Home based work, supply chain disruptions, and the (possibly) unanticipated drivers of pandemic job losses

It is tough to make predictions, especially about the future -- Lawrence Peter "Yogi" Berra

In the early days of the COVID-19 pandemic, efforts were made to predict the level and distribution of economic losses that it would bring. As massive shutdowns began, some predicted that the risk of job loss depended on factors such as the amenability of various occupations to work and serve customers remotely without the need to interact face-to-face. Early forecasters also factored in contractions in aggregate demand and downturns in contact-intensive sectors like travel and entertainment (e.g., BLS 2021 for the US and World Bank for Africa).

As the crisis has worn on, however, other, less predictable factors have begun to emerge, and the earlier emphasized predictors have receded in importance. Disruptions to supply chains — an issue currently making headlines  — have not only caused problems for rich country consumers, but as a new study shows, have led to an important decline in the demand for workers in middle income economies.

The study identifies the COVID-era drivers of formal sector job losses in two middle-income economies — Jordan and Georgia — and contains findings that may surprise observers.

Although declining labor supply has become a big part of the story in richer economies, in middle-income economies like Jordan and Georgia, shocks to labor demand have been the predominant cause of job losses.  Among all the possible sources of these shocks, COVID-related closures, falling export demand, and supply chain disruptions are the most significant determinants of job losses. Infection risk to customers also played an important role in Georgia, with its older population and higher per capita income (in PPP terms) than in Jordan. The results suggest that firms have not been able to replace lost sales from closures and that for many, declines in final consumer demand and supply chain disruptions have compounded both sales and job losses.

The second possibly surprising finding is the lack of evidence that firms with a higher share of employees working from home were better able to preserve jobs during the COVID-19 pandemic.  Employees of a firm work together to produce and sell their products, so when a firm’s sales suffer, all its workers are at risk, whether or not they can operate remotely. Still, jobs in sectors requiring higher levels of education tended to be at lower risk; and therefore, because women holding formal sector jobs in both countries tended to have more education than their male counterparts, women lost a lower share of formal jobs than men.

Finally, perhaps surprising to some is the similar magnitude of impact across two different economies of a change in sales on jobs. In both countries, for each week of closure, approximately 1.2 percent of permanent formal jobs were lost, and for each percentage point of lost sales, firms’ workforce levels fell on average 0.4 percent.

Still, each country responded differently to the crisis, and although the initial level of job losses was similar, Georgia’s formal private sector job levels rebounded, along with sales, in subsequent months, whereas in Jordan, they did not (first figure below). Georgia’s exports and shorter COVID closures contributed to this rebound, partially offsetting the effects of infection risk; and supply chain disruptions had a lesser impact than in Jordan (second figure).

The paper not only helps to understand what caused the levels and distribution of formal private sector job losses due to COVID-19, but also to predict the level and distribution (by education, gender & sector) at least a few months ahead.  Evidence on causal impact can also inform prediction models and, especially, policies in different economic circumstances in ways that mere correlations cannot.

Our study also brings to light a blind spot in our understanding of labor market dynamics in developing countries, not only in periods of crisis, but also in “normal” times. We know little about within-firm occupational structures and skill complementarities or how firms set the mix of skills demanded as market and technology conditions change. More detailed firm level data on these questions would not only improve predictions of the effects of future crises on jobs but would also provide a better basis for assessing labor market policies and the skills demands of the future.