OECD Employment Outlook 2021

Job retention (JR) schemes seek to preserve jobs at firms experiencing a temporary decline in business activity by reducing their labour costs and supporting the incomes of workers whose hours are cut back. While there are important differences in their design and implementation across countries, a crucial aspect of all JR schemes is that employees’ contracts remain in force even if their work is fully suspended (OECD, 2020[1]).

JR schemes have been the main instrument used to contain the fallout of the COVID‑19 crisis on jobs in most OECD countries. Their use has been unprecedented, with take‑up as a share of dependent employment in May 2020 about ten times as high as during the peak of the global financial crisis (OECD, 2020[1]). By investing in JR schemes, governments sought to protect firms and workers against the costs of government-imposed restrictions and broader social distancing measures and contain the impact of the health crisis on the economy and society at large. They did so while faced with great uncertainty about the duration and depth of an unprecedented, sudden and global crisis that threatened the lives and livelihoods of many of their citizens. The purpose of JR schemes during the COVID‑19 crisis was therefore considerably broader than during the global financial crisis when they were used as a cost-effective tool to preserve jobs that were temporarily at risk of being destroyed (Boeri and Bruecker, 2011[2]; Cahuc and Carcillo, 2011[3]; Hijzen and Venn, 2011[4]).

The aim of this chapter is to provide a preliminary assessment of the role of JR schemes during the COVID‑19 crisis. The central question that the chapter attempts to address is how governments can promote job retention on the one hand without deterring workers from moving from declining firms and sectors into ones that are expanding. As such, it touches on important questions related to the cost-effectiveness of JR schemes as well as their implications for the strength of the economic recovery. How many jobs have been saved thanks to JR schemes and how many would have been maintained even in the absence of public support? To what extent did JR schemes slow the creation of quality jobs by locking workers into firms facing structural difficulties? Answering these questions is not straightforward. The synchronised nature of the health crisis across countries and the plethora of policy responses deployed by governments to stem its economic and labour market impact severely limits the possibility of providing reliable estimates of the effects of JR schemes.

To provide a preliminary assessment of the role of JR schemes during the COVID‑19 crisis, this chapter makes use of a variety of approaches. Section 2.1 offers an institutional analysis of the main features of job retention schemes based on the OECD Policy Questionnaire on Working Time Regulation and Short-Time Work Schemes (see Chapter 5 for a description of the questionnaire). Section 2.2 provides a statistical portrait of their use during the initial phase of the COVID‑19 crisis and the factors affecting it. Section 2.3 presents tentative evidence on the effects of job retention schemes on employment and hours worked as well as their possible implications for job creation. Section 2.4 provides a comprehensive discussion of the policy challenge of promoting job retention while supporting job reallocation. Section 2.5 concludes with some remarks on how schemes might be adapted as the crisis evolves following the principles summarised by the three T’s: timely, targeted and temporary.

As the COVID‑19 crisis took off, nearly all countries took steps to ensure that JR schemes provide timely and broad-based support for all firms and workers that were affected by social distancing restrictions, with limited attention to their fiscal costs or their potential implications for the recovery. Virtually all governments using JR schemes have eased access to pre‑existing schemes and increased their generosity or, if no pre‑existing scheme was present, introduced comprehensive temporary new ones. Mexico was the only country without a universal JR scheme.1 In several countries, social partners were involved in the design and implementation of the JR schemes (Box 2.1). For a detailed overview of the key features of JR schemes as they operated during the initial period of the COVID‑19 crisis, see Annex Table 2.A.1.

Countries have used a variety of different approaches to job retention during the COVID‑19 crisis (Table 2.1). Traditionally, short-time work (STW) schemes that directly subsidise hours not worked, such as the German Kurzarbeit, have been the main instrument to promote job retention during economic downturns. However, since the outbreak of COVID‑19, a number of countries have introduced temporary wage subsidy (WS) schemes that subsidise hours worked but can also be used to top up the earnings of workers on reduced hours, such as the Job Keeper Payment in Australia. A crucial aspect of all JR schemes is that employees keep their contracts with the employer even if their work is fully suspended.

Most of the STW schemes that existed already before the crisis either do not impose any significant limits on the permissible reduction in working time or only allow for partial reductions (e.g. Sweden, the United States). One reason why schemes only allow for partial reductions is to promote work-sharing and spread the costs of adjustment across the workforce. Most new STW schemes that were introduced in response to the crisis take the form of furlough schemes that only subsidise jobs whose hours are temporarily reduced to zero (e.g. Denmark, Slovenia, and the United Kingdom).2 Such schemes might be easier to implement quickly and be less susceptible to abuse based on the misclassification of part-time workers. However, by restricting support only to jobs that are fully suspended, they are also necessarily more rigid and exclude the possibility of sharing the costs of adjustment across the workforce through broad-based working time reductions (OECD, 2020[1]).

A number of countries have introduced temporary WS in response to the COVID‑19 crisis that can be used by firms for hours worked (like standard WS) as well as for hours not worked (like STW schemes) – e.g. Australia, Canada, Estonia, Ireland, New Zealand. WS are reserved for firms experiencing a significant decline in revenue. In some countries, the size of the actual subsidy only depends on the wage bill (before programme participation) and not the decline in business activity. Since such subsidies are not contingent on reducing working time, employers have strong incentives to request support for all eligible workers in the firm, raising potentially important questions about the way they are targeted (OECD, 2020[1]). A number of countries have therefore introduced mixed wage subsidies that do not just depend on the previous wage, but also the reduction in business activity, similar in spirit to STW schemes. For example, the Netherlands introduced a temporary WS that is proportional to the reduction in sales. Canada reformed its temporary WS scheme in July 2020 by making the subsidy partially proportional to the decline in sales.

Most countries provided generous JR support to firms and workers in the immediate aftermath of the COVID‑19 crisis. As a result, the cost of hours not worked was largely borne by governments (61% and 68% of labour costs for average and low-wage workers respectively on average across countries).3 However, workers often bore a significant part of the costs of reduced working hours (28% and 21% of labour costs for average and low-wage workers on average), while the costs for firms were usually smaller (7% on average) (Figure 2.1).

During the early stage of the COVID‑19 crisis, a majority of countries set to zero the cost of hours not worked, allowing firms to adjust labour costs in line with the decline in business activity. This tended to hold in countries with STW schemes as well as those with WS schemes. When some business activity remained possible, WS schemes allowed for even larger reductions in labour costs than STW schemes since they subsidise hours worked as well.4 In about half of STW countries, the cost of hours not worked was already zero, while in several others it was reduced to zero as the COVID‑19 crisis struck (e.g. France, Germany, and Italy). However, in some countries, employers have continued to bear some of the cost of idle workers. The costs of hours not worked amount to around 30% in Korea and Lithuania, and 25% in Denmark and Japan. Even in the countries where employers bear some of the costs, JR schemes allowed for significant adjustments in labour costs during the crisis, alleviating liquidity constraints and limiting the number of workers at risk of dismissal. The actual impact of JR schemes on the number of jobs saved also depends on other factors, including the timeliness of support (Box 2.2).

Workers typically bore a substantially larger share of the cost of hours not worked than firms in terms of lower wages and social security entitlements, but still were considerably better off than workers on full-time unemployment benefits, even in the case of a complete work stoppage (see Section 2.4). Workers at the average wage absorbed 28% of the costs of hours not worked on average across countries.5 The incomes of low-wage workers – defined here as workers earning 67% of the average wage – were usually better protected due to the role of benefit caps for workers with incomes above a certain threshold (e.g. Italy, Spain and the United Kingdom) or lump-sum subsidies (e.g. Australia, New Zealand, the United States).6 Low-wage workers bore 21% of the costs of hours not worked on average across countries. In countries with a statutory minimum wage, this sometimes served as a floor for benefits, so that minimum-wage workers did not experience any loss in income when put on short-time work (e.g. Belgium, France, Greece).7 In countries without a statutory minimum wage, concerns about low incomes were sometimes addressed through temporary ad hoc top-ups for low-wage workers (e.g. Switzerland). More than half of countries with a pre‑existing STW scheme increased the replacement rate for hours not worked in response to the COVID‑19 crisis. All in all, JR schemes played an important role in preventing financial hardship, particularly among low-income families, and supporting aggregate demand.

