“…In the benchmark specification, the time window around every reform is defined to be six months, although we perform a number of sensitivity checks to test the robustness of the results to this assumption. 22 The discontinuity that model (1) exploits arises from the 20 Given time is the running variable with a treatment date as the threshold, our design reflects the regression discontinuity in time approach described by Hausman and Rapson (2018). Notice, however, that our application does not lack cross-sectional variation (as we have the universe of women having a child around the cutoff date) and does not ignore the time-series properties of the data (as we analyze both short-and long-run effects).…”