Spatial Analysis of Health Expenditure Effects of labor on Economic Growth in Iran (2011-16)

This paper aims to analyze spatial effects of the government health expenditure of labor on economic growth in 30 provinces of Iran during 2011-16. To this end, the Generalized Method of Moments (GMM) and the Arellano-Bover/Blundell-Bond estimators are used. A dynamic panel data (DPD) model is also used to increase the accuracy of results. The Spatial Durbin Model (SDM) is specified in the form of a Spatial Dynamic Panel Data model (SDPD), or the Generalized Spatial Generalized Method of Moments ‎ (SGMM), and is estimated by using GeoDA and Stata softwares. The results show that the estimated coefficient of lagged per capita income is negative and statistically significant in conventional econometric model, which confirms the conditional convergence hypothesis in Iran's provinces. The coefficients of labor and health expenditure variable are positive and significant in both models. Accordingly, if the health expenditure increases by one unit, the level of productivity of the workforce will rise and lead to an increase in gross domestic product. The life expectancy has a very small but significant effect on the per capita income of the provinces. The coefficient of the spatial coefficient of labor and health expenditure (W * health) is significantly different from zero. Thus, W * health has a positive and significant effect on per capita income and growth rates. The spatial lag coefficient of the dependent variable (rho) is negative and significant, which indicates negative spatial (proximity‎) effects on per capita income among provinces. The effect of education level of the workforce on the economic growth is positive and significant in different regions. Of variables with direct relations, the least impact devotes to the state budget and most impact belongs to health expenditure. The birth rate has no effect on the economic growth of the regions. Estimation of spatial model in addition to conventional form increases the validity of the results of panel data models