Robust Determinants of Public Sector Size under Uncertainty: The Study of OIC Member Countries (Bayesian Model Averaging approach)

This article examines the robust determinants of public sector (government) size in the Organization of the Islamic Conference (OIC) member countries (includeing Iran) during the years of 1996-2013, under uncertainty of model. For this purpose, is used of the 24 variables that effect on the government size according to theoretical and empirical studies in three categories: economic, political and social variables. Bayesian Model Averaging approach (BMA) is used, due to convenience feature to consider the uncertainty assumption of model. With estimation of 100000 regressions and Bayesian averaging of coefficients, robust determinants of government size are specified. According to the results, population and dependency ratio variables, respectively, with positive and negative symptoms have a certain effect on public sector size. The impact of land area and economic globalization variables on public sector size with a positive sign is relatively important. Coefficients and significant of dummy variables used in the model also shows that countries with a federal structure and countries with presidential systems, have a smaller public sector size and oil-exporting countries, have larger public sector size than the other countries studied.