Woman standing in front of a restaurant with her hands on her hips wearing a face mask

What puzzling effect did COVID-19 have on the US labour market?

This blog was originally published on the World Economic Forum.

  • Research from LinkedIn suggests that most people chose to remain within their own industry during the pandemic, seeking better jobs rather than moving to new industries.
  • Although the Beveridge curve, which captures the relationship between vacancies and unemployment, has shifted out, there is no indication that the process of matching workers and jobs has permanently declined.

  • The pandemic accelerated skills-based hiring, with 45% of companies focusing efforts on hiring based on skills and competencies.

The COVID-19 pandemic triggered job loss on a scale not seen since the Great Depression — unsurprising, given the near-total shutdown of economies worldwide.

But more perplexing was the recovery. In the US, the months and years after the pandemic struck was a period marked by millions of jobs, without enough workers to fill them.

One way to see this is through the Beveridge curve, which captures the inverse relationship between job vacancies and unemployment. Normally, changes in economic activity show up as movements along  the Beveridge curve, with stronger labour demand going hand in hand with lower unemployment and higher vacancies. However, in the United States, the Beveridge curve has shifted out since the start of the pandemic, meaning many more job openings are available than previously at the same level of unemployment.

Source: Data come from the BLS, are seasonally adjusted, and span the period Jan 2001 - June 2022.

The conventional interpretation of an outward shift in the Beveridge curve is a labour market mismatch between jobs and skills — due, for example, to the reallocation from one industry to another. This was the subject of intense debate among policymakers after the 2008-2009 recession. At that time, the argument went that there wasn’t much that monetary policy could do about the disconnect and the Federal Reserve wouldn’t repair it.

Some economists are echoing those sentiments today, suggesting that “sectoral reallocation” may be at play. If true, the resulting frictions from moving costs, re-training and learning about one’s ability, among others, can make the matching of workers and jobs a slow and difficult process.

Vacancies beget vacancies

However, LinkedIn’s data offers an alternative interpretation for the trend of plentiful jobs and not enough workers. It suggests that much of the increase in reallocation occurred within industries, with workers quitting to move to new jobs in the same sector, leaving their old jobs vacant, which firms repost, and so forth.

A vacancy chain emerges as an initial increase in job vacancies at some firms triggers new vacancy postings when jobs are filled by recruiting workers from other employers. The existence of vacancy chains is therefore central to understanding the procyclical behaviour of quits and the associated outward shift in the Beveridge curve. These findings mean the magnitude of the outward shift in the Beveridge curve may be exaggerated.

We saw clearly during the pandemic a record-level rise in the share of temporary unemployment. If those who were temporarily unemployed were not searching for jobs, but instead were waiting to be recalled, then it is necessary to separate them out from those who are searching for jobs when considering the performance of the labour market.

LinkedIn’s data on applications sent by job seekers suggest that the temporary unemployed did not affect the hiring process in the same way as those who lost their jobs permanently. When the share of temporary unemployed hit 80 percent of total unemployed at the start of the pandemic, job search intensity, measured by the number of applications sent per applicant, did not increase much. Workers on temporary layoff were likely not searching and should be excluded from the stock of unemployed when examining the dynamics of the Beveridge curve shift.

Once the temporarily unemployed are removed from the equation, the outward shift in the Beveridge curve becomes much less pronounced, with the remaining gap mostly driven by the increase in quits.


Source: Data come from the BLS, are seasonally adjusted, and span the period Jan 2001 - June 2022.

Reinterpreting the curve

So, why did so many people quit their jobs? It was because they could.

Part of the increase in quits can be attributed to workers concluding that it was easy to get another job and that job switching would result in better pay or working conditions. This phenomenon is indicative of what is termed the Great Reshuffle. That is, workers weren't just sitting on the sidelines — they were opting to move into new, more appealing jobs.

Reshuffling can be between industries, but much took place within the same industry. The unusual nature of the COVID-19 shock, with some sectors practically shut down completely, raises the question of whether the cross-industry share of job reallocation dominates the surge in total reshuffle. LinkedIn’s data on job transitions provides little support to this argument and suggests that the bulk of the pandemic-induced reallocation was happening within the same industry sectors.

The restaurant industry provides a salient example of within-industry reallocation that has likely been driven by upward pressure on wages. While some restaurants permanently closed, delivery-oriented chains experienced a demand boom.

The future of the labour market

Some have suggested that the increase in the intensity of reallocation is upending the usual relationship between job vacancies and unemployment. In this view, the contemporaneous presence of growing and declining sectors would imply higher worker reallocation and more difficulty matching jobs to workers.

However, LinkedIn’s data shows that much of the increase in reallocation has taken place within industries. This coincided with an unprecedented rise in quits, partly due to COVID-19 and partly because workers believed that it was easy to get a better job. The last two factors will undoubtedly be eased as aggregate demand softens, suggesting that the outward shift in the Beveridge curve can be largely reversed.

While the pandemic may have forced some workers to consider new industries, it also accelerated other trends that will undoubtedly improve, rather than harm, the efficiency of matching jobs to workers. One trend that accelerated during the pandemic was skills-based hiring, with 45% of companies on LinkedIn currently relying on skills data to search for and identify candidates.

By focusing on skills, employers can not only increase the size of their talent pools, allowing them to pinpoint non-traditional applicants for hard-to-fill roles, but also the quality of the matches will improve. As such, there is little evidence to suggest that workers reallocation has led to a decline in the efficiency of matching workers to jobs, and now, companies are getting smart about how they recruit the best candidates.

These findings are explored in further depth in a new research paper by LinkedIn.