This paper estimates how workers retained through short-time work (STW) during the COVID-19 pandemic would have fared had STW not been available.
Using Austrian matched employer-employee administrative data, we compare workers who became unemployed during the initial nationwide lockdown with observationally similar workers who entered STW instead. To strengthen the credibility of the comparison, we focus on jobs plausibly at risk of dismissal, exclude likely non-crisis unemployment using a machine-learning prediction model trained on pre-crisis labour market histories, and exploit quasi-experimental variation in firms' early adjustment strategies while conditioning on rich worker-, firm- and shock-related characteristics.
Although many unemployment entrants returned quickly to employment, unemployment had persistent adverse effects relative to STW. Three years later, unemployment entrants were less likely to be employed, much less likely to remain with their pre-crisis employer, and accumulated substantially lower labour income than comparable STW participants.
These findings suggest that STW can substantially reduce medium-run employment and earnings losses following sudden and unexpected demand shocks. They support the use of well-targeted job-retention schemes in response to future large-scale economic shocks.
The paper provides novel individual-level counterfactual evidence on unemployment relative to job retention through STW during the COVID-19 pandemic. By focussing on workers whose jobs were plausibly at risk of dismissal and comparing unemployment entrants with observationally similar STW participants rather than workers who remained continuously employed without STW, it provides a more credible assessment of the worker-level effects of STW.
