Cutting Fertility? The Causal Effects of Supply-Determined Cesarean Section on Fertility

Cesarean section rates have reached huge proportions in many countries around the globe, while costs and benefits of the procedure are discussed controversially. In addition to the substantial monetary costs of unnecessary surgeries, the medical literature finds associations of Cesarean section with negative health effects on the infant, but also the mother. In particular, there is evidence that first time mothers delivering by Cesarean section have lower probability to have a second child. These correlations, however, can be due to selection into procedure. Because randomized control trials are not feasible for ethical reasons, causal evidence can only be obtained with quasi-experimental methods. This paper identifies the causal effects of Cesarean section, exploiting variation in practice styles between hospitals and days of the week as an instrument.

Using the Austrian birth register, i.e. the universe of live births from 1995 to 2007, we find large geographical variation in Cesarean rates. Women in certain hospital areas exhibit strikingly high probabilities for a Cesarean section (e.g. 37%), while Cesarean rates in other hospital areas are modest (e.g. 20%). We exploit these differences in hospitals' treatment styles as a source of exogenous variation in mother's probability to deliver by Cesarean section. In particular, we compute the Cesarean rate in non-close communities within the same hospital area and use it as our instrument. To mitigate concerns that regions with high Cesarean rates are special in any unobserved characteristic and account for time-invariant region-specific unobservables, we estimate additional specifications with geographical fixed effects. Moreover, we control for several important socioeconomic and health-related characteristics of infant and mother.

Our preliminary results suggest that the probability of having another child is reduced after a Cesarean section. The probability is decreased by 9 percentage points two years after birth, and the effect does not subsequently diminish. The causal effect after seven years indicates that a mother is 13 percentage points less likely to give birth to another child. Our results are robust to different sample selections, the use of hospital fixed effects and the use of two alternative instruments that exploit i) the peak in the probability of delivery by Cesarean section observed on Friday and ii) differences in the treatment style of the chief physician.