Abstract. This paper tests the tax smoothing hypothesis for Tunisia using annual data for the period of 1972-2015. According to this approach, an optimal fiscal rule is to smooth tax rates over time and to finance temporary difference between government expenditure and tax revenue by debt creation. Tax smoothing implies that the tax rate behaves as a random walk and changes in the tax rate are nearly unpredictable. For this reason, two tests are used. The first, is a unit root test performed on Tunisian data to examine the null hypothesis of non-stationary of the tax rates. The results show that the null hypothesis of the unit root cannot be rejected, indicating that the tax rate is nonstationary and, thus, it follows a random walk. The second is the cointegration test which indicate that the future tax rate is cointegrated with the current permanent government expenditure rate. Therefore, Tunisia is trying to keep close correspondence between these two variables. In this context, the best policy option would be an initiation of prompt action program of tax base expansions and expenditure rationalization.

Keywords: tax smoothing, Tunisia, fiscal policy, permanent expenditure.