Robustness Checks
Technically already the two other control groups of the served as a robustness check, however Bauernschuster & Schlotter use other techniques to argue in favour of both estimation techniques.
Robustness of the IV
To underline the robustness of the IV approach, the authors show that the parents below and above the cut-off do not vary significantly in most character traits and additionally they provide estimates for the predicted (by all characteristics except the cut-off rule) childcare attendance rate and try to show that based on observables the increase in predicted childcare attendance runs smoothly around the cut-off. The jump in the actual attendance seems to be driven by the cut-off rule shock. To further test the IV approach, the two authors apply a placebo test by regressing the model on the employment status in period \(t\ -1\) instead of period \(t\) as done in equation (2). The results of the IV estimate, displayed in Table 3 show that the coefficient is not statistically different from zero. This means that mothers whos youngest child turned three shortly after the cut-off date are not different from mothers whose youngest child turned three shortly before the cut-off date in terms of employment before the cut-off rule becomes relevant. The autors conclude that this further validates the use of the IV approach.