Has COVID-19 distorted international comparability of unemployment rates?
By Benoît Arnaud (Benoit.ARNAUD@oecd.org), Statistics and Data Directorate (OECD)
COVID-19 has had a profound impact on economic activity across the globe, with GDP falling precipitously almost everywhere where stringent confinement measures have been implemented. However, official statistics on unemployment paint a more differentiated picture. In some countries, unemployment rates increased when lockdown measures were imposed but in others they fell. Differences in the immediate impact of the crisis on unemployment rates across countries have also translated into differences in the evolution of rates as the impact of the crisis has unfolded.
In the United States, for example, unemployment rates spiked significantly at the height of lockdown measures in April and have gradually, and significantly, fallen since then, whilst in Europe, rates increased steadily through to July, with a marginal decline appearing only in recent months (Figure 1).
Figure 1: Official unemployment statistics point to diverging trends across countries
Official unemployment rate, seasonally adjusted
The casual observer might conclude from this that the employment situation has been improving in the US but deteriorating in Europe, and that differences in underlying economic structures and conditions – for example, resilience of firms, the regulatory environment, government support packages, the strength of the economy, etc. – can explain the divergence. But is there another explanation? Could differences in measurement, at least in part, be a factor?
Furloughed workers: What do the international guidelines say?
Broad comparability of unemployment data across OECD countries is achieved through adherence to International Guidelines from the International Conference of Labour Statisticians (ICLS). One aspect of those guidelines, of particular relevance to current official estimates of unemployment, concerns the treatment of persons on temporary lay-off or furloughed workers, i.e, those ‘employed’ persons who, in their present job, were ‘not at work’ due to economic reasons for a short duration but maintained a job attachment during their absence (ILO, 2013 and 2020).
The guidelines treat furloughed workers as being employed when:
- the expected total duration of the absence is up to three months (which can be more than three months, if the return to employment in the same economic unit is guaranteed)
- workers continue to receive remuneration from their employer, including partial pay, even if they also receive support from other sources, including government schemes.
In turn, furloughed workers not satisfying the criteria above are classified as unemployed, if they are actively looking for work and are available for work, or, otherwise, as outside of the labour force. i.e. neither employed nor unemployed.
Furloughed workers: What do countries actually do?
In practice, departures from these guidelines in national practices do exist. In the United States for example, persons on temporary lay-off are classified as ‘unemployed’ if they expect to be recalled to their job within six months. Indeed, for the latest US figures “people who were effectively laid off due to pandemic-related closures were counted among the unemployed on temporary lay-off” without further testing for their return to their previous job (BLS, 2020). In Canada, persons on temporary lay-off are also classified as ‘unemployed’ even if they have a date of return or an indication that they will be recalled by their employers. In Europe, on the other hand, only those furloughed workers that don’t have an assurance to return to work within three months and receive less than half of their salary are treated as unemployed, if they are also available to start work in the next two weeks (and are ‘actively looking’ – and those not actively looking, even if available, are excluded from the labour force entirely and are classified as ‘inactive’).
Ordinarily, these differences have only a marginal impact on international comparability and, indeed, on measures of unemployment and employment, as furloughed workers typically account for a relatively insignificant part of the population. However, lockdown measures imposed by countries to contain the spread of COVID-19 have significantly increased their numbers.
Available to work?
This is not the only area of unemployment statistics where COVID-19 has created measurement and interpretation problems. ICLS guidelines define the unemployed as “all those of working age who were not in employment, carried out activities to seek employment during a specified recent period and were currently available to take up employment given a job opportunity”.
However, when severe lockdown measures are in place it may be impossible to satisfy the ‘available to work’ or ‘employment seeking’ criteria, which may, in turn, create counter-intuitive movements in unemployment statistics: in some European economies for example (Figure 2), unemployment rates fell in the second quarter of 2020.
Figure 2: Official unemployment statistics in some countries showed surprising falls at the height of the crisis 1 , 2 , 3
The flip-side of these falls is an increase in persons outside of the labour force, and persons not in the labour force who didn’t seek work in the previous four weeks but who wished and were available to work, i.e. persons marginally attached to the labour force (Figure 3).
Figure 3: Persons outside the labour force increased
Persons outside the labour force. Persons aged 15 and over, percentage of population, seasonally adjusted
Can we adjust for COVID-19 distortions to improve international comparability?
