Neighborhoods, networks, and delivery methods

We examine the contribution of information transmission among pregnant women to geographic variation in C-sections in Lombardy, Italy. Defining networks as pregnant women living in the same municipality, we observe that if the incidence of C-sections within the womans network is one standard deviation higher over the 12 months preceding delivery, then her probability of delivering by C-section is 0.007 percentage points (3%) higher. This result is mainly a network effect on Italian women, while it arises from both network and neighborhood effects on foreign women. Both groups respond to additional information, such as the incidence of C-section complications. The selection of pregnant women across hospitals does not uniquely explain our results, which are robust to alternative sample selections and specifications.