Pesko, Michael, and Christopher F. Baum. “The self-medication hypothesis”. Boston College Working Papers in Economics 865, 2016. http://hdl.handle.net/2345/bc-ir:104590.
We use single equation and simultaneous instrumental variable models to explore if individuals smoke during times of stress (i.e., motivation effect) and if they are successful in self-medicating short-term stress (i.e., self-medication effect). Short-term stress is a powerful motivator of smoking, and the decision to smoke could trigger biological feedback that immediately reduces short-term stress. This feedback confounds estimates of the relationship between stress and smoking. Omitted variables, such as genetic or social factors, could also suggest a spurious correlation. We use data on self- reported smoking and stress from 240,388 current and former smokers. We instrument stress with temporal distance from September 11, 2001 (using date of interview). We instrument smoking with cigarette accessibility variables of cigarette price changes and distance to state borders. In the absence of accounting for feedback and other forms of endogeneity, we find that smoking is associated with increases in short-term stress. This is opposite of our theoretical prediction for self-medication. However, when we account for endogeneity we find no evidence of smoking affecting short-term stress. We do find a consistent positive effect of stress on smoking.