Description:
We use seasonal ARIMA methods to study the imposition and removal of national uniform social distancing restrictions in response to Covid-19 in New Zealand for six crime types in six cities. We then use the estimated models to forecast counterfactual crime trajectories. Novel elements include cleanly defined lockdown periods, two distinct lockdowns with meaningful gaps between them, and sizeable periods after each one to allow for dynamics. We find that social restrictions initially lower offending, subsequent lockdowns have smaller impacts on offending, "bounce back" occurs in criminal offending after their removal, and bounce back is faster from subsequent lockdowns.