Most countries took measures to allow easy access to JR schemes. Applications can now be made mostly online, approval processes have been automated in many countries (Box 2.2) and eligibility requirements for either firms or workers have sometimes been eased.

To reduce the risk of supporting jobs that do not need support, i.e. “deadweight effects”, countries often condition eligibility to having a valid economic justification. A formal economic test based on a quantitative threshold is required in the case of all WS schemes. STW schemes do not always require an economic justification and, if there is one, this is usually less formal (not based on a quantitative threshold).8 Since subsidies are conditional on reducing working hours, which only makes sense when business activity is down, a formal economic justification may be less relevant in the context of STW. To promote access, a number of countries with STW, temporarily weakened the criteria for a valid economic justification or even completely removed the need for one. This reflected reduced concerns about deadweight effects during the initial lockdown when economic activity came to a virtual standstill. Requiring an agreement with a trade union or worker representative can also help to alleviate deadweight effects, while at the same time ensuring a sound process. Since participation is costly for workers, a firm-level agreement can help to prevent firms from claiming support when there are no jobs at risk. The need for a firm-level agreement is quite common in countries with STW schemes but never present in countries with WS schemes.

In addition to easing eligibility for firms, several countries have extended eligibility for workers. Eligibility may be limited to “insured” workers, i.e. workers who meet the minimum contribution requirements for unemployment benefits, or workers with a permanent contract, i.e. jobs that would be expected to last for a long time in the absence of the temporary shock. The focus of JR schemes on workers with recent work experience or permanent jobs is, in principle, consistent with the rationale of such schemes to preserve firm-specific knowledge that would be costly to rebuild if the worker is laid off. However, it also risks deepening labour duality, i.e. the gap in employment protection between those on open-ended and temporary contracts (Hijzen and Venn, 2011[4]). Newly introduced JR schemes typically do not impose such restrictions, while a number of countries with pre‑existing ones have removed or relaxed minimum contribution requirements (e.g. Japan, Turkey) or extended coverage to workers on temporary contracts (e.g. Finland, France and Switzerland).9 Almost all countries now cover workers on temporary contracts and a number of countries even cover certain categories of self-employed (e.g. Australia, Latvia, Poland), typically without minimum contribution requirements. However, formal eligibility to JR schemes for these categories does not guarantee actual access. Evidence presented in Chapter 1 indeed suggests that adjustments for workers in temporary contracts still occurred mostly through job destruction.

As a result of these measures, JR support during the early period of the COVID‑19 crisis was easily accessible, widely available and exceptionally generous. These measures no doubt helped to provide timely support to all firms and workers whose economic activities had suddenly been reduced or even completely come to a halt. To an important extent, this reflects the specific nature of the crisis that was driven by government-imposed economic restrictions and social distancing measures to contain the epidemic and the need to shield firms and workers as much as possible against their economic and social consequences. However, it may also reflect the tendency of JR schemes to temporarily become more cost-effective during an economic downturn (Hijzen and Venn, 2011[4]). The efficiency cost of JR support may be reduced during a recession since many more jobs are at risk, reducing the risk of supporting jobs that do not need support, while the social cost of locking workers in unviable jobs is temporarily lower since the prospects of moving quickly to more productive jobs are weakened as a result of the decline in job vacancies. At the same time, the benefits from preventing layoffs during a recession may increase due to the longer expected duration of unemployment and the increased risk of “scarring”. Importantly, both arguments suggest that the measures taken by governments to promote the use of JR schemes should be temporary and be adapted when social distancing restrictions are withdrawn and economic activity can be resume. Section 2.4 discusses how this can be done.

Partly as a result of the measures taken by governments to promote access to JR schemes and increase their attractiveness, the use of JR schemes reached unprecedented levels following the outbreak of the COVID‑19 crisis.

The use of both new and old JR schemes was widespread during the first wave of the COVID‑19 pandemic. Across the OECD, take up as a share of dependent employment peaked (mostly in April 2020) at around 20%, supporting approximately 60 million jobs, more than ten times as many as during the global financial crisis (Figure 2.2).10 Take‑up rates were lower in the few countries with STW schemes that only allow partial reductions in hours. This small group of countries includes the US which relied on the unemployment benefit system more than any other country to deal with the labour market shock of the pandemic.11 By contrast, in countries with general STW schemes, peak take‑up rates tended to be considerably higher, reaching or exceeding 30% in France, Belgium, Switzerland and Italy. All countries with WS schemes except Poland had peak take‑up rates above 20%, with New Zealand reaching a stunning 66%. High take‑up in countries with WS schemes could reflect strong incentives to apply for support for all workers in the firm rather than just those at risk, but also the ease with which such schemes could be accessed. During the global financial crisis, newly introduced JR schemes typically had difficulty achieving high levels of take‑up (Hijzen and Venn, 2011[4]).

The use of JR support declined quickly as most countries relaxed restrictions over the summer of 2020. Average take‑up fell to 7% by September 2020, just before several countries began to see a resurgence of the epidemic. The decline was particularly strong in countries with high initial take‑up. For example, France, Belgium, Switzerland and Italy saw take‑up plummeting from higher-than-average values of 30% to values in line with the new average of 7% in September 2020. In some countries, the decline was due to the termination of temporary schemes (e.g. Estonia, New Zealand), while in others this may also have reflected in part the gradual phase out of measures to promote take‑up (e.g. France, the United Kingdom). Take‑up rates declined much less in some of the countries with WS schemes. Indeed, by September, take‑up was higher in Canada, Australia, Ireland and the Netherlands than in any other country except Greece. This likely reflects that WS schemes are less targeted to firms with reduced working hours, but also the fact that these countries eased restrictions less than others during the third quarter of 2020 (Figure 2.3).

The average use of JR support remained around 6% in February/March 2021 as countries experienced diverging health developments a year after the start of the crisis. Countries that were hit by a second or even a third wave of the coronavirus such as France, Italy, and the United Kingdom saw increases in take‑up – although to levels well below the peak of spring 2020. In Greece, take‑up reached levels very close to the peak seen at the start of the crisis. Other countries, such as Canada and Sweden, saw continued declines in take‑up.

Differences in the use of JR support across countries largely reflect the intensity of the crisis rather than differences in their broader institutional settings. Peak take‑up rates are correlated with both the stringency of the lockdown measures and the fall in GDP (Panels A and B of Figure 2.3).12 For example, take‑up was very low in countries like Korea and Japan which managed to contain the first wave of the epidemic and therefore resorted to less stringent government restrictions and experienced a smaller fall in output than most other countries. By contrast, peak take‑up rates do not correlate with institutions such as employment protection legislation (EPL) and collective bargaining systems. Indeed, during the COVID‑19 crisis, even countries that traditionally relied on external flexibility, such as Australia, Canada, New Zealand and the United Kingdom, have deployed JR schemes on a massive scale. This is likely due to the nature of the crisis that in many countries suddenly forced large swaths of firms to temporarily cease their activity and stands in contrast with previous evidence that stricter EPL is associated with higher take‑up of JR schemes (Boeri and Bruecker, 2011[2]; Hijzen and Martin, 2013[7]).

As the first wave receded in most countries, the use of JR schemes declined in line with the relaxation of the lockdown measures and the rebound in GDP in the third quarter of 2020 (Panels C and D of Figure 2.3). For example, the large declines in take‑up in France, Italy, Belgium and Switzerland took place as these countries greatly relaxed the restrictions and GDP regained some of the lost ground. By contrast, Ireland, Canada and Australia made relatively smaller changes to their restrictions and saw smaller changes in take‑up which, as noted above, resulted in them having higher-than-average use of JR schemes in September 2020. The strong responsiveness of take‑up to enhanced economic conditions suggests that the unprecedented levels of take‑up are unlikely to persist once the economic recovery takes hold. However, one would expect take‑up to become more persistent as it falls to lower levels since firms with temporary difficulties will resume their activities, while firms with structural problems will continue to rely on JR support.

