Individual heterogeneity plays a key role in explaining variation in self-reported well-being and, in particular, health satisfaction. It is hypothesised that the influence of this heterogeneity varies over levels of health and increases over the life-cycle. These hypotheses are tested with data on health satisfaction from 22 waves of the German Socioeconomic Panel (GSOEP).Nonlinear fixed effects methods that allow for unobserved heterogeneity are not readily available for categorical measures of well-being. One common solution is to revert to conditional fixed effects methods, at the price of a high degree of information loss. Another common solution is to ignore the association between unobserved heterogeneity and socio-economic status by using pooled or random effects models, at the price of potential bias.We use a generalization of the conditional fixed effects logit, that allows for individual-specific reporting bias, heterogeneity in health endowments, and heterogeneity in the impact of income on health satisfaction. Adjusting for unobserved heterogeneity accounts for the relationship between income and very good health, but not between income and poorer health states. The income gradient for older age-groups is more strongly affected by controlling for unobserved heterogeneity: revealing an increasing influence of heterogeneity on health satisfaction over the life-span.
Panel data generalized conditional fixed effects logit generalized ordered logit health GSOEP