The Effect of Education on Health Behaviors and Obesity in Turkey: Instrumental Variable Estimates from a Developing Country

This study investigates the causal effect of education on health behaviors and obesity in Turkey, which is a middle-income developing country. Health Surveys of the Turkish Statistical Institute for the years 2008, 2010 and 2012 are used. The health behaviors considered are smoking, alcohol consumption, fruit and vegetable consumption, and exercise, and one health outcome, namely, obesity. We examine the causal effect of education on these health behaviors and on obesity. An instrumental variable approach is used in order to address the endogeneity of education to health behaviors. Educational expansion of the early 1960s is used as the source of exogenous variation in the years of schooling. Our main findings are as follows. Education does not significantly affect the probability of smoking or exercise. As the education level increases, the probability of alcohol consumption and the probability of fruit and vegetable consumption also increase. Nevertheless, some of the initial results change when we perform robustness checks. In order to check the robustness of our results, we narrow the time period around the policy years, since the impact of political and socio-economic developments may change the preferences of the individuals and hence invalidate our results. The evidence from the samples used for robustness checks suggests that there is a negative significant relationship between the probability of smoking and the years of schooling. Therefore, we can state that education can be a relevant policy instrument to fight smoking in Turkey. However, it is not a sufficient policy instrument to combat obesity, unlike in developed countries. This study provides a baseline for further research on the various aspects of health behaviors in Turkey. ResumeCette etude examine le lien de cause a effet entre l'instruction et les comportements lies a la sante ainsi que l'obesite en Turquie, pays en developpement a revenu intermediaire. Nous utilisons les enquetes sur la sante de l'Institut statistique turc pour les annees 2008, 2010 et 2012. Les comportements lies a la sante qui sont consideres sont le tabagisme, la consommation d'alcool, la consommation de fruits et legumes, l'exercice physique et l'un des resultats en terme de sante, a savoir l'obesite. Nous examinons l'effet causal de l'instruction sur ces comportements de sante et l'obesite. L'approche de la variable instrumentale est utilisee pour traiter l'endogeneite du niveau d'instruction par rapport aux comportements de sante. L'expansion de l'instruction du debut des annees 1960 est utilisee comme source de variation exogene au cours de la scolarite. Nos principales conclusions sont les suivantes. Le niveau d'instruction n'a pas d'incidence significative sur la probabilite de fumer ou de faire de l'exercice. mesure que le niveau d'instruction augmente, la probabilite de consommation d'alcool et celle de fruits et legumes augmente egalement. Neanmoins, certains des resultats initiaux changent lorsque nous effectuons des controles de robustesse. Afin de verifier la robustesse de nos resultats, nous reduisons la periode aux annees directement avant et apres des decisions politiques cles car l'impact des developpements politiques et socio-economiques peut modifier les preferences des individus et, par consequent, invalider nos resultats. Les preuves provenant des echantillons utilises pour les controles de robustesse suggerent qu'il existe une relation negative significative entre la probabilite de fumer et le nombre d'annees de scolarite