Do Cross-Section Dependence and Parameter Heterogeneity Matter? Evidence on Human Capital and Productivity in Greece

This paper estimates production functions for Greek regions over 1981-2003, using a novel human capital dataset. We construct rich human capital series, where data for employees are decomposed according to their education level. Our empirics include recent non-stationary panel techniques, allowing for cross-section dependence and parameter heterogeneity in production technology, along with fixed effects and dynamic panel estimators. We show that ignoring cross-section dependence and parameter heterogeneity has a serious distorting impact on the estimated results, both qualitatively and quantitatively. Our evidence shows that tertiary education has a strong positive association with labor productivity growth, while secondary education exhibits a negative relationship. Primary education, public capital and net agglomeration do not display any relation with growth. Overall, findings suggest that policy makers should account for spillovers alongside technology heterogeneity and direct their efforts towards the expansion of tertiary education in poor regions in order to promote convergence.