Database description: COVID Monitor (COVID)
Table of contents
Concepts and definitions
The main concept for the indicators presented in this database is actual weekly hours worked. The concept of hours actually worked within the System of National Accounts (SNA) production boundary relates to the time that persons in employment spend directly on, and in relation to, productive activities; down time; and resting time during a specified time reference period. It excludes time not worked during activities such as leave (annual leave, public holidays, sick leave, parental leave, etc.), commuting time, and time spent in certain educational activities.
The percentage of hours lost due to the COVID-19 crisis is compared to the baseline (the latest pre-crisis quarter, i.e., the 4th quarter of 2019, seasonally adjusted) and adjusted for population aged 15-64. The figures reported should not be interpreted as a quarterly or an inter-annual growth rate.
Full-time equivalent employment losses are also provided. This indicator is constructed by dividing the number of weekly hours lost due to COVID-19 and dividing them by 40 or 48. Hence, they provide an illustration of the magnitude in hours lost, by expressing them in full-time jobs.
An additional two indicators are available: Total weekly hours worked by employed persons and weekly hours worked divided by population 15-64. These time series start in 2005, the estimates combine nowcast results for 2020 and onwards with historical time series data on hours worked and population from ILOSTAT.
The number of working hours lost is estimated by making use of a “nowcasting” model. This is a data-driven statistical prediction model that provides a real-time measure of the state of the labour market, drawing on real-time economic and labour market data. In other words, no scenario is specifically defined for the unfolding of the crisis; rather, the information embedded in the real-time data implicitly defines such a scenario. By using data that are available almost in real time, it allows us to predict hours worked that are published with substantial delay or simply not available.
The data in the nowcasting model include a variety of indicators of economic activity and of the evolution of the labour market, such as:
- labour force survey data
- administrative data on the labour market, such as registered unemployment
- up-to-date mobile phone data from Google Mobility Reports
- Oxford’s COVID‑19 Government Response Stringency Index
- data on the incidence of COVID-19
Drawing on available real-time data, the model estimates the historical statistical relationship between these indicators and hours worked per person aged 15–64, and uses the resulting coefficients to predict how hours worked adjusted for population aged 15–64 change in response to the most recent observed values of the nowcasting indicators. Multiple candidate relationships were evaluated on the basis of their prediction accuracy and performance around turning points to construct a weighted average nowcast.
For countries for which high frequency data on economic activity were available, but either data on the target variable itself were not available or the above methodology did not work well, the estimated coefficients and data from the panel of countries were used to produce an estimate.
An indirect approach is applied for the remaining countries: this involves extrapolating the change in hours adjusted for population aged 15–64 from countries with direct nowcasts. The basis for this extrapolation is the observed mobility decline from the Google Community Mobility Reports and the Oxford Stringency Index, since countries with comparable drops in mobility and similar stringent restrictions are likely to experience a similar decline in hours worked adjusted for population aged 15–64. From the Google Community Mobility Reports, an average of the workplace and “retail and recreation” indices was used. The stringency and mobility indices were combined into a single variable using principal component analysis. Estimates from the first quarter of 2022 and onwards incorporate other high-frequency variables, such as quarterly GDP growth projections.
Additionally, for countries without data on restrictions, mobility data, if available, and up-to-date data on the incidence of COVID-19 were used to extrapolate the impact on hours worked adjusted for population aged 15–64. Because of countries’ different practices in counting cases of COVID-19 infection, the more homogenous concept of deceased patients was used as a proxy of the extent of the pandemic. The variable was computed at an equivalent monthly frequency, but the data were updated daily based on the Our World in Data online repository.
Finally, for a small number of countries with no readily available data at the time of estimation, the regional average was used to impute the target variable.
See the Annex of the latest ILO Monitor: COVID-19 and the world of work for information and statistical approach used to estimate the target variable for each country.
Interpretation and uses
Actual hours of work remain the most comprehensive and internationally comparable indicator of labour market activity. Because considerable differences exist at the country level in the composition of changes in working hours due to employment changes (and hence unemployment and inactivity) or adjustment of work-week hours, focusing on the traditional headline indicators such as the unemployment rate alone would result in a very incomplete picture. Population adjustment is also necessary for comprehensiveness and international comparability. Average global population growth during the last decade was approximately 1 per cent annually, with wide variation among countries. To properly capture work activity, changes in working hours need to account for this change to ensure that the level increase in population is not driving growth in hours worked (for the same reason, employment is often adjusted for population aged 15–64, using the employment-to-population ratio indicator). The ILO nowcasting model uses population aged 15–64 to adjust hours worked.
Because of the exceptional situation, including the scarcity of relevant data, the estimates are subject to a substantial amount of uncertainty. The unprecedented labour market shock created by the COVID-19 pandemic and the subsequent recovery are difficult to assess by benchmarking against historical data. Furthermore, at the time of estimation, consistent time series of readily available and timely high-frequency indicators, including labour force survey data, remained scarce.
These limitations result in a high overall degree of uncertainty. For these reasons, the estimates are being regularly updated and revised.