Selecting Plans to Enable Moral Hazard: Anticipatory Selection in Health Insurance

Build an empirical test for selection on anticipated moral hazard in health insurance. Use variation in plan features that plausibly change the “exploitable” space—prior authorization strictness, referral requirements, network breadth, out-of-network billing protections, or coding audit intensity—and ask whether plan choice is disproportionately higher among enrollees with traits predictive of ex post opportunistic behavior (e.g., prior patterns of high-intensity care, propensity to visit high-upcoding providers). This idea brings the mechanism of selecting into contracts anticipating moral hazard, previously shown in crop insurance, to health insurance and makes it testable with plan features that facilitate ex post behavior. It adds a new pathway to disentangle adverse selection from moral hazard by explicitly modeling selection on the ability to exercise moral hazard, not just selection on risk. This can reconcile puzzling cases where measured moral hazard looks small but adverse selection persists, suggesting people may choose plans for exploitable rules rather than to increase total utilization. The research has clear policy implications: tightening exploitable plan features or standardizing them could reduce harmful selection without harsher underwriting, providing insurers and regulators a new lever to reduce selection distortions while preserving risk protection.

References:

1. Experimental identification of asymmetric information: Evidence on crop insurance in the Philippines. Snaebjorn Gunnsteinsson (2020). Journal of Development Economics.