Prior to the outbreak of the COVID-19 crisis, furloughed workers represented 0.5% of the labour force in the United States and less than 0.2% in the European Union. 4 In the second quarter of 2020, immediately after confinement measures entered into force in many countries, these shares jumped to 9.3% in the United States and 9.6% in the EU27. However, as described above, the treatment of furloughed workers in the United States differs significantly from that in the EU, resulting in the diverging trends of Figure 1. Excluding temporary lay-offs from the United States unemployment statistics (and instead treating them as they generally are in the EU) reveals very similar upward trends in unemployment rates (Figure 4). On this basis the unemployment rate was 1.9 percentage points higher in the United States in October compared to February, while the increase over the same period in the European Union was 1.1 percentage points.
Figure 4: Adjusting for furloughed workers reveals similar, upward, trends in unemployment in the US and Europe
Unemployment rate excluding temporary lay-offs, seasonally adjusted
Source: OECD estimates based on US Current Population Survey; Eurostat Monthly Unemployment Rate.
Further adjustments for the treatment of persons marginally attached to the labour force (i.e including them in the labour force and as unemployed) reinforces the pattern of upwards movements in underlying unemployment rates. In the EU for example, (as a share of the labour force plus the marginally attached), this extended measure of unemployment was 1.4 percentage points higher in the second quarter of 2020, compared to the first quarter, whilst in the US, it was 1.2 percentage points higher (Figure 5).
Figure 5: Broader measures of unemployment point to a stronger crisis impact on the labour market
Unemployed (excluding temporary lay-offs) and marignally attached, as a share of the labour force plus marginally attached , seasonally adjusted
Note : In this chart, European Union total includes Austria, Belgium, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Poland, Portugal, the Slovak Republic, Slovenia, Spain and Sweden.
Source: OECD estimates based on US Current Population Survey; EU Labour Force Survey.
The broader labour underutilisation rate 5 is compiled by combining the above information with another component of the population: the underemployed. This population, which consists of full-time workers working less than usual during the survey reference week for economic reasons and part-time workers who wanted to but could not find full-time work, increased significantly between the first and the second quarter of 2020, leading to increases in labour underutilisation rates in the European Union and the United States of, respectively, 8.7 and 13.6 percentage points (Figure 6).
Figure 6: Labour underutilisation rates have grown significantly
Labour Underutilisation Rate, as a percentage of labour force, seasonally adjusted
The significant impact of COVID-19 on components of the labour market that are, in normal times, relatively insignificant, calls for caution in interpreting current estimates of official unemployment rates and, in particular, cross country comparisons.
To better understand the evolution of the labour market, greater awareness is needed of current measurement differences such as those highlighted above, but also on other existing complementary indicators that provide a fuller picture of the labour market. Many of these are produced by official statistics agencies but are only rarely used in analyses. These include:
– Potential labour force, which includes persons available to work but not seeking and persons seeking work but not immediately available.
– The extended labour force, 6 which is the sum of the labour force and those seeking work but not immediately available and those available to work but not seeking.
– Total hours actually worked
– Full-time/Part-time employment
- BLS (2020), Frequently asked questions: The impact of the coronavirus (COVID-19) pandemic on The Employment Situation for March 2020, U.S. Bureau of Labor Statistics, March 2020, Washington DC. https://www.bls.gov/cps/employment-situation-covid19-faq-march-2020.pdf.
- ILO (2013), Resolution concerning statistics of work, employment and labour underutilization, 19th International Conference of Labour Statisticians (ICLS), Geneva. https://www.ilo.org/wcmsp5/groups/public/—dgreports/—stat/documents/normativeinstrument/wcms_230304.pdf.
- ILO (2020), COVID-19: Guidance for labour statistics data collection, International Labor Organisation (ILO), Geneva. https://www.ilo.org/wcmsp5/groups/public/—dgreports/—stat/documents/publication/wcms_741145.pdf.
- OECD dashboard of household statistics. http://www.oecd.org/sdd/na/household-dashboard.htm.
- OECD Short-Term Labour Market Statistics. https://stats.oecd.org/Index.aspx?DataSetCode=STLABOUR.
- Statistics Flash on April 2020 Employment and Unemployment released on 3 June 2020.↩
- Considering the population aged 20-64.↩
- The number of unemployed, marginally attached and underemployed, can be aggregated into a single measure, expressed as a ratio of the labour force, to form the labour underutilisation rate, as published in the OECD Household Dashboard (http://www.oecd.org/sdd/na/household-dashboard.htm).↩
- Expressed as a percentage of the extended labour force, the sum of the potential labour force with the number of unemployed and the number of underemployed part-time workers, forms the labour market slack, as presented in the Eurostat news release https://ec.europa.eu/eurostat/documents/2995521/11410470/3-08102020-AP-EN.pdf/074d0df8-8784-68aa-ac02-21584b702826.↩