Dynamic analysis of regional convergence in Indonesia

This study examines income convergence among provinces in Indonesia using dynamic panel data approach. The results show that static and dynamic panel data approaches produce different results of convergence patterns. Consistent with the theory, the Ordinary Least Square (OLS) and fixed-effects estimators provide the upper and lower bounds. The first-differences generalized method of moments (FD-GMM) provides invalid estimators which are lower than the coefficient from the fixed effects estimators due to the weak instruments problem. The system-GMM (SYS-GMM) estimators are found to be unbiased, consistent and valid. They show that convergence process prevails among provinces in Indonesia for the period 1983 – 2003. However the speed of convergence is relatively very slow (0.29) compared to other studies in developing countries