To gain a better understanding of the impact of the different policy strategies in Germany and the United States on the labour market, we conduct a business-cycle analysis comparing, on the one hand, the Coronavirus Recession with the Great Recession in both countries and contrasting the German experience with that of the United States on the other hand. Then, we look at the case of Germany in the Coronavirus Recession with particular interest in the relative importance of working-time instruments (overtime, regular working time, working time accounts (WTA), and STW) in safeguarding employment, before examining the role of external flexibility in the United States via temporary and permanent lay-offs during the Covid-19 pandemic.
Germany and the United States: a business cycle analysis
For a business-cycle analysis of the cyclical variations in economic activity, employment, productivity, and working hours in these two countries, we must first determine the peak and trough of the Great Recession and the latest recession in Germany and the United States. We follow the business cycle dating of the NBER’s Business Cycle Dating Committee for the United States and of the German Council of Economic Experts (GCEE) for Germany.
In Germany the economic downturn of the Great Recession started after the first quarter of 2008 (peak) and ended in the second quarter 2009 (trough). According to the NBER, in the United States it started already after the fourth quarter 2007 (peak) and also ended in the second quarter 2009 (Table 2). As for the latest downturn, the Coronavirus Recession, in both countries economic activity peaked in the fourth quarter 2019.Footnote 4 The NBER dated the trough of economic activity to the second quarter 2020. For Germany, the GCEE has not yet determined the trough of economic activity. However, the development of economic activity in Germany in 2020 and 2021, especially GDP growth, also points towards the second quarter 2020 as time of the economic trough. Therefore, in the further analysis we assume that 2020q2 is also the trough of economic activity in Germany, but present data for both countries until the end of 2021.
Given the determination of the Great Recession and the Coronavirus Recession in Germany and the United States by the GCEE and the NBER, respectively, we focus on cyclical variations in economic activity in the following business-cycle analysis. Therefore, we extract the cyclical and trend component using the Hodrick-Prescott Filter (HP-Filter). As is common practice for quarterly data, we use a HP-Filter with a smoothing parameter (lambda) equal to 1600 to detrend the quarterly time series from 1991 to 2021. It is well known that the HP-Filter, like other filter methods, suffers from an end-point problem. Since our main focus is on the slump until 2020q2 and we use six additional data points, the impact of this end-point problem is still there but of a smaller importance for the analysis of the recession periods. However, we are careful in interpreting results after 2020q2 and closer to the end of the data set. For completeness and clarity, we present results up to 2021q4, the end of our dataset.
For Germany and the United States, Fig. 3 examines the economic dynamics of the cyclical components of real GDP, employment, productivity and working time, i.e., average hours worked per employee, during the Great Recession (Germany: Panel A, and United States: Panel C) and the Coronavirus Recession (Germany: Panel B, and United States: Panel D). All figures are normalised to the respective beginning of the two economic downturns, i.e., for the Great Recession 2008q1 for Germany and 2007q4 for the United States, and for the Coronavirus Recession 2019q4 for both economies.
Due to the economic shock caused by the Covid-19 pandemic, the Coronavirus Recession was much more severe. From 2019q4 to 2020q2, cyclical real GDP contracted by 12.3% in Germany and by 11.4% in the United States as a direct consequence of the Covid-19 pandemic. In the Great Recession, the corresponding cyclical decline in output from peak to trough was 8.1 and 5.1%, respectively.
As in the Great Recession, most of the economic shock in Germany was absorbed by internal flexibility in the labour market via a temporary working-time reduction and labour hoarding in the form of a procyclical decline in labour productivity. However, this time the relative contribution of internal flexibility was even larger than in the Great Recession. From peak to trough, the cyclical reduction in the average number of hours worked per employee was twice as high as in the Great Recession (− 8.8 vs. − 3.4%). In Germany, productivity reacted much stronger in the Great Recession than in the Coronavirus Recession (− 5.6 vs. − 2.1%). Even though speed and intensity of job losses were more pronounced in the Coronavirus Recession, in both economic recessions cyclical employment continued to decline, even after the trough of the business cycle. Overall, employment declined cyclically by 0.6% from 2008q1 to 2009q2; by 2010q1 it had fallen by a further 0.7%. Thereafter, cyclical employment started to recover. In the Coronavirus Recession, cyclical employment declined by 1.4% from 2019q4 to 2020q2 and a further 0.4% by 2021q1. It then started to recover over the remaining quarters of 2021.
