Living Rationally Under the Volcano? Heavy Drinking and Smoking Among the Elderly

Most rational addiction models focus on how drinking and smoking are made when young. Yet, the costs of drinking and smoking generally come later in life. We focus on the decisions of the elderly where individuals know their propensity for addiction but are uncertain about their future earnings and helath status. Using data from the Health and Retirement Survey, we estimate a dynamic stochastic model of heavy drinking and smoking of the elderly. Individuals make decisions not only based upon the current effects of heavy drinking and smoking, but also the future effects of drinking and smoking on earnings, health, and mortality. We are especially interested in the identification of the discount factor. We show how the likelihood function varies with the discount factor and also how behavior decisions vary from not estimating a dyanmic model. In particular, we find that the dynamic model forecasts more drinking and smoking as well as individuals living longer. This is because individuals know when, and when not to, engage in heavy drinking and smoking behavior.