Fig. 15

Cumulative impulse responses of infection growth I to policy P for two alternative models. The impulse responses are estimated using the recursively ordered VAR(4) in (4) with \(k=3\) and \(P=\Delta\) \({\text {KSI}}^+\), \(B=\Delta \ln {\text {NTRX}}\), and \(I = \ln R_e\) with canton and time fixed effects. The top panel shows the cumulative IRs of Infection Growth I to Policy P of the baseline model (blue) and the alternative specification (red). The bottom panel shows the difference between the baseline and alternative IRs. The alternative models are characterized by: a two additional control variables concerning vaccination, namely the (lagged) change in the number of fully as well as partially vaccinated residents (left), and b a shorter sample length (right). Both variants do not include the interaction terms on vaccination progress. The data span from September 28, 2020, until April 18, 2021 for model (a), and until January 17, 2021 for model (b). The shaded areas represent 95% confidence intervals based on a wild bootstrap procedure with 5000 repetitions. The horizontal axes depict the time horizon in weeks. The vertical axes are shown in units of \(\ln {\text {NINF}}\)