Key Points:
- The pandemic brought lasting structural changes to the economy, particularly to occupations jobseekers are most interested in.
- While jobseeker-employer mismatch by occupation soared at the pandemic’s start, it did eventually return to pre-pandemic levels.
- Countries with short-time work schemes were slower to return to pre-pandemic mismatch levels than those without.
The mismatch between job seeker interest and job postings for different occupations soared during the early days of the COVID-19 pandemic. While mismatch eventually returned to pre-pandemic levels as labor supply adjusted to changes in labor demand, some compositional shifts in the mix of job postings and job seeker clicks lasted. The speed at which the economy readjusted was significantly slower in countries that had (or implemented) short-time work schemes intended to encourage employers — especially those in the most impacted industries, including manufacturing and retail — to keep their workers.
Prior to the pandemic, many countries had policies already in place meant to preserve jobs in businesses experiencing difficulty. But how well-suited were these policies for a sudden shock that left lasting changes to employment supply and demand? A new paper by researchers from the Organisation for Economic Cooperation and Development, Indeed, and George Washington University uses Indeed data to examine how employment retention policies held up to the near-overnight shuttering of in-person businesses. It compares labor reallocation across 55 occupations and 19 countries based on their employment retention policies both during and after the pandemic.
To do this, the authors use a measure of short-time work schemes — where workers are put on a reduced work schedule and receive partial compensation from their governments for work hours lost. The authors then analyze a monthly metric measuring occupational mismatch for countries with and without short-time work schemes. This metric is based on a weighted sum of the difference between the number of job postings in each sector, and the amount of interest in that sector (measured as clicks) between January 2019 and January 2023.
The pandemic brought a lasting labor market shift
The COVID-19 pandemic shook the foundations of the global economy and job markets in ways we’ve never experienced, with lasting effects. The mix of available jobs changed dramatically, virtually overnight — fewer in-person roles in food prep and retail, balanced by more healthcare jobs and more remote-eligible jobs. At the same time, job seeker interest in available jobs also shifted (fewer job seekers clicking on retail jobs, more people clicking on software jobs).
During the first three months of the pandemic (February, March & April 2020), 6% of both postings and clicks reallocated across occupations as job seekers and employers responded to the closure of in-person businesses and many offices went remote. To put into perspective how rapid that change was, the analysis found that had the speed of reallocation remained at its pre-pandemic pace, a similar 6% reallocation would have taken about two years for clicks and more than three years for postings. And many of these changes lasted through at least 2022, and likely beyond, as digital labor practices and tools stayed broadly popular and labor demand and job seeker interest both resettled into different occupations.
Manufacturing job postings, which were highly impacted early in the pandemic, ended 2022 representing a higher share of total job postings than they had prior to the pandemic in 2019. Sales and customer service jobs, which represented a relatively larger share of total job postings in 2019, ended 2022 with a smaller overall share of total postings. On the clicks side, interest in software development and education jobs remained well above pre-pandemic levels, while interest in sales and administrative assistance roles ended 2022 below pre-pandemic norms. This signals a potential change in the kinds of jobs typically attractive to low- and mid-skilled workers in particular.
These changing compositions contributed to a spike in the so-called “mismatch” between job seeker interest and job postings for different sectors in the early days of the pandemic.
The pandemic temporarily supercharged occupational mismatch
Economists use the term “mismatch” to describe when job seekers and available positions aren’t aligned. When there’s a high amount of mismatch, some sectors have lots of job seekers but few openings, while others have lots of openings but few job seekers. In the simplest example, suppose we split the economy into just two sectors: services and manufacturing. If 30% of job seekers are clicking on postings in the service sectors, but postings in that sector only make up 10% of all job postings, there would be a mismatch of 20%.
There is always some level of mismatch in the economy — available jobs never perfectly match up with job seeker desires, and vice versa. Prior to the pandemic, the level of mismatch across all 19 nations analyzed was about 26% — a little more than a quarter of clicks and postings were not fully aligned. As jobseekers shifted their interests both in response to the pandemic and to the change in job postings available, mismatch increased by nearly 4 percentage points, to almost 30%. Over the course of the pandemic, sectors that had an initial excess in job seeker interest (comparatively more clicks from job seekers), including management and retail, saw an increase in postings. And those with excess demand (comparatively more job postings) contracted, including loading and stocking, and food services, allowing mismatch to return to pre-pandemic levels by mid-2022.