In contrast to the economic developments observed in Germany, in the United States external flexibility bore the brunt of adjustment in response to the economic shock as a consequence of the Covid-19 pandemic (Panel D). From peak to trough, cyclical employment decreased by 13%. However, in contrast to the cyclical development in Germany, employment then started to recover. This also contrasts the cyclical behaviour of employment in the Great Recession when it continued to decline beyond the trough quarter.
Average hours worked per employee decreased by 1.0% in the Coronavirus Recession, while labour productivity cyclically increased by 2.1% from peak to trough. This also contrasts with developments in the United States during the Great Recession, when internal flexibility from cyclical reductions in working hours per worker and changes in labour productivity together accounted for about a quarter (− 1.2%) of the labour market adjustment relative to the cyclical decline in real GDP (Panel C). In the Great Recession, from peak to trough working time per employee decreased cyclically by 1.7%, a larger decline than during the Coronavirus Recession. But in the latest contraction the speed of the working-time reduction was faster than in the Great Recession. However, the major difference in the cyclical labour-market responses between the two recessions lies in the development of cyclical labour productivity in the United States. Cyclical labour productivity behaved slightly pro- to acyclical in the Great Recession and anticyclical in the Coronavirus Recession.
Overall, this section has shown that the German and US labour market reacted quite differently during the last two recessions. In Germany, internal flexibility dominated labour market adjustment, while in the United States it was external flexibility. This finding fits with the descriptions of the policy responses outlined above.
Germany: internal flexibility
In Germany, several instruments of internal flexibility are available at the establishment level to temporarily adjust the number of hours worked per employee in response to changes in the economic environment, such as overtime, working-time accounts, temporary changes in regular working time and STW. Figure 4 therefore shows the development of cyclical working time per employee and its components regular working time, paid and unpaid overtime, STW, as well as WTA, again detrended with the HP-filter ((lambda) = 1600) if the component has a trend. Over the period from the beginning of 2005 to the end of 2020, working time and all its components follow a clear cyclical pattern. However, while all these components contributed to the safeguarding of employment during the financial crisis (Herzog-Stein et al. 2018), this is no longer the case in the Coronavirus Recession.
Short-time work (STW)
In terms of safeguarding jobs, STW has two important dimensions: the number of workers in STW and the intensity of STW, i.e., the number of reduced working hours per short-time worker due to STW. Comparing the development of STW in both recessions, two aspects stand out particularly. First, policy makers reacted fast and made the use of STW more attractive for establishments immediately after the outbreak of the Covid-19 pandemic at the end of the first quarter 2020. This had the effect of introducing STW on a uniquely large scale in both dimensions of STW. In April 2020, the month with the highest incidence of STW in the Coronavirus Recession, almost 6 million, or 17.9% of all employees subject to social security contributions were in STW. The average loss of working time for a short-time worker was nearly 50%. In employment equivalents this corresponded to 9.1% of all employees subject to social security contributions (Fig. 5).
Although the number of employees in STW declined steadily after April 2020, there were still more employees in STW in October 2020 than at the peak of the Great Recession. As a result of the second wave of the Covid-19 pandemic, the number of workers in STW rose again from November 2020 and peaked in February 2021 before declining again.
Consequently, in the Coronavirus Recession there was a rapid cyclical reduction in average working time per worker of 2.4 h already in 2020q1 alone (relative to 2019q4). This is comparable in its magnitude to the cyclical working-time reduction induced by the use of STW from peak to trough in the whole Great Recession of 3.3 h per worker—of which 3.1 h were reduced in the first two quarters of 2009 relative to the last quarter in 2008.
