The Relationship between Government Expenditure and Economic Growth in Africa: Evidence from Dynamic Common Correlated Effects Panel Estimators

Abstract

The study on the relationship between government expenditure and the economy growth in Africa by employing a dynamic common correlated effects estimator, the ECM-ARDL approach. This model is robust to both heterogeneity and cross-section dependency. Three versions of Wagner’s hypothesis were investigated. The two estimators were considered, the dynamic common correlated effects - mean group (DCCEMG) and the dynamic common correlated effects – pooled mean group (DCCEPMG) estimators. Based on the results from residual cross-sectional dependence test and the residual mean square error, the DCCEMG was preferred. ECM-ARDL(1 2) was considered the most preferred optimal lags for model (I & II) and ECM-ARDL(1 1) for model III. The coefficient for the three models, lnRGDP for Peacock-Wiseman (model I), lnRGDPp for Gupta (model II) and lnRGDP for Goffman (model III) were positive and statistically significant in predicting real government expenditure and real government expenditure per capita as the case may be except for model II which was not statistically significant. For the three versions error correction mechanism showed the appropriate sign and were statistically significant at 1% level, suggesting causation running from economy growth to government expenditure. The coefficient expressed as percentage were 23.07%, 20.17% and 57.67% for model (I, II & III) respectively and indicated the speed of adjustment to equilibrium by the system annually. The results also suggested broad adoption of Wagner’s hypothesis among Africa countries.