In the last quarter of 2020, the use of JR schemes continued to follow variations in government restrictions and economic conditions, but less closely than during the previous two quarters (Panels E and F of Figure 2.3). The weakening relationship between take‑up, government restrictions and economic conditions across countries is likely to reflect a variety of factors. While many countries saw a resurgence of the epidemic in Q4 2020, the nature, scope and enforcement of the new restrictions varied considerably between countries (as well as within countries). In addition, by the end of 2020, firms and workers had learned much more about managing the risk of contagion while performing different activities and could therefore behave differently from when the pandemic first hit. Finally, the policy context evolved as countries continued to fine‑tune their interventions to support firms and workers, including through adjustments to JR schemes.

The use of JR schemes was particularly important in sectors that were most affected by government restrictions (Figure 2.4, Panel A – see also Chapter 1). In hotels and restaurants, more than 50% of jobs were supported by JR schemes in Q2 2020 compared with 19% on average in the countries considered. Other sectors that were heavily affected by government restrictions include arts and entertainment as well as wholesale and retail. In sectors where telework is possible such as finance the use of JR support remained rather limited. While take‑up declined strongly from Q2 2020 to Q3 2020, it declined relatively more in wholesale and retail as shops were allowed to resume activity in many countries, while it declined relatively less in manufacturing, possibly due to the role of the COVID‑19 crisis for consumption and investment. The widespread use of JR schemes in services stands in sharp contrast to the experience during the global financial crisis. In France and Germany, 80% of the actual use of JR support in France and Germany was concentrated in manufacturing during the global financial crisis, compared with around 20% during the COVID‑19 crisis (OECD, 2020[1]).

The use of JR schemes across sectors also determines to an important extent their use across demographic groups (Panel B). Their use was considerably higher for young workers than prime‑age or older workers and, in some cases, there is some evidence of significant use for workers with temporary contracts albeit at a lower level than for permanent workers (see Box 2.3). These patterns are likely to reflect the fact that a disproportionate share of young and temporary workers is employed in hotels, restaurants, retail and arts – sectors which made most intensive use of JR support.13 The use of JR schemes was similar between men and women, but if anything slightly higher for women.14 The patterns by industry and demographic group are strikingly consistent across the countries for which disaggregated data are available.

This section presents a preliminary discussion of the possible effects of JR schemes during the first two quarters of the COVID‑19 crisis on job preservation as well as their possible adverse effects on job creation and reallocation.

Providing an accurate estimate of the number of jobs saved by JR support requires a well-defined and credible counterfactual for what would have happened during the COVID‑19 crisis in the absence of JR schemes. This is far from obvious due to the unprecedented and synchronised nature of the COVID‑19 crisis that followed from the combination of legal restrictions imposed by governments and behavioural responses by workers and firms to the epidemic. Moreover, governments across the OECD responded decisively through a range of measures to contain the impact of the health crisis on the economy and the labour market by providing liquidity to firms, via JR schemes but also tax deferrals, direct income‑support to workers and households and many other measures (Chapter 1). These factors greatly complicate the task of defining a meaningful counterfactual based on either previous crisis episodes or by making comparisons across countries that differed in their use of JR support and not much else.

Bearing these limitations in mind, this sub-section presents some tentative insights on the plausible range of the effect of JR support on jobs saved.15 The estimates are based on two simple pair-wise correlations which relate the change in JR use during the COVID‑19 crisis to respectively the change in hours worked per employee and the change in the number of employees. The correlation between the change in average hours worked and the use of JR support provides an indication of the number of jobs saved under the assumption that firms only used the scheme to support jobs that they would have otherwise terminated (i.e. no efficiency losses). The correlation between JR support and the change in the number of employees provides an indication of the number of jobs saved under the assumption that employment growth is not affected by any other factors that also correlate with JR take‑up (i.e. other policy interventions, pre‑existing policies and institutions, the size and nature of the shock). While the latter set-up in principle allows for efficiency losses, it is also clear that there are important confounding factors that are not easily taken into account. To the extent that larger shocks cause both an increase in the use of JR support and a decline in employment, the implied jobs effect is most likely underestimated and should be interpreted with due caution.16

The correlations of JR support on the one hand and average hours worked and employment on the other, point to a broad range of plausible estimates of the jobs impact of JR schemes during the COVID‑19 crisis (Figure 2.6).17 As expected, the use of JR schemes is associated with a strong reduction in hours worked per employee across countries (Panel A). The fitted line suggests that a 10% increase in the use of JR support is associated with a reduction in hours worked per employee of about 4% (on average between Q2 and Q3), equivalent to an average reduction in working time for those on JR support by 40%.18 As discussed above, under the assumption of no efficiency losses, this also implies a jobs impact of 4%, which could therefore be considered an upper bound to the true effect on jobs. The correlation between the use of JR support and employment is weaker: a 10% increase in the use of JR support is associated with a 1.2% higher level of employment (Panel B). Taken at face value, a relatively small effect of JR schemes on jobs may reflect the ability of firms to retain workers independently of JR support, due to complementary measures taken by governments to provide liquidity to firms. However, as mentioned above, there are good reasons to believe this is an underestimate of the true jobs impact.19

The implications of these different jobs estimates of the use of JR schemes for the evolution of employment during the COVID‑19 crisis are visualised in Figure 2.7. It compares the actual change in the total number of employees with the implied counterfactual changes that would have occurred in the absence of JR support. The counterfactual changes have been obtained by adjusting the actual change in employment for the estimated effect of JR use on hours and employment. Across the OECD, the number of employees fell sharply by about 4% between Q1 2020 and Q2 2020, and then started to recover gradually. The adjusted employment series based on the correlation between the change in JR support and the change in average hours worked indicates that in the absence of JR support – not allowing for possible efficiency losses – the fall in the number of employees in Q2 2020 might have been as large as 11%. The adjusted employment series based on the correlation of the change in JR support and the change in the number of employees across countries suggests that, in the absence of JR schemes, the decline in the number of employees would have been almost 50% larger than the actual change in employment, resulting in a decline in employment of more than 6%. The large differences in the evolution of employment after adjusting for the estimated associations of JR schemes with average hours worked and employment clearly highlight the high degree of uncertainty that surrounds these estimates. A better understanding of the role of JR schemes during the COVID‑19 crisis may be obtained as time passes and more detailed data become available.

Given the difficulty of coming up with a well-defined and credible counterfactual of what would have happened in the absence of JR schemes using country level data, there have been only few attempts to quantify their effects so far and those that have done so rely on detailed microdata. An early evaluation of the Australian JobKeeper Payment finds that it saved the job of one in five employees who received the payment or about 700 000 jobs over the period April to July 2020 (Bishop and Day, 2020[10]). Given the modest size of the lump-sum subsidy, this effect seems large, suggesting that the JobKeeper Payment was an effective instrument to preserve jobs. Evidence for the Paycheck Protection Program in the United States remains inconclusive, with some studies finding very small and others finding very large employment effects (see Box 2.4).

The main concern with the widespread use of JR schemes is that, when used for too long, there is a risk that they undermine job creation in the recovery and slow down the reallocation of jobs from low to high performing firms.20 For these concerns to have materialised so far, three conditions need to be met. First, JR schemes must have supported unviable jobs in firms with structural difficulties (low productivity firms). Second, in the absence of JR support, these workers would have been laid off and would swiftly have found a job in another firm. Third, it must have become more difficult for firms to fill their job vacancies as a result of the use of JR schemes. Each of these issues will be reviewed in turn below, based on the available data and preliminary evidence for the first year of the COVID‑19 crisis.

