Our study reexamines standard econometric approaches for the detection of information asymmetries on insurance markets. We claim that evidence based on a standard framework with 2 equations, which uses potential sources of information asymmetries, should stress the importance of heterogeneity in the parameters. We argue that conclusions derived from this methodology can be misleading if the estimated coefficients in such an `unused characteristics' framework are driven by different parts of the population. We show formally that an individual's expected risk from the perspective of insurance, conditioned on certain characteristics (which are not used for calculating the risk premium), can equal the population's expectation in risk - although such characteristics are both related to risk and insurance probability, which is usually interpreted as an indicator of information asymmetries. We provide empirical evidence on the existence of information asymmetries in the market for supplementary private health insurance in the UK. Overall, we found evidence for advantageous selection into the private risk pool; ie people with lower health risk tend to insure more. The main drivers of this phenomenon seem to be characteristics such as income and wealth. Nevertheless, we also found parameter heterogeneity to be relevant, leading to possible misinterpretation if the standard `unused characteristics' approach is applied.

JEL classification: I11; I13; C18