The labour market policy response to COVID-19 must leverage the power of age

Shigeru Fujita, Giuseppe Moscarini, Fabien Postel-Vinay 15 May 2020

a

A

In our previous column (Fujita et al. 2020) we argued that saving aggregate matching capital is a priority during the current COVID-19 crisis. But we emphasised that these measures should also guide desirable reallocation of employment from impacted to essential sectors, some of which also present health risks to workers. While the news lament the astronomical number of newly unemployed workers, these still amount to just a few months of churning at normal pace. Reallocation is particularly pronounced in the US, where in an average month of 2019, close to four million employed workers switched jobs, as we document in Fujita et al. (2019). Even more people left the labour force, 1.7 million lost their jobs and became unemployed, and about as many made the opposite transition (Bureau of Labor Statistics). In Europe, transitions are notoriously less frequent, but still the order of magnitude of the unemployment stock every quarter. Government policies implemented in the last two months in Europe and the US, mostly as blanket insurance to workers and businesses, are contributing toward the goal of preserving matching capital, not to reallocation. We propose practical ways to improve upon these emergency measures and their next version (such as, for example, The Paycheck Security Act currently sponsored by US Senators Sanders, Warner, Jones, and Blumenthal), so that they address also reallocation.

The following empirical observations are key: COVID-19 health risk, matching capital, and worker reallocation all have distinct, albeit different, age profiles. Our recent work, referenced above, addresses the latter.

For health risk by age, a rough but useful statistic is the number of deaths per 100,000 residents. The US (CDC), UK (ONS) and Italy (ISS) report minuscule death rates below age 35, rising strongly after age 45. By absorbing valuable hospital resources for weeks on end, older workers also exert a negative externality on the multitude of people who need health care for other reasons. Therefore, protecting older workers from COVID-19 is a public health priority.

While safer health-wise, younger workers are always much more likely either to lose their job and become unemployed, or to switch jobs. Turnover is strongly declining until age 35, and then stabilises (see Rubinstein and Weiss 2006). Job-to-job transitions are an order of magnitude more likely among young workers. Job shopping occurs early and explains about half of wage growth in the first ten to 15 years of a worker’s career (Bagger et al. 2014). Finally, age correlates positively with tenure on the job, which in turn correlates positively with earnings (see Altonji and Williams 2005). Whether the latter is due to the accumulation of firm-specific human capital on the job, or to selection of good surviving matches, is a long- standing research question. In either interpretation, however, longer-tenured workers have more firm-specific human capital, so we should strive to protect it. The fuel for employment reallocation and for the build-up of ‘matching capital’ are young workers.

Under a third interpretation, unrelated to productivity, firms backload compensation and reward seniority to motivate new hires to work hard and to retain them (Burdett and Coles 2003). But the empirical evidence in Fujita and Moscarini (2017) shows that firms are more likely to recall previously long-tenured workers after a jobless spell, and recalled workers are more likely to remain at the company thereafter than new hires. Explaining these phenomena as part of a grand incentive scheme would require a formidable amount of commitment by the employer and should prevent precisely the observed high turnover of less senior employees. Therefore, we are confident that tenure and pay correlate with firm-specific human capital and productivity, and society should protect older job matches first.

We conclude that age, better than tenure alone, predicts the social value of a match. Even more accurate is work experience, such as pension contribution history plus years of post-secondary education. We should aim to preserve the jobs of more experienced workers and let the less experienced workers fill the gaps that open.

Other factors that correlate with age, such as marital status, job characteristics, and cyclical participation to employment, may account for the observed high labour market mobility of young workers. In Fujita et al. (2019), we use monthly Current Population Survey data from 1995-2018, and regress a job-to-job move indicator on a rich set of demographics, including age classes, month dummies for seasonality, a labour market business cycle indicator, and dummies for ‘source’ sector and occupation broadly classified. While most of our work addresses the (huge) impact of the Respondent Identification Policy introduced in 2007 on non-response rates, we find that some confounding factors partly explain the age profile of labour market mobility. For example, young people tend to switch jobs more often because they are less tied to marital relationships. But the strong negative and convex age pattern is confirmed.

The current crisis requires a special type of reallocation. Following Kaplan et al. (2020), we classify sectors as either S(ocial), if demand depends on risky social interaction, such as live entertainment, or C (regular Consumption, e.g. agriculture) otherwise. Three-digit occupations are classified as either Essential (E) or non-Essential, as indicated by each local authority, and further as either Flexible (F) or Rigid (R), depending on the feasibility of telecommuting (Dingel and Neiman 2020). All Essential occupations are, by inspection, Rigid. So, we have six groups: E-S (e.g. police), E-C (construction), R-S (some retail), R-C (engineers), F-S (teachers) and F-C (lawyers). See Torrejón Pérez et al. (2020) for a similar classification.

