Abstract: In markets where consumers seek expert advice regarding purchases, firms seek to influence experts, raising concerns about biased advice. We combine a model of supply and demand with a local instrumental variables strategy based on regional spillovers from academic medical center conflict-of-interest policies to estimate the distribution of marginal treatment effects of pharmaceutical firm payments on physician prescribing, accounting for frictions like market power, negotiated prices, and insured demand. We find substantial heterogeneity across physicians in expected response to payments. Firms target payments to physicians with: a larger expected response, larger patient panels, and lower than average prescribing of the focal drug. Counterfactual estimates of the equilibrium response to a payment ban suggest that payments improve allocation by offsetting the distortion of high prices for on-patent drugs. We explore sensitivity of welfare estimates from our model, varying to the extent to which payments and the interactions that accompany them improve vs. distort prescribing. Total surplus typically increases with payments, but most of the gain accrues to manufacturers. Consumers only gain from payments if they mostly improve prescribing.