Chronic Fiscal Deficits and Chronic Inflation in the Burundian Economy: An Empirical Test of the Validity of the Olivera–Tanzi and Patinkin Effects Using ARDL Models

Abstract

For nearly two decades, the Burundian economy has faced challenges in public finance management. On the one hand, public spending has increased dramatically, and revenues are no longer sufficient to cover it. As a result, the economy is experiencing recurring budget deficits. On the other hand, inflation continues its upward trend, becoming increasingly chronic and pervasive, steadily eroding the purchasing power of the average citizen. Although this situation is underappreciated, the coexistence of twin problems—chronic deficits and chronic inflation—raises fundamental questions about the impact of inflation on public finances, particularly with respect to the theoretical "Olivera-Tanzi" and "Patinkin" effects. Therefore, this study aims to empirically analyze the Olivera-Tanzi and Patinkin effects in the Burundian economy. To achieve this objective, the study uses real tax revenues and real public expenditures as dependent variables. Inflation is the main independent variable, whereas real GDP, the money supply, and the real interest rate are control variables that may influence the relationship between the main independent variable and the dependent variable. The data are quarterly, spanning 2004:Q1 to 2024:Q2. The ADF and PP unit root tests showed that the models’ series were integrated of order I(0) and I(1), respectively. This indicates that the ARDL approach and the bounds test for cointegration should be employed. Additionally, after applying the CUSUM and CUSUMSQ stability tests to both models, it was found that the real tax revenues (the Olivera-Tanzi effect) model exhibited potential breaks, particularly from the second quarter of 2015 onward. This break corresponds to the political violence that plagued the country from the beginning of that quarter. Accordingly, this paper introduced a dummy variable (PV = Political Violence) into the ARDL model of real tax revenues. The Olivera–Tanzi effect model results indicate that, in the short run, a one–unit increase in the contemporaneous inflation rate leads to a 0.68% decrease in real tax revenues, and a one–unit increase in the past inflation rate leads to a 0.54% decrease in real tax revenues. Hence, in the short run in Burundi, the Olivera–Tanzi effect holds. In the long run, a 1% change in real GDP leads to a 2.80% change in real tax revenues. In the long run, the presence of political violence resulted in a 23.4% decrease in real tax revenues relative to the period without political violence. According to the Patinkin effect model, contemporaneous inflation does not significantly affect real public expenditures. In contrast, a 1% increase in the money supply is associated with a 1.91% decrease in real public expenditures, as observed in the second quarter. In the short run, a 1% increase in real GDP leads to a 1.33% increase in real public expenditures. However, this effect appears to change over subsequent quarters. Furthermore, in the long run, a 1% increase in real GDP is associated with a 2.39% increase in real public expenditures. Conversely, a 1% point increase in inflation leads to a 1% increase in real public expenditures. This result, therefore, reinforces the Olivera–Tanzi effect rather than the Patinkin effect. Thus, in Burundi, the Patinkin effect postulate does not hold.