JR schemes may end up supporting unviable jobs in firms with structural difficulties for two main reasons. First, the JR schemes may support firms with pre‑existing structural difficulties. This is likely to be small relative to the number of jobs supported by JR schemes. As documented above, the use of JR support has been widespread across sectors and was closely related to presence of economic restrictions since the start of COVID‑19 crisis. While a number of countries that relied heavily on JR schemes exhibited lower layoffs than in normal times, suggesting that JR schemes and other firm-support measures have tended to preserve some jobs that in the absence of the crisis would have been terminated, the number of these “missing layoffs” so far is likely to have been rather small.21^, 22 Second, JR schemes may have supported jobs that had become permanently unviable as a result of the COVID‑19 shock itself: it did not just represent a temporary shock but also a permanent one, requiring the reallocation of jobs across firms and sectors.23 While it is too early to say to what extent the COVID‑19 crisis is likely to induce permanent changes, it is likely to accelerate a number of important pre‑crisis trends related to the spread of online activities, the incidence of working from home and the automation of production (see Chapter 1).

Even in the event that JR schemes were supporting many permanently unviable jobs, this does not necessarily mean that they were also slowing job creation and reallocation. If there is weak effective demand for workers in permanently unviable jobs, letting them go would merely increase unemployment, without any effect on job creation and reallocation. This is more likely when job vacancies are relatively low and the number of unemployed persons competing for these job vacancies is relatively high. This is visualised in Figure 2.8 using Beveridge Curves which trace out combinations of job vacancies and unemployed job seekers during the COVID‑19 crisis (from 2020M1 to 2020M12) and, to put this in perspective, also during the global financial crisis (from 2008M6 to 2011M12).24 It shows that as a result of the COVID‑19 crisis, the ratio of vacancies to the number of unemployed initially sharply declined as job vacancies plummeted and the number of unemployment jobseekers tended to increase. As time passed, job vacancies started to recover while unemployment continued to edge up, resulting in an outward movement of the Beveridge Curve. With the exception of the United States, however, job vacancies remained well below their pre‑crisis levels until the end of 2020. At least in part, this is likely to reflect weak labour demand related to the high degree of uncertainty about the short-term outlook. The continued weakness of labour demand mitigates concerns about the possible adverse effects of JR schemes on job creation and reallocation. Indeed, the resulting increase in the expected duration of unemployment and the competition for jobs among the unemployed provides a justification for temporarily increasing the generosity of JR support (and unemployment benefits) in the context of an economic downturn (Landais, Michaillat and Saez, 2018[18]; Hijzen and Venn, 2011[4]).25

A further question is whether supporting jobs in firms with structural difficulties has undermined job creation by making it more difficult for firms to fill their vacancies. To shed some light on this question, so-called “matching functions” are estimated, which describe the ease with which job openings can be filled (or unemployed job seekers can find jobs) for a given level of labour market tightness (Petrongolo and Pissarides, 2001[19]; Federal Reserve Bank of San Francisco et al., 2011[20]). Figure 2.9 documents the actual evolution of the job-filling rate, the ratio of hires over vacancies, as well as its expected evolution based on actual changes in labour market tightness (see Box 2.5 for details). The figure provides two insights. First, actual job-filling rates have tended to increase since the start of the COVID‑19 crisis. This is the typical pattern observed during economic downturns as more workers are competing for fewer job vacancies. Importantly, the increase in the job-filling rate suggests that it has not become more difficult to fill job vacancies. Second, the job-filling rate tended to increase less than what might have been expected based on its relationship with labour market tightness in the period before the COVID‑19 crisis. While this may point to a decline in matching efficiency, it is unlikely to be related to the unprecedented use of JR schemes. Indeed, there is no clear indication in the data that the use of JR schemes played a significant role in explaining the‑smaller-than-expected increase in the job filling rate (Figure 2.10).

The limited increase in the job-filling rate might, instead, be due to the nature of the crisis itself. The asymmetric impact of the crisis across sectors with different skill requirements might have produced a mismatch between skills of unemployed jobseekers and those required by employers, as observed during previous recessions (Şahin et al., 2014[21]). Indeed, one of the objectives of JR schemes is to prevent the build-up of such skill mismatches. Moreover, unemployed workers might have limited their search efforts due to concerns linked to the severity of the health and economic situation – see Chapter 1. For example, evidence for Sweden shows that job-search declined more sharply than vacancies in the period immediately after the COVID‑19 outbreak, making it harder to fill jobs despite a larger number of unemployed per vacancy (Hensvik, Le Barbanchon and Rathelot, 2021[22]).

With job vacancies and job search remaining depressed in most countries, JR schemes mainly helped preserve jobs and limit the surge in unemployment, probably without a significant impact on workers’ flows towards expanding firms. As a result of often strict social distancing restrictions, many jobs were temporarily at risk, limiting the risk of subsidising jobs that do not need support (deadweight costs), while the risk of significantly stifling job creation was small since the scope for effective job search and job mobility was limited. This justified the measures taken by governments to promote the use of JR support during the initial phase of the COVID‑19 crisis.

Going forward, concerns about the possible adverse effects of JR schemes on job reallocation are growing, but withdrawing the support too early or too widely risks generating a sudden wave of layoffs.26 A careful scaling back of JR support should be implemented flexibly, closely following the evolution of the economic and health situation, and likely requires a differentiated approach across sectors. Firms and sectors whose ability to operate remains constrained by health recommendations or legal restrictions should continue to receive strong JR support. In other sectors, however, continuing to provide generous support after the end of health and legal restrictions carries an increased risk of supporting unviable jobs and consequently potentially undermine job creation and reallocation as economic activity resumes. Here, the challenge for governments is to adapt JR schemes to provide incentives for firms and workers to use support only for jobs that are temporarily at risk but remain viable in the longer term. Indeed, as firms with temporary difficulties are likely to leave support when activity can resume – as witnessed during the third quarter of 2020 – the share of firms with structural problems is likely to increase.

This section discusses the dimensions along which job retention schemes can be adapted and how the changes can be implemented keeping into account the uncertain evolution of the crisis and its differentiated impact across sectors. It focuses on the following three key issues:

• Limiting the maximum duration of support

• Enhancing the targeting of support

• Supporting workers in jobs at risk

While support should remain in place as long as government-imposed health measures restrict economic activity, once these measures are withdrawn and economic activity can resume, support should become time‑limited to reduce the risk of subsidising jobs that no longer need support or jobs that have become unviable. Since STW schemes provide relatively strong incentives for resuming working hours once business conditions have improved, the main purpose of maximum limits for those schemes is to avoid supporting jobs with structural problems. By contrast, WS are not conditional on reducing working time and therefore provide strong incentives for drawing support as long as possible irrespective of the financial need or the viability of the job. To limit the risk of subsidising jobs that do not need support (i.e. deadweight losses), firm eligibility needs to be re‑assessed regularly in WS schemes. Conditional on eligibility requirements being met, limits to the effective maximum duration serve the same purpose as in STW schemes, i.e. reducing the risk of preserving jobs that have become unviable in the longer term.

The maximum duration of support varies considerably across JR schemes. In countries with STW schemes, the maximum duration varies from about three months in Japan, Latvia and Portugal to 18 months in Switzerland up to 24 months in Germany. In Germany, firms which started using STW before 31 December 2019 can continue to do so for up to 24 months without reassessment of eligibility until the end of 2021 (compared to 12 months in normal times). While the maximum duration of support in Switzerland is also quite long, its continued use is subject to regular reassessments of firm eligibility (every 3 or 6 months).27 In countries with WS schemes, the maximum duration of support tends to be short (around 3 months) or firm eligibility is re‑assessed regularly. In countries such as Australia, the Netherlands and the United Kingdom, several extensions have been introduced, allowing for additional support for firms that remain eligible, albeit in some cases under less generous conditions.

Limiting the maximum duration prevents that jobs are supported for too long, particularly those with limited prospects of recovering, but does not ensure that subsidies go to the right jobs, i.e. jobs that are temporarily at risk but will be able to resume in the not too distant future. Firms and workers typically have a better sense of the viability of jobs than governments, particularly in the current context where social distancing restrictions remain important and market signals are weak. Financial incentives for firms and workers can help enhance the targeting of JR support to jobs that are temporarily at risk, but viable in the longer term.

