Bayesian VAR

JEL Code: E44, E62, C12, C32