Second, while the immediate response in STW was already comparable to the Great Recession, at the trough of the Coronavirus Recession in the second quarter 2020, STW reduced the average working time per worker by 18.4 h compared to the peak quarter 2019q4, more than five times the working-time reduction due to the use of STW in the Great Recession. On average, STW is accounting for around 89% of the total reduction in hours worked per worker from 2019q4 to 2020q2.
In the two recessions, employees subject to social security contributions were affected differently by STW in the individual economic sections (Fig. 6). A comparison between May 2009 and April 2020, the months with the highest incidence of short-time work in both downturns, shows that this time not only was the number of short-time workers significantly higher, but in the economy as a whole STW was used more heavily (columns in Fig. 6). While more than 80% of short-time workers were employed in manufacturing during the Great Recession, it was only about 31% during the Coronavirus Recession.Footnote 5
In the Coronavirus Recession, STW is also used more intensively across all economic sections (dots in Fig. 6). In the total economy, the intensity of the use of STW in April 2020 was nearly twice as high as in May 2009. The average intensity of STW use was particularly high in the services sector, exceeding 70% in sections ‘Accommodation and food service activities’ (I), ‘Arts, entertainment and recreation’ (R), and ‘Other service activities’ (S). In the past, STW intensity of 100% was not common. In the Coronavirus Recession it was used only modestly, despite the severity of the crisis. According to Kruppe and Osiander (2020) using information on the individual STW intensity from a survey in May 2020, 24.1% of all STW-workers reported a loss in hours of 100%, but still more than half a loss in hours of only up to 50%.
In contrast to the importance of internal flexibility and especially the use of STW, external flexibility—unlike in the United States (see Sect. 3.2.1)—hardly played a role in Germany between March and April 2020. The overall change in employment, measured by the sum of employees subject to social security contributions and workers only marginal employed, was only about − 1% on average (diamonds in Fig. 6). Only in section H (Transportation & Storage) there is a substantial drop in employment of − 7.5%.
Working-time accounts (WTA)
Together with STW, they were the most important instrument of internal flexibility during the Great Recession. Like STW, the use of WTA at that time reduced the average working time per employee by 3.3 h in total or 0.7 h per quarter from peak to trough. In the Coronavirus Recession, the contribution of WTA to the temporary reduction in average hours worked per worker is this time much smaller than in the Great Recession. From peak to trough, WTA contributed 1.7 h, or on average 0.8 h per quarter, to the reduction in average hours worked per worker in the latest downturn.
At first glance, this is unexpected, as WTA became more common over time and 56% of all employees had WTA in 2016 (Ellguth et al. 2018). However, one possible explanation could be the respective economic dynamics in the boom periods before the two recessions.
In the upswing before the Great Recession, WTA were filled, providing firms with a considerable working-time-account buffer for the following downturn. In contrast, in the long boom period before the Coronavirus Recession, working time was closer to its long run trend with smaller cyclical variations. As a result, opportunities to increase the balances in the WTA were more limited than in the boom period before the Great Recession. Therefore, the working-time reductions due to WTA account only for 8% of total working-time reduction in the latest recession from 2019q4 to 2020q2.
In general, paid and unpaid cyclical overtime vary between ± 1 h per quarter over the business cycle. Unpaid overtime was most important at the beginning of the considered period (Fig. 4). After the minor economic slowdown in Germany related to the so-called Euro Crisis from 2011q3 to 2013q1, it lost its relevance for cyclical fluctuations. Interestingly, unlike unpaid overtime, the cyclical variation of paid overtime continues after the Great Recession and can also be observed in the Covid-19 pandemic.
In the Coronavirus Recession, the contributions of paid and unpaid overtime to the cyclical reduction in working time from 2019q4 to 2020q2 on a quarterly basis (− 0.3 h vs − 0.2 h per quarter) is similar to that in the Great Recession (− 0.2 h and − 0.2 h per quarter), but together accounting only for less than 5% of the total working-time reduction per worker during that time period, in contrast to nearly 20% in the Great Recession.