While the cost of reducing working time was set to zero in response to the COVID‑19 crisis in most countries, governments should consider (re)introducing some cost-sharing with firms in sectors where restrictions are withdrawn and activity can resume. While this would reduce the attractiveness of STW for firms in general, it would also strengthen incentives to use support only for jobs that are likely to re‑start after the crisis and to resume regular work schedules as soon as possible. A number of countries have introduced co-financing for firms, while others have increased the rate (Figure 2.11). For example, since 1 June 2020, firms in France were required to pay 10% of the cost of hours not worked (15% of the benefit) in sectors that were no longer subject to economic restrictions.28 This is expected to increase to 24% from 1 July 2021 (40% of the benefit). The United Kingdom had started to gradually increase the cost to employers for keeping workers on furlough from July 2020, but as the health situation deteriorated in the fall, largely reverted to its original version with firms only required to pay social security contributions over hours not worked. Employers in Germany will be liable to pay 50% of social security contributions from July 2021 and 100% from January 2022, which is the usual rule during short-time work. In Norway, the waiting period during which firms have to pay full wages has been increased from 2 to 10 days since November 2020. Latvia introduced a new scheme in response to the second wave of the virus which requires firms to cover up to 20% of the costs of hours not worked, while employers did not have to contribute anything with the temporary scheme that was operational during the first wave.

The optimal contribution of firms for hours not worked depends on efficiency considerations related to the risk of slowing job reallocation and equity considerations related to the risk of reinforcing labour market duality.

If the objective is to prevent supporting jobs that are permanently unviable, a small co-payment by firms should be sufficient. This would help to reduce the risk of supporting permanently unviable jobs from the start of programme participation rather than only after the maximum duration of support. In practice, the objective of co-financing is likely to go beyond preventing permanently unviable jobs by limiting support to jobs that are viable after the maximum duration of support. To the extent that jobs that are not viable after this period will be terminated, it may be preferable to avoid supporting such jobs in the first place. This requires a more significant co-payment by firms. To provide a broad idea of the approximate co-financing rate, Box 2.6 considers a set of numerical examples. For STW schemes with a maximum duration of 12 months, requiring firms to pay 8.5% of the costs of hours not worked may be enough to limit support to jobs for which the costs of closing and re‑opening them exceeds one month of pay. Having a shorter maximum duration or restricting support to jobs with larger replacement costs would be consistent with a larger co-payment.

While co-payments by firms increase the targeting of support to viable jobs (and firms), they can have unintended consequences. First, co-payments may reinforce firms’ financial difficulties and increase the risk of bankruptcies. To avoid increasing the risk of bankruptcies in solvent but illiquid firms, co-financing by firms could take the form of a delayed payment or (zero‑interest) loan (Cahuc, Kramarz and Nevoux, 2018[23]; Burdett and Wright, 1989[24]).29 Second, co-payments by firms risk reinforcing labour market duality by limiting support to workers with high replacement costs, e.g. permanent workers with significant firm-specific skills and severance pay entitlements, and laying off workers with low replacement costs, e.g. temporary jobs or jobs with limited skill requirements. Indeed, Tilly and Niedermayer (2017[25]) show using administrative data for Germany that take‑up is increasing in tenure and experience.30 This reflects weak incentives for firms to hoard such workers rather than concerns over the viability of these jobs in the future. A potentially more nuanced but also much more complex and so far untested approach would be to link the co-payment to tenure or contract status.31

While WS schemes typically have short maximum durations, they have been extended multiple times in a number countries, raising similar issues as STW in relation to reallocation. To address this issue to some extent, governments could ensure that employers bear some of the cost of hours not worked – at least for large reductions in working hours – by requiring firms to pay a fraction of a workers’ usual wage regardless of hours worked with the subsidy set to cover only part of that pay.32 The WS scheme operated in the Netherlands mimics STW schemes that require firms to share some of the cost of hours not worked. While workers continue to receive 100% of their earnings during periods of reduced working time, employers receive at most 90% of the wage, depending on the decline in business activity. Since 1 October 2020, firms are allowed to reduce the wage bill by 10%, be it through quits or layoffs during a 3‑month period, without a reduction in the subsidy. The implications for job retention and job reallocation of this measure are discussed in Box 2.7.

Balancing job retention and reallocation also requires striking the right mix of in-work and out-of-work support in the form of JR and unemployment benefits (UB). In most countries, JR benefits for hours not worked exceeded regular unemployment benefits during the initial period of the COVID‑19 crisis (Figure 2.13).33 The largest differences with UB can be found in countries with temporary JR schemes such as Denmark and the Netherlands, which offer full income protection to workers, as well as countries with means-tested UB such as Australia, New Zealand and the United Kingdom. Higher replacements rates for workers increase the attractiveness of short-time work in comparison to (full) unemployment and the willingness of workers, including those not directly at risk of being laid off, to accept a reduction in working hours as part of a STW scheme.34 One argument why JR benefits should be more generous than UB in the beginning of a deep crisis may be that it helps to limit congestion in the labour market, as many unemployed search for a limited number of jobs (Giupponi and Landais, 2018[26]; Lalive, Landais and Zweimüller, 2015[27]), and prevents the PES from being overwhelmed (OECD, 2020[1]). In other words, the flattening-the‑curve argument that has been used to justify lockdown measures to contain the health crisis equally applies to the use of JR schemes to contain the jobs crisis.35

As concerns about congestion in the labour market diminish and those about the cost-effectiveness of JR support become more important, there may be a case for rebalancing JR with UB support, by reducing the gap between STW benefits and regular unemployment benefits, notably in countries with particularly generous JR benefits. These changes would help contain the overall cost of JR schemes, and might improve the targeting of STW schemes to jobs at risk of being destroyed by reducing the willingness of workers to accept STW. Lower subsidies might also increase incentives for workers to resume normal working hours or actively look for another job altogether (see also discussion on financial incentives to promote job mobility below). Even with a smaller difference when compared to unemployment benefits, JR is likely to remain attractive because it preserves the employment contract and the non-wage benefits linked to it (fringe benefits, social security, including access to health insurance in some countries). In France’s general STW scheme, the gross replacement rate has been reduced from 70% to 60% in April 2021, while in the Australia and the United States the lump-sum benefit was reduced.

There may also be an argument for letting benefits decline over the support period in countries where the maximum duration is relatively long. One justification for this could be that the risk of supporting permanently unviable jobs increases with the duration of support. Another is that since in most countries unemployment benefits decline over the spell already, a similar benefit schedule may be needed for JR benefits to prevent the gap between the two from increasing over time. To the best of our knowledge, no country operates a JR scheme with a declining benefit schedule. Interestingly, Germany temporarily operates increasing benefit schedules. This is based on the observations that i) during the COVID‑19 crisis workers in the services sector are hit particularly hard, ii) these workers are less likely to be covered by top-ups in sectoral collective agreements and iii) the risk of financial hardship is likely to increase over the period of reduced activity. Korlsrud (2018[28]) recently proposed a similar argument for having increasing benefit schedules for unemployment benefits (see Hijzen and Salvatori (2020[29]), for a discussion of this issue in the context of unemployment benefits).

How to adjust the level of support available for firms and workers has been a major challenge for governments. The consensus over the course of the COVID‑19 crisis has been to protect firms and workers as much as possible from the direct effects of health-related restrictions on economic activity.36 As a result, firms typically were able to reduce working time in line with the decline in business activity at no or limited costs and benefits for workers were exceptionally generous (OECD, 2020[1]). However, this also meant that adjustments needed to be made as economic restrictions were withdrawn, particularly in countries where significant special measures were introduced in response to the crisis. This was complicated by the fact that restrictions were often not universal, but limited to specific sectors or regions and the risk of a relapse in the form of a second or a third wave loomed. Adjustments typically related to the generosity of support to firms and workers, i.e. the contribution rate by firms and the replacement rate to workers for hours not worked. In some cases, adjustments have also been made with respect to firm eligibility (tightening eligibility as government measures are withdrawn or economic conditions improve).

