But this negative association is statistically fragile and sensitive to model specification (see e.g. Levine and Renelt, 1992, and Tanzi and Zee, 1997). In their widely cited critique, Levine and Renelt (1992, p. 951) have emphasized two reasons for non-robustness: (i) The overall size of government cannot capture the different implications of different government activities. Hence, several authors use disaggregated data to investigate the growth effects of specific categories of public expenditure and taxation (see e.g. Kneller et al., 1999). (ii) Ignoring government efficiency may yield inaccurate measures of the actual delivery of public services. Hence, Angelopoulos and Philippopoulos (2005) construct an indicator of public sector efficiency and provide evidence that there is a non-linear effect of government size in the sense that “efficient” and “inefficient” governments are situated on different sides of a Laffer curve in macroeconomic performance.
An obvious shortcoming of cross-section and panel studies is that they rely on the assumption of common coefficients across countries although different countries may have different structures. This naturally leads to time-series studies of single countries.4 But time-series studies are still liable to the criticisms of Levine and Renelt (1992) discussed above. In addition, they may suffer from two well-known problems: (a) If the series are non-stationary, the regressions might be spurious. (b) The effect of government size might be biased due to endogeneity if higher government spending is triggered by negative income shocks.
Our paper uses annual data for the period 1960-2000 to study the growth effects of fiscal policy in Greece.5 In doing so, we will try to deal with all the above problems. Namely, we will first pay attention to the time-series properties of the data to avoid the spurious regression problem and to investigate for possible breaks or regime switches; second, we will test for exogeneity of government size; third, we will examine whether different categories of government expenditures and taxes have different implications for growth; fourth, we will examine whether the data reveal a non-linear growth effect of fiscal policy depending on the mix between government size and government quality.