Das Dokument ist öffentlich zugänglich im Rahmen des deutschen Urheberrechts.

We propose a reweighting-estimation-transformation (RWET) approach to estimate the impacts of COVID-19 on job postings in Australia. Contrary to the commonly used aggregation-based method on counting data, our approach can be used in a relatively 'thin' market, such as Australia. In a thin market, the number of job postings is relatively small, and the share of empty cells increases substantially when aggregating the data into finer categories. Using Australian job postings collected by Burning Glass Technologies and the RWET approach, our empirical evidence shows that the overall labour demand in Australia as of July 2020 is slowly recovering from its lowest 45 per cent dip at the beginning of May. Our results also suggest that the impacts of the pandemic are relatively evenly distributed across skill levels, but vary substantially across states, industries and occupations. Our findings of the dynamics on the demand side of the labour market suggest that skill-targeted policies might not be as effective as policies targeted at the state and industry levels to facilitate economic recovery.