To reduce uncertainty about the degree of JR support available in the future, one promising avenue is to relate the degree of JR support more directly to the evolution of the health and the economic situation. Most countries already make use of dashboards to monitor the health and economic situation and inform policy. In principle, these could be used to explicitly relate the level of JR support to the health and economic situation. As long as social distancing restrictions remain significant, the health situation, possibly differentiated across regions, may be the main determinant for the degree of support. To take account of the differential effect of social distancing measures across sectors or firms, information on sales may be used as a complement. Once social distancing restrictions have been withdrawn, the strength of the economic recovery could determine to what extent prolonged support is necessary. This could be based on a set of readily available indicators of the state of the labour market, such as statistics on vacancies, hires and separations, the unemployment rate or broader measures of labour market slack accounting for underemployed and marginally attached workers (issues that are particularly important in the current context – see Chapter 1).

Based on these health and economic indicators, the generosity of JR support to firms and workers could be allowed to evolve gradually. For example, the cost of hours not worked for firms could be allowed to increase gradually to its desired long-term level in countries with permanent schemes and 100% in countries where JR support is supposed to remain temporary. Similarly, the benefit for workers for hours not worked could be gradually brought in line with that for unemployed workers in countries where this is considerably higher. To provide a maximum degree of transparency and predictability about the degree of JR support available to firms and workers, it is important to not just monitor the health and economic situation, but also provide forecasts of their evolution in the near term and their implications for the availability of JR support under different scenarios of the short-term outlook.

To take account of the fact that the impact of economic restrictions varies across sectors, regions and firms, some countries differentiated JR support. France for example applied different rules for sectors that remained subject to government-imposed restrictions and those that were not. Firms in sectors without major government restrictions were required to contribute 10% of the cost of hours not worked (expressed in usual gross wages) from 1 July 202037 and 24% from 1 July 2021, while this has remained zero in sheltered sectors. Moreover, from 1 May 2021, support without co-financing by firms in sheltered sectors has been restricted to firms experiencing significant reductions in revenue (more than 80%), with other firms being gradually brought into the general regime.38 Portugal introduced a new temporary scheme in August 2020 that requires firms to pay for 30% of the costs of hours not worked when working time is reduced by less than 60%, while there is no co-financing for firms with larger reductions in working time. Moreover, firms with a reduction in sales of more than 75% firms receive an additional WS to contribute to the cost of hours worked.39^, 40

While the main aim of JR schemes is to preserve jobs, they will not be successful in all cases as some jobs may have become permanently unviable. Some workers in subsidised jobs may have limited career prospects and remain at risk of losing their job eventually. Government can promote the career prospects of workers by investing in the viability of their current job or facilitating job mobility across firms.

The permanent mobility of workers from subsidised to unsubsidised jobs41 can be promoted through the provision of effective public employment services (PES), by engaging employers in supporting job mobility among workers in jobs at risk and by strengthening the incentives of workers for job mobility.42

Governments can promote job mobility among workers in subsidised jobs by encouraging or requiring workers on short-time work to register with the PES and benefit from their support (e.g. job-search assistance, career guidance and training (cf. Chapter 3). OECD analysis shows that early interventions – including those before job displacement takes place – can be very effective in promoting smooth job transitions (OECD, 2018[30]).

Mandatory registration requirements exist in only few countries, and typically concerns countries where JR support is part of the overall UB system and subsidies are paid directly to the worker (e.g. Finland, Ireland (STW), and Norway).43 Mandatory job-search requirements are even less common and where they exist, they do not tend to be enforced in practice (e.g. Finland, Norway). This is not surprising since workers on JR schemes are supported to keep them in their existing job. Consequently, the main purpose of mandatory registration requirements in the context of JR schemes may be to encourage workers to make use of PES rather than to impose job-search requirements.

About half of countries allow for the voluntary registration of workers on JR support with the PES and enable such workers to access their services (e.g. job-search assistance, career guidance, counselling). This includes both countries with STW schemes (e.g. Belgium, Canada, Germany, Italy, and Switzerland) and countries with WS schemes (e.g. Australia, Canada). While little is known about the actual number of people in subsidised jobs who register with the PES, let alone their level of engagement with any of their activities, there are good reasons to believe that this tends to be rather modest in practice.44Governments can promote registration with the PES by reaching out to firms that make use of JR support with information about the process and the potential benefits this can provide. For example, the Ministry for Social Development in New Zealand contacted firms using JR support with information about training opportunities.

Firms could also play an active role in promoting the mobility of workers at risk of dismissal. For example, the Netherlands requires employers using JR support to stimulate career development among employees. Concretely, employers are obliged to inform the works council when using JR support and to contact the PES when dismissing a worker for economic reasons. These notification requirements provide a natural starting point for discussing possible development options. Development measures can be financed through education and development funds (“O&O fondsen”) managed by the social partners or through additional funding made available by the government. Sweden has a well-established system of Job Security councils that can be activated to provide income support and employment services in the case of collective dismissal (OECD, 2018[31]).

Financial incentives could also be used to promote the mobility of workers from jobs supported by JR schemes to other jobs. For example, it may be possible to temporarily reduce employee social security contributions for workers starting a new job in a firm not receiving JR support or to temporarily provide an in-work benefit. These job mobility “bonuses” compensate workers for the cost of changing jobs, while at the same promote the reallocation of workers from subsidised jobs to expanding firms. A similar bonus could be made available for persons moving from unemployment into work as well as new labour market entrants (e.g. youth). The fiscal cost of such a measure would most likely be limited if kept for a short period of time (say 3 months) since the cost-saving effect of reduced JR (or unemployment) benefits would most likely outweigh the cost of the bonus (Cahuc, Carcillo and Le Barbanchon, 2019[32]). In Spain, it is possible to keep JR benefits for a limited period of time after having moved to another firm.45

Participation in training while on reduced working hours can help workers improve the viability of their current job or improve the prospect of finding a new job. Several countries actively encourage training during short-time work by providing financial incentives to firms or workers (e.g. France, Germany, Japan, Portugal) or setting expectations (e.g. Netherlands). In France, employers were fully reimbursed for the costs of training and for 70‑80% since March 2021. In Germany, training costs are reimbursed up to 100%, with larger subsidies for smaller firms. In addition, from July 2021, as the obligation to pay social security contributions on hours not worked is re‑introduced, firms that place their workers on training will have them discounted by 50% till July 2023. In the Netherlands, employers applying for JR support have to declare that they actively encourage training since June 2020, while the government has taken additional measures to make on‑line training and development courses freely available. Formal training requirements as a condition for JR support currently do not exist.46 By contrast, in Italy, Greece and Chile, training is not allowed during subsidised hours. This most likely reflects concerns about abuse in relation to training provided within the firm to the extent that this makes it difficult to distinguish between work and training, particularly when provided informally. To some extent, this can be addressed by allowing only for formal training. Any remaining concerns about abuse should be weighed carefully against the costs of excluding training while on reduced working time completely.

There is limited information on the use and nature of training whilst on JR support during the COVID‑19 crisis. Evidence from the global financial crisis suggests that the use of training was negligible in most countries (Hijzen and Venn, 2011[4]). To some extent this reflects the difficulty of engaging employed adults in training more generally. Only about 40% of all adults participates in training in normal times (OECD, 2020[33]). However, it is also likely to reflect the difficulty of organising training in such a way that it can be combined with temporary reductions in working time. This is easier when training courses are targeted at individuals rather than groups, delivered in a flexible manner through online teaching tools and their duration is relatively short (OECD, 2020[34]). France has been one of the countries that has been most successful in engaging workers on JR support with training (Box 2.10). In November 2020, almost one in five workers on JR support participated in training.

