ABSTRACT This article analyzes disease-specific moral hazard effects in the demand for physician office visits and explores whether optimal insurance for physician services should be designed to have disease-specific cost sharing. Generalized method of moments is implemented to address the endogeneity of private health insurance, and the nonnegativity and the discreteness of physician services use. The results indicate that the moral hazard effect varies considerably across disease-specific specialist care. The strongest moral hazard (for no-condition related specialist visits) is almost twice the moral hazard effect of the weakest (for chronic condition related specialist visits). Although the findings indicate some variation in the moral hazard effect across disease-specific general practitioner visits, the variation is less considerable. The main policy implication is that optimal insurance for physician services should be designed to have differential cost sharing based on disease status rather than to have uniform cost sharing. (ProQuest: ... denotes formulae omitted.) INTRODUCTION Private health insurance policies provide for a price reduction at the time the insured purchases medical care. Consequently, an insured individual would respond to the price decrease by purchasing more medical care than he would have purchased at the market price, ceteris paribus. This effect of insurance on medical care demand is known as the ex post moral hazard effect (Arrow, 1963; Pauly, 1968).1 It refers to the effect of insurance on the net price of medical care and to the consequent incentive effects on medical care consumption.2 This article analyzes disease-specific moral hazard effects (i.e., the effect of health insurance on the demand for medical care associated with similar disease states) in the demand for physician office visits. The article disaggregates speciahst and general practitioner visits into three disease-specific components and seeks to identify whether there is any variation in the magnitude of moral hazard across them. The outcome has implications for optimal health insurance design, in particular, whether health insurance for physician services should be designed to have differential cost sharing that depends on the respective demand characteristics of individual diseasespecific service categories. One contribution of this article is to the recent literature that analyzes disease-specific moral hazard effects.3 An individual's preferences toward physician care may vary by disease state, which suggests that physician services associated with different diseases may have different marginal benefit configurations; therefore, the price responsiveness of the individual-consumer physician care demand may vary by disease state. Consequently, the moral hazard effect may vary across physician services associated with different disease states, since if the individual-consumer demand for a physician service is less (highly) elastic, the extent of moral hazard would be lower (higher). Disease-specific physician services variables are constructed using the three-digit ICD-9-CM disease classification codes.4 Employing data from the Medical Expenditure Panel Survey, each specialist and general practitioner service category is classified in three groups: services associated with chronic conditions, services associated with acute conditions, and services that are not associated with any medical conditions. Another contribution of this article is to the health insurance design literature that emphasizes the state specificity of optimal insurance policies. The conventional theory of the demand for health insurance, which suggests that individuals buy insurance to achieve certainty with regard to the consumption of other goods, motivates diseasespecific cost sharing. According to this theory, although the risk avoidance aspect of insurance benefits consumers, it also creates a welfare loss, since the moral hazard consumption involves consumption of medical services whose value to the consumer is less than its cost of production (Pauly, 1968; Feldstein, 1973; Feldman and Dowd, 1991; Marining and Marquis, 1996). …