Objectives: Moral hazard and adverse selection have been widely studied in relation to health insurance coverage choices. A related concept is that of selection on moral hazard, which implies that individuals select a specific coverage according to their different behavioural responses to health insurance. Another way to see this is a selection of health insurance coverage– i.e. the co-payment level – based on heterogeneous utilization “slopes” associated with different levels of coverage, rather than merely on heterogeneous utilization levels. However, the extent of selection on moral hazard in different health insurance markets remains an open empirical question. Using a mix of data from the Swiss Household Panel, and publicly available information from the health insurance markets regulator, we try to measure selection on moral hazard in the Swiss statutory health insurance. Switzerland has a peculiar consumer-driven health system, with a managed competition system where insurers offer plans in a strictly regulated market to all citizens, under a universal health insurance mandate.
Methods: We use the (log) number of doctor visits as proxy for healthcare utilization and study the response to unexpected injury or illness, which we argue to be plausibly exogenous conditionally on contemporaneous and lagged individual characteristics. Furthermore, to control for unobserved individual characteristics and overcome the inherent endogeneity between coverage choice and risk “types” related to adverse selection, we use both a FE panel design and an instrumental variable approach. We use both specifications to assess heterogeneity in responses across coverage levels, comparing individuals switching deductible from 2018 to 2019 to those maintaining the same level of coverage.
Results: Our results provide mild support for the mechanisms of selection on moral hazard. On average, individuals switching from highest to lowest deductible levels consume more healthcare, compared to individuals who had already selected the same lowest deductible level. The difference in conditional averages between these two groups – based mainly on IV estimates – suggests responses to the health shock between 1.3 and 1.5 stronger for the group of switchers. Analogously, we find a lower response in terms of doctor visits among individuals switching from lower to higher deductible, compared to “high-risk” individuals who had already selected the lowest coverage possible.
Discussion: The results seem to confirm the predictions of simple models of selection on moral hazard in health insurance, suggesting some scope for improvement in the regulation of the Swiss mandatory health insurance market.
Back to Research