Regular working time
Unlike in the Great Recession, there is not really a cyclical response in regular working time to reduce working hours in the Coronavirus Recession. The cyclical component of regular working time per worker even slightly increased average working hours per worker by on average 0.3 h from 2019q4 to 2020q2. Overall, this observation might be explained by the dominance of STW, which made further adjustments to working time unnecessary.
In conclusion, although external flexibility again was of minor importance and instruments of internal flexibility played a crucial role in the safeguarding of employment in both the Great Recession and the Coronavirus Recession in Germany, a closer look at various working-time components shows marked differences between the two recessions. While in the Great Recession several instruments contributed markedly to the temporary decline in hours worked per worker, in the Coronavirus Recession STW is the instrument that has contributed by far the most to the reduction in working hours (Fig. 7).
In the Great Recession, STW and WTA contributed equally to the cyclical reduction in working time from peak to trough (− 3.3 h each). Paid and unpaid overtime and a temporary reduction in regular working hours both reduced cyclical working time by an additional two hours. In contrast, while most instruments responded as expected in the latest downturn, in absolute and in relative terms STW was by far the main driver to safeguard employment in the Coronavirus Recession (− 18.4 h). WTA was again the second most important instrument of internal flexibility used. However, its quantitative importance was smaller, reducing average working hours per employee by 1.7 h. The same is true for paid and unpaid overtime which together reduced average working hours by another 0.9 h. Reductions in regular working time do not contribute to the cyclical reduction in working time in the Coronavirus Recession. The observed dominance of STW in the attempt to safeguard employment in the Coronavirus Recession is in line with the made discretionary policy changes governing the use of STW. It is conceivable that the extended and simplified use of short-time work “crowded out” to some extent the use of other measures like e.g., WTA since already in March 2020 no negative balances on WTA were required anymore as eligibility criteria for the use of STW.
Finally, the important impact of the use of STW on unemployment respectively employment is best seen by looking at the seasonally adjusted inflow rate from employment into unemployment and the exit rate from unemployment into employment on a monthly basis (Fig. 8). From February to April 2020 the inflow rate from employment into unemployment increased from 0.5 to 0.7% and declined than quickly back to 0.5% in June, while the exit rate from unemployment into employment decreased from 8.3 to 4.7% from February to May 2020 and did not reach its pre-pandemic level until the end of 2021. This can be seen as some indication that the massive use of STW was able to prevent large and prolonged flows from existing employment into unemployment.
United States: external flexibility
Although STW programs exist in about half of the U.S. states and STW utilisation was much higher than in the past, the use of STW has overall not played a major role in the United States (Krolikowski and Weixel 2020). Here, the focus was rather on external flexibility. However, for the first time, the use of temporary lay-offs, i.e. laid-off individuals who expect to be recalled by their former employers (Gallant et al. 2020), was the prominent tool for dealing with the crisis.
While temporary unemployment has been between 0.4 and 1.2% throughout the years and even during the economic and financial crisis it played no prominent role with respect to the overall increase in unemployment, the share of workers on temporary layoffs jumped to 11.5% in April 2020, accounting for almost 80% of all unemployed persons (Fig. 9). While unemployment declines slowly during an economic recovery, the work-finding rateFootnote 6 for the temporarily laid-off unemployed is usually twice as high as for the unemployed. Accordingly, unemployment fell faster this time than in previous recoveries (Hall and Kudlyak 2022). Thus, the rate of temporary unemployment halved from April to July 2020, while the jobless unemployment rate increased by 1.2 percentage points.
External flexibility via temporary lay-offs was the main means of the labour-market adjustment in the Coronavirus Recession, but complementary to this, companies also used some measures of internal flexibility by reducing the working hours of their employees. While unlike in Germany there are no detailed information on the average number of working hours lost per worker due to the Covid-19 pandemic in the United States, an additional survey conducted by the BLS beginning in May 2020 as part of the Current Population Survey provides a good insight into the extent to which workers in non-agricultural industry were affected by the crisis. In this survey, workers were asked whether they had been unable to work due to the pandemic in the last four weeks and whether they had received any kind of payment from their employers. Unfortunately, no information was asked about the form of compensation paid or the exact number of hours lost as a consequence of the pandemic. Thus, it is also not known whether workers who were compensated for working time lost due to the pandemic received any payments via one of the short-time work programs at the state level.
