Demand Induction with a Discrete Distribution of Patients

We test the "induced demand" model on cesarean section, relying on a natural experiment of proportional fee reduction of all delivery procedures. This fee change does not imply any substitution effect and allows a proper measure of the income effect. We extend the demand induction model and derive its testable implications by introducing a discrete class of patients towards which physicians can discriminate their inducement behaviour. The empirical test is performed estimating a simultaneous binary probit model in which the probability of c-section delivery depends on the type of hospital chosen and the latter is allowed to be endogenously determined.