But while the amount of mismatch did eventually return to pre-pandemic norms, the overall composition of postings and clicks did not.
Why some countries were slower to adjust to pre-pandemic mismatch levels
Labor markets in the 19 countries analyzed eventually readjusted and returned to pre-pandemic mismatch, but the time it took for reallocation to take place differed based on a given country’s employment retention policies. Countries that utilized short-time work schemes were significantly slower in reducing mismatch than those that did not. These schemes incentivize employers to retain workers until economic activity picks up, and are commonly used to soften unemployment during times of temporary downturns. But in the case of the COVID-19 pandemic, short-time work schemes made it harder for the economy to adjust to lasting changes in labor demand. Businesses in expanding sectors had a hard time finding workers as businesses in contracting sectors held onto them.
Interestingly, wage subsidies — another form of job retention scheme that pays a fixed subsidy for each employed worker, irrespective of whether workers are working full contractual hours or are put on a reduced work schedule — had no discernable impact on mismatch.
Why did short-time work schemes slow labor market adjustment?
It’s easy to see the logic of short-time work schemes, especially when addressing abrupt cross-country shutdowns like during the pandemic. What many could not predict was the lasting changes to the demand for labor in the months following closures. These schemes help businesses hold on to workers and workers to hold onto their jobs in hopes that the incentive to let go of workers is temporary and mismatch would return to pre-pandemic levels. But while mismatch did return, the distribution of clicks and postings did not. In countries that paid employers to hold onto employees they would have otherwise let go by putting them on reduced work schedules, workers in fields with a drop in demand held off on finding new employment, slowing down labor market adjustment. By contrast, businesses covered by wage subsidy schemes may have let go of some workers because they could obtain the subsidy without holding on to all workers.
Conclusion
Shifts in jobseeker interest away from in-person sectors, and demand for workers in sectors that initially had more interest than postings, allowed mismatch to return to pre-pandemic levels by the end of the pandemic, despite lasting changes to the composition of clicks and postings. While this is true for all countries observed, countries that incentivized employers to hold onto workers through short-time work schemes were statistically much slower to adjust to changes in labor demand than those that did not. It was impossible to predict both the onset and impacts of the COVID-19 pandemic, and there’s a chance that short-time work schemes might have preserved good labor demand and supply matches if the economic impacts of lockdowns had been shorter-lived. Instead, these shifts in the market were longer-lasting, and, in this case, these policies made it harder for job seekers to adjust their interest to high-demand sectors and close the mismatch gap.
Methodology
We measure occupational mismatch based on the difference in shares of job postings and job seeker clicks using the Duncan and Duncan dissimilarity index, a formula we’ve used in past research. Clicks allow us to focus on the forward-looking interest of job seekers rather than their past work experience. Clicks also capture the intensity of job seeker interest since a single job seeker may click on many different job postings. The difference in shares means that we’re comparing the distribution of job seekers and job postings rather than raw numbers. Zero mismatch does not mean there is an appropriate job seeker to fill every job posting. Instead, zero mismatch would mean that the relative difficulty of hiring or finding a job is the same across all sectors. By contrast, high mismatch means in some sectors it’s hard to hire, while in other sectors it’s hard to find a job.
Economy-wide mismatch is driven by large categories. The formula captures the aggregate differences in the shares of job postings and clicks by category. Furthermore, mismatch rises with more categories, which explains why mismatch is lower here, where we use 55 sector categories, compared with our past research at the normalized job title level of 6000+ categories.
Countries included in the analysis with a short-time work scheme were Austria, Belgium, Switzerland, Germany, Spain, France, Great Britain, Israel, Italy, Japan, Luxembourg, Netherlands, Sweden, and the United States. Countries included in the analysis with a wage subsidy scheme were Australia, Canada, Ireland, Poland, and the Netherlands (the Netherlands had both short time and wage subsidy schemes during the pandemic recovery period).