This classification takes into account the social aspect of an economic activity on the demand side. On the supply side, workers face health risks too. For example, meat processing workers, a R-C occupation, operate in close physical proximity. Mongey et al. (2020) offer a coarser classification along these lines - hazardous or not - using the American Time Use Survey (ATUS). Many essential occupations and sectors are hazardous, but some (e.g. agriculture) are not. So, we have in total 12 groups. Just like local authorities, we ignore commuting time, which can also be measured with ATUS.

Policy must facilitate reallocation of employment, especially young workers, to Essential occupations, and to R-S-H occupations that are only temporarily affected (many retail workers). The natural source are workers in Non-Essential R-S-H occupations (leisure and hospitality) which cannot be performed remotely and are hazardous to both workers and customers. F-S occupations (teachers), may be more subject to temporary disruptions during lockdowns, and job protection there is desirable. R-C (ground maintenance) and F-C (IT specialists) occupations should receive no special support, unless they are R-C-H (food retail), which should be encouraged to hire young workers.

While in principle reallocation is desirable, it will be easier along paths of least resistance. If the reallocation between any pair of sectors or occupations is desirable, but very low in normal times, the distance in skill requirements might be too large, and thus policies might be less effective in that specific direction.

Three tools can direct employment reallocation: European-style furlough subsidies to keep middle age workers (35-44, low mobility, high human capital, long horizon to use it, moderate health risk) on payroll during mandatory but temporary lockdowns; wage subsidies to younger workers (below 35, high mobility, low health risk) in the socially desirable, but riskier, activities; and Unemployment Insurance add-ons for older workers (45 and up, low mobility, high human capital, shorter horizon to use it, high health risk) specialised in activities that are persistently affected. These tools should be deployed across sectors and occupations depending on where these fall in our classification. Older workers can still work regularly in many flexible and non-hazardous occupations. Supporting the incomes of young workers can only facilitate their ability to live independently and reduce within-family transmission.

A critical issue is implementation. The first UI intervention in the US encountered enormous bottlenecks at the state level. European countries already have systems in place to subsidise furloughs and wages. Age discrimination laws can be by-passed with fiscal tools. For example, the US Treasury can selectively forgive payments of Payroll and Medicare taxes and the withdrawal of Federal income taxes, which the payroll processing company can divert directly to the worker pay check, whether on furlough or at work. Payroll taxes are regressive, so their holiday also attains a redistribution goal.

We leave the specific implementation to local authorities. In this column, we lay out the general principle, along with specific definitions of occupations and age classes. The hardest determination is whether an S(ocially affected) sector suffered a temporary or persistent decline. Are people going back to dine in restaurants and fly around, until the health crisis is resolved? Current estimates put that goalpost at 8-24 months from now. Keeping on life support, through furlough subsidies, sectors and occupations that are likely to suffer demand declines for many quarters is not socially desirable.

References

Altonji, J and N Williams (2005), "Do Wages Rise with Job Seniority? A Reassessment", Industrial and Labor Relations Review 58(3): 370-397.

Bagger, J, F Fontaine, F Postel-Vinay and J-M Robin (2014), “Tenure, Experience, Human Capital, and Wages: A Tractable Equilibrium Search Model of Wage Dynamics”, American Economic Review, 104(6): 1551-96.

Burdett, K and M Coles (2003), “Equilibrium Wage‐Tenure Contracts”, Econometrica 71(5): 1377-1404.

Bick, A and A Blandin (2020), “Real Time Labor Market Estimates During the 2020 Coronavirus Outbreak”, Manuscript, Arizona State University.

Coibion, O, Y Gorodnichenko and M Weber (2020), “Labour markets during the Covid-19 crisis: A preliminary view”, VoxEU.org, 14 April.

Dingel, J and B Neiman (2020), “How Many Jobs Can Be Done At Home?”, University of Chicago Becker Friedman Institute White Paper.

Fujita, S and G Moscarini (2017), “Recall and Unemployment”, American Economic Review 102(7): 3875-3916.

Fujita, S, G Moscarini, and F Postel-Vinay (2019), “Measuring Employer-to-Employer Reallocation”, manuscript, Yale University and University College London.

Fujita S, G Moscarini, and F Postel-Vinay (2020), “The labour market policy response to COVID-19 must save aggregate matching capital”, VoxEU.org, 30 March.

Kaplan, G, B Moll and G Violante (2020), “Pandemics According to HANK”.

Mongey, S, L Pilossoph and A Weinberg (2020), “Which workers bear the burden of social distancing policies?”, University of Chicago Becker Friedman Institute Working Paper No. 2020-51.

Rubinstein, Y and Y Weiss (2006), “Post Schooling Wage Growth: Investment, Search and Learning”, in Handbook of the Economics of Education 1: 1-67.

Torrejón Pérez, S, M Fana, I González-Vázquez and E Fernández-Macías (2020), “The asymmetric impact of COVID-19 confinement measures on EU labour markets”, VoxEU.org, 9 May.

a

A

Topics:  Covid-19 Labour markets

Tags:  coronavirus, COVID-19, labour market, reallocation, matching capital

Economic Advisor and Economist, Federal Reserve Bank of Philadelphia

Professor of Economics, Yale University

Professor of Economics, University College London

Events

CEPR Policy Research