Nonparametric Estimates of Demand in the California Health Insurance Exchange

We develop a new nonparametric approach for discrete choice and use it to analyze the demand health insurance in California Affordable Care Act marketplace. The model allows endogenous prices instrumental variables, while avoiding parametric functional form assumptions about unobserved components of utility. estimate bounds on effects changing premiums or subsidies coverage choices, consumer surplus, government spending subsidies. find that $10 decrease monthly premium would cause decline between 1.8% 6.7% proportion subsidized adults with coverage. reduction total annual surplus be $62 $74 million, savings yearly subsidy outlays $207 $602 million. impacts linking age, finding shifting from older younger buyers increase average potentially large enrollment. also impact removing highly?subsidized plans Silver metal tier, where we is consistent wide range possibilities. comparable mixed logit models tend yield price sensitivity estimates toward lower end bounds, producing can both higher than depending specification random coefficients.