Self-assessed Health and Cigarette Smoking in China

Self-assessed health (SAH) is often elicited in surveys in lieu of expensive health assessments made by medical professionals. Measures like this are quite often used in explaining decisions to undertake various activities, but their validity as an exogenous variable has been questioned. If the SAH is improperly treated as an exogenous variable this could bias other coefficients (e.g. price) that are important for policies that attempt to curb smoking. This paper takes a first step in sorting out the relationship between SAH and smoking behavior: we examine whether the variation in SAH can be explained as an endogenous dependent variable. An ordinal endogenous treatment is modeled with an ordinal smoking variable, using the copula approach to accommodate a non-Gaussian distribution in the error terms. The treatment approach avoids a wide variety of selection issues that could bias coefficients. The empirical model is estimated for a random sample of the adult males from nine Chinese provinces in the 2006 China Health and Nutrition Survey. For our sample the SAH has negative effects on the probabilities of smoking heavily, and such negative effects are most pronounced in the effect of the higher health category.