Modeling Medicare-Eligible Veterans’ Demand for Outpatient Services: A Two-Stage Approach

Abstract

We apply a utility consistent, combined multinomial choice and count data model to identify factors that affect Medicare-eligible veterans’ outpatient health care usage and provider choice. Our first stage count data regression shows that a shadow cost index for visits calculated from the second stage multinomial choice model is significant in determining the Medicare-eligible veterans’ demand of outpatient services. In the second stage, we specify a multinomial choice model to study veterans’ allocation of outpatient visits between the VA and non-VA health care facilities, and find that veterans’ actual out-of-pocket cost and the distance to the health care facility have significantly negative effects on the probability of choosing the alternative. A number of other factors including family income, means of transportation, race, employment, and disability status are also found to be important at various stages of the decision making. We find some evidence of moral hazard effect in the market for private supplemental health insurance and prescription drugs, but adverse selection does not seem to be a problem for this special population.