Analysis of Asymmetries in the Tax-Spending Nexus in Burundi

Abstract

This study examines the asymmetries in the tax-spending nexus for Burundi using a three-variable model. The study employs a threshold cointegration test with asymmetric adjustment advanced by Enders & Siklos (2001). The findings indicate that government spending, taxes and grants are cointegrated with asymmetric adjustment. Causality tests from the estimated asymmetric error correction model suggest that in the short run there is an independent relationship between government spending and taxes in Burundi. Pertaining to the impact of grants, the results show that grants encourage the government of Burundi to spend more, but, conversely, they also discourage tax revenue, which is known as the tax displacement hypothesis. The findings further show that only government spending responds to budgetary disequilibrium, and this occurs when the budget situation is worsening. This implies that in Burundi, to restore the equilibrium when the budget situation is worsening, the adjustment is made by reducing spending. A policy intuition arising from this study is that, to reduce budget deficits, Burundi should reduce its grant-dependence and improve its tax collection system as well as cut its spending in sectors where it is not productive and reallocate it to more productive sectors.