Nonlinear Causality between Education and Health: the Role of Human Development Index

This study employs a panel smooth transition vector error correction model (PST-VECM) to explore the education-health causality. This model can grasp the characteristics of the causality in nonlinearity, heterogeneity, and time-variation. Empirical evidence from the 123 sample countries during 2001–2013 finds that the causality is nonlinear, time- and country-varying in both the long run and short run, depending on the development level of a country (measured by the human development index, HDI) in each period. For the developing and underdeveloped countries, the short run causality between education and health is unidirectional, running from education to health, whereas the long run causality is bidirectional. For the developed countries, the causality is bidirectional in both the short run and long run, and the two-way effect would be deferred from short run to long run. The policies for countries with different HDI scores to improve education and health outcomes are also provided.