ABSTRACT
Endogenous regressors: the evaluation problem revisited The problem restated Chapter 1 introduced an example of the evaluation problem: how to estimate the ‘treatment effect’ of ownership of private insurance on the use of healthcare. From a policy point of view it is of critical importance to know whether, and to what extent, an observed association between insurance and utilisation is due to selection effects or to a direct utilisation effect or both. If the observed insurance effect is entirely due to self-selection of those more likely to use healthcare, then private insurance coverage merely acts as a marker for such propensity and reducing private insurance options will not reduce the use of care. If, on the other hand, the insurance effect is mostly due to the utilisation effect of increased coverage, then the expansion or reduction of private insurance options will have an impact on use.
