Unpriced and unseen: private information and taxi insurance purchases in Taiwan

The role of information asymmetry in automobile insurance markets has been studied extensively. However, not all empirical studies found correlation between driver risk and chosen coverage. There are two explanations for this (Cohen and Siegelman 2010): (1) information asymmetry may not characterise environments in which consumers cannot act on private information; (2) most studies measure not risky driver behaviour, but ex post realisations of such behaviour, e.g. claims. In Taiwan’s taxi insurance market, insurers cannot freely use rating factors. Usage density (e.g. daily distance), a direct measure of risk, therefore constitutes private information. Using ordinary least squares models, we find that usage density is positively correlated with total premiums. Using logistic models, we find high-density drivers more likely to buy additional policies. Thus, usage density is positively correlated with insurance demand, consistent with the risk-coverage hypothesis. Unlimited-mileage coverage may disproportionately attract high-risk drivers, an argument in favour of usage-based insurance.