The Demand for Health Care in Switzerland - A Latent Variable Approach

Abstract

In this paper, we develop and test a structural model of the demand for health and health care in Switzerland, treating health status as a latent (unobservable) variable which is characterized fully by its causes and indicators. Robinson and Ferrara (1977) first suggested this approach. Subsequently, it was used by Wolfe & Van der Gaag (1981), Van de Ven & Van der Gaag (1982), and Leu et al. (1984), among others. Analyzing health care demand in a structural model with latent variables provides two major advantages. First, it allows explicit specification of the complex relationship between health, income, health insurance, and the demand for health care. In our model, health and earned income are determined simultaneously, and health care demand is specified as a function of both health and income. The model distinguishes between three income sources - earned income, transfer income, and wealth income. The first two are modelled as endogenous. All three are specified as permanent income. Second, treating health as a latent variable allows inclusion of a large number of health indicators (partial health measures) for the overall unobservable health status. We use more than 100 variables covering physical, mental, and social health. Using factor analysis (principle components; varimax rotation), we reduce these variables to 28 factors which are included as indicators in the measurement model. Because of the structure of the model, we are able to obtain a one-dimensional overall health status measure (index) for each individual in the sample. This, in turn, allows comparisons of health status over time or for different population subgroups. The method can be used also for medical technology assessment (Leu 1984).