Job retention (JR) schemes have been the main policy tool used by most OECD countries for stemming the labour market impact of the COVID‑19 crisis. This has helped limit the increase in unemployment and prevent financial hardship among workers and their families. At the same time, preliminary evidence suggests that concerns that JR schemes might significantly slow the reallocation of jobs to high-performing firms and undermine job creation have not materialised so far. Job vacancies remain depressed on the whole and there is no evidence that the use of JR schemes has made it more difficult to fill them. The risk of growing dependency on JR benefits seems limited. During periods when business activity was able to resume, take‑up declined quickly. Nevertheless, as jobs in firms facing temporary difficulties restart and are no longer covered by the programmes, the share of supported jobs in firms with structural problems is likely to increase. This is why JR support has to remain temporary and become more targeted on jobs in temporary need of support due to the effects of social distancing restrictions rather than those at risk because of pre‑existing structural difficulties. Indeed, well-designed JR schemes need to be timely, targeted and temporary:

Timely. To be effective, JR support needs to be timely. The unprecedented levels of take‑up soon after the outbreak of the COVID‑19 crisis in many countries suggest that by and large this was the case. Moreover, and in contrast to the experience during the global financial crisis, high take‑up was achieved both with pre‑existing schemes and new ones that were introduced in response to the COVID‑19 crisis. In almost all countries, applications could be made on-line and the approval process was largely automated. However, there are important differences across countries in the time between application and the actual payment of the subsidy. Reducing the delay in payments to a minimum is crucial for the effectiveness of JR support. Where possible, payments should be made partly in advance. Proof of eligibility, when required, and the circumstances determining the amount of the subsidy can be verified ex post.

Targeted. As the COVID‑19 crisis took off, most countries took steps to ensure that JR schemes provide broad-based support for all firms and workers that were affected by social distancing restrictions, with limited attention to their fiscal costs or implications for the recovery. However, this also increased the risk of subsidising jobs that did not need support and jobs that would have disappeared even in the absence of the pandemic. To prevent JR schemes from becoming an obstacle to job reallocation and job creation in the recovery, it is important that their targeting is enhanced by strengthening the conditions for eligibility and bolstering incentives of firms and workers to limit support to jobs that are at risk, but remain viable in the medium term. Where necessary the rules can be differentiated between sectors and firms according to the impact of social distancing restrictions on their activities.

Temporary. JR schemes are an important tool for limiting excessive layoffs in the context of a temporary reduction in business activity, but should not be used to support firms with structural difficulties as this risks undermining job reallocation and the creation of good jobs. While support should remain in place as long as government measures restrict economic activity, the use of JR support should be time‑limited once economic activity can resume. It may be preferable that firms with structural problems restore financial health by restructuring. Workers who lose their jobs should be supported through effective income‑support schemes and re‑employment support provided by the public employment services (cf. Chapter 3).

Adapting JR schemes to the evolving crisis is challenging. The degree of uncertainty over the short-term outlook remains very high and the effects of social distancing restrictions vary widely between sectors and regions. This raises important questions about the speed with which temporary support measures are withdrawn and for whom. One promising avenue may be to link the degree of JR support more directly to developments in the health situation and economic activity. Most countries are already making use of dashboards to monitor the health and economic situation and these could be used to inform policy on JR support. As long as social distancing restrictions remain significant, the health situation could be used as the main determinant for the degree of support, possibly on a regional basis. Meanwhile, information on sales could be used to decide the level of support for individual sectors or firms. Once social distancing restrictions have been withdrawn, the strength of the economic recovery could determine the extent to which prolonged support is provided. The generosity of JR support to firms and workers could therefore be adjusted gradually, based on developments in the health and economic situation.

Beyond the refinements needed to JR schemes over the near term, an important question is also whether it is necessary to have a permanent scheme in place so that this can be scaled up if necessary, or whether a temporary scheme could simply be introduced when the need arises. As mentioned above, several countries without pre‑existing JR schemes have been able to provide timely support to firms and workers, in large part thanks to the use of digital technologies in the application and payment process. While this is clearly positive, it did not come without any costs. To achieve high levels of use, new schemes had to be sufficiently simple and necessarily tended to be somewhat crude. For example, most new short-time work schemes only provided support in the case of a temporary suspension of work, while some temporary wage subsidy schemes only made lump-sum payments. This suggests that there may be a case for having a limited permanent scheme in place that can be scaled up quickly in the context of a major crisis and cut back gradually as economic activity picks up. The nature and scope of such a scheme is likely to vary between countries depending on their preferences and existing institutional settings.

There are several important issues left for future work. The most important is no doubt to provide a more comprehensive ex-post evaluation of the effectiveness of JR schemes in preserving jobs and supporting job creation during the COVID‑19 crisis once data become available. A key aspect of such any such evaluation should be to analyse the effectiveness of JR schemes in protecting different types of workers from the risk of job losses, as well as the longer-term career paths. While JR schemes have helped to prevent a surge in unemployment, it is not clear to what extent they have been equally effective in protecting the jobs of different groups of workers.  Evidence presented in Chapter 1 suggests that those in high-pay occupations were much likely to keep their jobs, even if their hours were reduced to zero hours, while those in low‑pay occupations were more likely to be laid-off. Moreover, despite efforts to open up JR schemes to workers on temporary contracts, concerns remain about the effectiveness of JR schemes to protect workers in such jobs. To some extent these differences in the effectiveness of JR schemes across different groups may be related to their design. For example, co-financing requirements for firms are likely to reinforce such concerns. Empirical evidence on the effectiveness of JR schemes in protecting different groups of workers will help fine‑tune their design but also provide a better understanding of their limits and how they should be complemented by other policy tools, including income support schemes.

[38] Adams‐Prassl, A. et al. (2020), “Furloughing*”, Fiscal Studies, Vol. 41/3, pp. 591-622, http://dx.doi.org/10.1111/1475-5890.12242.

[11] Autor, D. et al. (2020), An Evaluation of the Paycheck Protection Program Using Administrative Payroll Microdata *.

[39] Barrero, J., N. Bloom and S. Davis (2020), COVID-19 Is Also a Reallocation Shock, https://oui.doleta.gov/unemploy/claims_arch.asp. (accessed on 18 February 2021).

[13] Bartik, A. et al. (2020), The Targeting and Impact of Paycheck Protection Program Loans to Small Businesses, National Bureau of Economic Research, Cambridge, MA, http://dx.doi.org/10.3386/w27623.

[10] Bishop, J. and I. Day (2020), How Many Jobs Did JobKeeper Keep?, https://www.rba.gov.au (accessed on 17 February 2021).

[2] Boeri, T. and H. Bruecker (2011), “Short-time work benefits revisited: some lessons from the Great Recession”, Economic Policy, Vol. 26/68, pp. 697-765, http://dx.doi.org/10.1111/j.1468-0327.2011.271.x.

[24] Burdett, K. and R. Wright (1989), “Unemployment Insurance and Short-Time Compensation: The Effects on Layoffs, Hours per Worker, and Wages”, Journal of Political Economy, Vol. 97/6, pp. 1479-1496, http://dx.doi.org/10.1086/261664.

[3] Cahuc, P. and S. Carcillo (2011), Is Short-Time Work a Good Method to Keep Unemployment Down?, https://pdfs.semanticscholar.org/17bc/c2fab64fd829f868667dac2186babf38bf2c.pdf (accessed on 25 October 2018).

[32] Cahuc, P., S. Carcillo and T. Le Barbanchon (2019), “The effectiveness of hiring credits”, Review of Economic Studies, Vol. 86/2, pp. 593–626, https://doi.org/10.1093/restud/rdy011.

[23] Cahuc, P., F. Kramarz and S. Nevoux (2018), When Short-Time Work Works, http://www.iza.org (accessed on 25 October 2018).

[12] Chetty, R. et al. (2020), The Economic Impacts of COVID-19: Evidence from a New Public Database Built Using Private Sector Data, National Bureau of Economic Research, Cambridge, MA, http://dx.doi.org/10.3386/w27431.

[37] Chodorow-Reich, G. et al. (2020), Bank Liquidity Provision Across the Firm Size Distribution, National Bureau of Economic Research, Cambridge, MA, http://dx.doi.org/10.3386/w27945.

