�鶹��ý

In retail and health care markets, demand declines with geographic distance to the establishment, but either transport costs or preferences correlated with distance (“home bias”) could cause this decline. Using hospital choices for childbirth, we find that, after controlling for home bias using fixed effects, estimates of the transport cost disutility fall by 50% relative to a standard logit model. We show that referrals are a likely source of home bias. Through two examples -- a simulated merger and a planned hospital move -- we demonstrate that controlling for home bias can imply greater substitution between geographically distant hospitals.