In May 2020, 20% of workers reported that they were affected by some kind of loss of working time,Footnote 7 and in June and July 2020 still more than 15 respectively 10% of workers experienced some loss of working time (Fig. 10). The majority of them reported that they were not compensated by their employers. Only less than a quarter received some compensation. Therefore, the reduction respectively loss in working hours in the United States took place in a way that is quite different from the short-time allowance in Germany.
Overall, these information on unpaid as well as on compensated temporary working time losses fit together with the macroeconomic evidence for the United States presented in Sect. 3.1. As shown above there was some cyclical reduction in the average working time per employee from peak to trough of 1.0% in the Coronavirus Recession. If we take into account that the Coronavirus Recession was much shorter than the Great Recession, the average individual working-time reduction per quarter was stronger this time. Given that “job-losses have disproportionally hit the low-wage workforce” (Bateman and Ross 2021) it is also likely that the reported working-time losses were concentrated among the low-skilled. Since low-skilled workers generally have a lower hourly labour productivity, this would explain the anticyclical increase in labour productivity observed in the Coronavirus Recession (see Fig. 4D).
More details about the American way of dealing with the Coronavirus Crisis are revealed by looking closer at the change in employment and the share of workers affected by working hours lost due to the inability to work in different economic sections of the US economy (Fig. 11). Given that for economic sectors no data for temporary layoffs are available the change in employment is used instead to indicate the intensity with which employers were hit by job losses across economic sectors. As in Germany, the economic sections have been affected differently by the Covid-19 pandemic, the service sectors more than the industry.
Interestingly, a combination of layoffs and reductions in working hours dominates in all sectors of the economy. Hence, the dominance of external flexibility in the labour market adjustment in the United States does not imply that firms do not use measures of internal flexibility, too. The information from the economic sectors indicates that economic sectors that were hit hard by the Covid-19 pandemic relied on external as well as on internal flexibility in response to the Coronavirus Recession. There is a strong positive correlation between layoffs and working-time reductions with a correlation coefficient of 0.9: in economic sectors with a larger reduction in employment there is also a larger share of workers with a loss of working hours due to the inability to work. However, this positive relationship is driven by the positive correlation between employment reductions and working-time losses without compensation; there is no correlation between the magnitude of employment losses and the size of the share of workers with renumerated working-time losses. This suggests that, in contrast to the experience in Germany, in the United States the burden of labour market flexibility in the Coronavirus Recession is borne primarily by workers.
In conclusion, the United States have relied again heavily on the use of external flexibility. However, there are also major differences in its response compared to the Great Recession as a breakdown of the change in the unemployment rate in the two crises reveals (Fig. 12).
Most of the change in the unemployment rate is determined by the number of employees losing their jobs, usually without being on recall. During the Great Recession, the unemployment rate rose by 4.5 percentage points. Only a small part of 0.5 percentage points was due to temporary layoffs. Another 3.1 percentage points of the increase in the unemployment rate was due to workers losing their jobs. The proportion of job leavers among the unemployed is hardly influenced by the business cycle and lies typically in a range between 0.5 and 0.6%. Hence, its contribution to the change in unemployment over time is negligible. Almost one percentage point of the increase was due to re-entrants and new entrants into the labour market during the Great Recession.
As shown in the analysis above the labour market response during the Coronavirus Recession was extraordinary and very different to the one observed in the Great Recession. For the first time, temporary layoffs played a prominent and dominant role in the United States. The unemployment rate rose by a total of 9.4 percentage points from peak to through, of which 8.8 percentage points were due to temporary layoffs and only 0.7 percentage points to workers who lost their jobs. Interestingly, fewer employees seemed to leave their job of their own accord, and no change in labour force entry was observable.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.