Earlier this year we released a report (see COVID-19: Labour implications dated 30 April 2020) setting out our forecast for job losses in 2020. Our main takeaway at the time was that 1.6 million jobs were going to be shed this year and it would have taken over three years for jobs to reach their pre-crisis peak. Since then, more employment statistics have been released and our forecasts have changed to reflect this. We now expect around 500 000 jobs to be shed this year and for jobs to reach their pre-crisis peak in the second half of 2023.
The large difference between our current and initial forecast is the result of several factors. First, our employment model is largely based on our view of GDP growth and because GDP declined significantly (-51% q-o-q saar) in the second quarter, this exaggerated the extent of job losses in our model. It should be noted that the relationship between a percentage drop in GDP and a percentage drop in employment was never one to one, but because of the outsized influence of GDP in our model, expected job losses were greater than actual job losses in the second quarter.
Second, the current labour environment is such that many employees maintained a ‘job attachment’ and were thus considered employed. This could mean that an employee is temporarily absent from work because of the pandemic but gets paid partially or in full because the employer deems the impact of COVID-19 on their business as a very temporary supply disruption that will be reversed soon and so does not let go of staff permanently. The delineation between permanent job losses and temporary job losses is unclear and so influences forecasts.
Third, there were many anomalies in the official statistics that added complexity to the forecasting process. For example, COVID-19 related data collection challenges were such that the latest (Q2) statistics from the Quarterly Labour Force Survey (QLFS) cannot be compared to previous quarters. Our initial forecast was based on the historical series and did not consider the current break in the series. The QLFS also reported that the unemployment rate in the second quarter dropped to its lowest level since the start of the QLFS, primarily because of a shrinkage in the labour force. Not only was this outcome unexpected, but it makes future forecasting very challenging. This is because insight into when households will return to the labour force in the coming months will be required to make accurate forecasts (for more detail see our report Labour Force Survey dated 29 September 2020) and this insight can only really be obtained by conducting a household survey.