[43] Cominetti, N. et al. (2021), Long Covid in the labour market: The impact on the labour market of Covid-19 a year into the crisis, and how to secure a strong recovery., Resolution Foundation, London, https://www.resolutionfoundation.org/app/uploads/2021/02/Long-covid-in-the-labour-market.pdf (accessed on 26 April 2021).

[16] Doniger, C. and B. Kay (2021), “Ten Days Late and Billions of Dollars Short: The Employment Effects of Delays in Paycheck Protection Program Financing”, SSRN Electronic Journal, http://dx.doi.org/10.2139/ssrn.3747223.

[6] Eurofound (2021), Involvement of social partners in policymaking during the COVID-19 outbreak, Publications Office of the European Union, Luxembourg.

[15] Faulkender, M., R. Jackman and S. Miran (2021), “The Job Preservation Effects of Paycheck Protection Program Loans”, SSRN Electronic Journal, http://dx.doi.org/10.2139/ssrn.3767509.

[20] Federal Reserve Bank of San Francisco et al. (2011), “Which Industries are Shifting the Beveridge Curve?”, Federal Reserve Bank of San Francisco, Working Paper Series, pp. 1.000-27.000, http://dx.doi.org/10.24148/wp2010-32.

[45] Forsythe, E. et al. (2020), “Searching, Recalls, and Tightness: An Interim Report on the COVID Labor Market”, Vol. NBER Working Paper Series/28083, http://www.nber.org/papers/w28083 (accessed on 5 March 2021).

[26] Giupponi, G. and C. Landais (2018), “Subsidizing Labor Hoarding in Recessions: The Employment & Welfare Effects of Short Time Work”, CEPR Discussion Papers, https://ideas.repec.org/p/cpr/ceprdp/13310.html (accessed on 12 January 2021).

[40] Gourinchas, P. et al. (2020), COVID-19 and SME Failures, National Bureau of Economic Research, Cambridge, MA, http://dx.doi.org/10.3386/w27877.

[17] Granja, J. et al. (2020), Did the Paycheck Protection Program Hit the Target?, National Bureau of Economic Research, Cambridge, MA, http://dx.doi.org/10.3386/w27095.

[8] Hale, T. et al. (2020), Oxford COVID-19 Government Response Tracker, Blavatnik School of Government, University of Oxford, https://www.bsg.ox.ac.uk/research/research-projects/covid-19-government-response-tracker.

[22] Hensvik, L., T. Le Barbanchon and R. Rathelot (2021), “Job search during the COVID-19 crisis”, Journal of Public Economics, Vol. 194, p. 104349, http://dx.doi.org/10.1016/j.jpubeco.2020.104349.

[7] Hijzen, A. and S. Martin (2013), “The role of short-time work schemes during the global financial crisis and early recovery: a cross-country analysis”, IZA Journal of Labor Policy, Vol. 2/1, http://dx.doi.org/10.1186/2193-9004-2-5.

[29] Hijzen, A. and A. Salvatori (2020), “Designing fair and work-oriented unemployment benefits: The case of Belgium”, OECD Social, Employment and Migration Working Papers, No. 237, OECD Publishing, Paris, https://dx.doi.org/10.1787/ac17d171-en.

[4] Hijzen, A. and D. Venn (2011), “The Role of Short-Time Work Schemes during the 2008-09 Recession”, OECD Social, Employment and Migration Working Papers, No. 115, OECD Publishing, Paris, https://dx.doi.org/10.1787/5kgkd0bbwvxp-en.

[14] Hubbard, R. and M. Strain (2020), Has the Paycheck Protection Program Succeeded?, National Bureau of Economic Research, Cambridge, MA, http://dx.doi.org/10.3386/w28032.

[28] Kolsrud, J. et al. (2018), “The optimal timing of unemployment benefits: Theory and evidence from Sweden”, American Economic Review, Vol. 108/4-5, pp. 985-1033, http://dx.doi.org/10.1257/aer.20160816.

[36] Kopp, D. and M. Siegenthaler (2019), “Short-Time Work and Unemployment in and after the Great Recession”, KOF Working Papers, Vol. 462, http://dx.doi.org/10.3929/ethz-b-000359533.

[27] Lalive, R., C. Landais and J. Zweimüller (2015), “Market externalities of large unemployment insurance extension programs”, American Economic Review, Vol. 105/12, pp. 3564-3596, http://dx.doi.org/10.1257/aer.20131273.

[18] Landais, C., P. Michaillat and E. Saez (2018), “A Macroeconomic Approach to Optimal Unemployment Insurance: Applications”, American Economic Journal: Economic Policy forthcoming, http://econ.lse.ac.uk/staff/clandais/cgi-bin/Articles/aejpol_applications.pdf (accessed on 22 February 2018).

[33] OECD (2020), Increasing Adult Learning Participation: Learning from Successful Reforms, Getting Skills Right, OECD Publishing, Paris, https://dx.doi.org/10.1787/cf5d9c21-en.

[1] OECD (2020), “Job retention schemes during the COVID-19 lockdown and beyond”, OECD Policy Responses to Coronavirus (COVID-19), OECD Publishing, Paris, https://dx.doi.org/10.1787/0853ba1d-en.

[9] OECD (2020), OECD Employment Outlook 2020: Worker Security and the COVID-19 Crisis, OECD Publishing, Paris, https://dx.doi.org/10.1787/1686c758-en.

[34] OECD (2020), “The potential of online learning: Early lessons from the COVID-19 crisis”, OECD Policy Responses to Coronavirus (COVID-19), OECD Publishing, Paris, https://doi.org/10.1787/ee040002-en.

[35] OECD (2019), OECD Employment Outlook 2019: The Future of Work, OECD Publishing, Paris, https://dx.doi.org/10.1787/9ee00155-en.

[30] OECD (2018), “Back to work: Lessons from nine country case studies of policies to assist displaced workers”, in OECD Employment Outlook 2018, OECD Publishing, Paris, https://dx.doi.org/10.1787/empl_outlook-2018-8-en.

[46] OECD (2018), Good Jobs for All in a Changing World of Work: The OECD Jobs Strategy, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264308817-en.

[31] OECD (2018), “The role of collective bargaining systems for good labour market performance”, in OECD Employment Outlook 2018, OECD Publishing, Paris, https://dx.doi.org/10.1787/empl_outlook-2018-7-en.

[44] ONEM/RVA (2020), Rapport Annuel 2020.

[19] Petrongolo, B. and C. Pissarides (2001), Looking into the black box: A survey of the matching function, American Economic Association, http://dx.doi.org/10.1257/jel.39.2.390.

[21] Şahin, A. et al. (2014), “Mismatch unemployment”, American Economic Review, Vol. 104/11, pp. 3529-3564, http://dx.doi.org/10.1257/aer.104.11.3529.

[41] Sharpe, S. (1994), “Financial Market Imperfections, Firm Leverage, and the Cyclicality of Employment”, American Economic Review, Vol. 84/4, pp. 1060-1074, https://ideas.repec.org/a/aea/aecrev/v84y1994i4p1060-74.html (accessed on 7 October 2019).

[5] The Global Deal for Decent Work and Inclusive Growth (2020), Social Dialogue, Skills and COVID-19, http://www.theglobaldeal.com/social-dialogue-skills-and-covid-19.pdf.

[25] Tilly, J. et al. (2017), Employment and Welfare Effects of Short-Time Work, https://jtilly.io/short-time-work.pdf.

[42] Van Audenrode, M. (1994), “Short-Time Compensation, Job Security, and Employment Contracts: Evidence from Selected OECD Countries”, Journal of Political Economy, Vol. 102/1, pp. 76-102, http://www.jstor.org/stable/2138794.

[47] Viviano, E. (2020), Alcune stime preliminari degli effetti delle misure di sostegno sul mercato del lavoro, Banca d’Italia, https://www.bancaditalia.it/pubblicazioni/note-covid-19/2020/Nota-Covid-19.11.2020.pdf.