An Assessment of Job Retention Support During the COVID‑19 Pandemic and its Aftermath in Belgium

3. The Belgian job retention scheme in OECD perspective

Abstract

Based on a new synthetic indicator of de jure job retention support, this chapter systematically compares of the job retention scheme in Belgium with that of other OECD countries and explores which design features drove job retention support take-up during the COVID-19 pandemic. It further evaluates how effective job retention schemes were in preserving employment relationships during the COVID‑19 pandemic.

In Brief

The Belgian job retention scheme in OECD perspective

Using a newly developed synthetic indicator of effective generosity of job retention support, Belgium’s job retention scheme ranked in the lower third of OECD countries during the peak of the COVID‑19 pandemic. The government covered 57% of labour costs for hours not worked at average wages, though a relatively low benefit cap resulted in higher effective income replacement rates for lower-wage workers.

Unlike most other countries that reduced scheme generosity as their economies recovered or phased out newly established programmes entirely, Belgium maintained its expanded provisions largely unchanged until July 2022, with full return to pre‑pandemic conditions only by July 2023. This meant that by end of 2022, Belgium operated one of the most generous and widely used job retention schemes across the OECD.

Cross-country analysis shows that containment stringency influenced job retention scheme take‑up more strongly than effective generosity during the COVID‑19 pandemic. Nevertheless, amongst design features, the cost to firms for hours not worked proved a particularly important factor in driving take‑up, with Belgium’s absence of direct co-financing requirements potentially contributing to persistent usage patterns.

Employment analysis across European OECD countries demonstrates that job retention schemes successfully averted considerable employment losses, particularly amongst occupations that could not be performed remotely. Without these schemes, employment would have declined by approximately 8% on average at the crisis peak. Belgium’s scheme was particularly effective, averting an estimated 12.9% employment loss. However, the scheme’s persistence beyond the acute crisis period resulted in small negative employment effects, as continued usage potentially hindered normal labour market adjustment processes.

3.1. Introduction

The COVID‑19 pandemic prompted an unprecedented expansion of job retention schemes across OECD countries, with governments rapidly scaling up existing programmes or introducing entirely new schemes to preserve employment relationships during widespread economic disruptions. However, these schemes varied considerably in their design, generosity, and usage patterns (DG EMPL, 2025[1]). This chapter places Belgium’s job retention scheme within the broader OECD context by developing a new synthetic indicator of effective generosity combining three key dimensions: eligibility (the share of workforce potentially covered), work-sharing arrangements (maximum permissible reduction in working hours), and financial generosity (government subsidy for hours not worked). It then systematically compares Belgium to other countries across all dimensions of this new indicator.

This chapter further investigates the determinants of scheme take‑up through cross-country regressions that relate usage rates to both the stringency of COVID‑19 containment measures and the effective generosity of job retention support. As such, it seeks to identify whether differences in take‑up reflect economic necessity driven by lockdown restrictions or institutional factors related to the design of job retention schemes.

Lastly, the chapter evaluates how effective job retention schemes were in preserving employment relationships during the COVID‑19 pandemic. The analysis exploits variation in scheme generosity across countries and time, as well as differences in teleworkability across occupations within countries. This methodology enables to identify employment effects by comparing outcomes in occupations most and least exposed to containment measures across countries with varying levels of job retention support.

3.2. The Belgian job retention scheme in OECD perspective

3.2.1. A new synthetic indicator of job retention support

This section situates the Belgian job retention scheme in OECD perspective based on a new synthetic indicator of the de jure level of support provided by job retention schemes. The indicator consists of three dimensions: i) eligibility expressed in terms of the number of workers that is potentially covered; ii) the share of working hours per worker that is potentially supported as measured by the maximum permissible reduction in working hours; and iii) the generosity of support by the government for hours not worked. Together these dimensions provide an indication of the maximum share of the overall wage bill in a country that can be subsidised by governments through job retention schemes. The indicator is available for each month from January 2020 to December 2022 for 27 OECD countries. The indicator relies as much as possible on official information based on OECD questionnaires completed by member states and the OECD Taxes and Benefits Model for the calculation of the generosity component. For more details on the construction of the synthetic indicator, see Box 3.1.

Box 3.1. A new synthetic indicator of the de jure effective generosity of job retention support

The new synthetic indicator of the de jure effective generosity of job retention support has three dimensions related to eligibility, work-sharing and generosity. Formally, effective generosity (EG) is measured as follows:

Equation 3.1

$EG= E \text{*} H \text{*} G.$

where E refers to the share of the workforce that is eligible for support, H the share of working hours per workers that is potentially support and G the generosity of support for hours not worked.

Eligibility (E) is measured in terms of potential take‑up as a share of private‑sector dependent employment (0‑1). It includes restrictions on workers like minimum contribution requirements for unemployment benefits (tenure) and contract status (temporary contracts) and restrictions with respect to the size or sector of firms. The eligibility component excludes restrictions related to the economic or financial performance of firms.

Work-sharing (H) is measured as the share of working hours per worker that is potentially supported, as defined by the maximum permissible reduction in working hours (0‑1). While most countries do not impose any such restrictions, some have used limits to promote work-sharing. The work-sharing component excludes restrictions to promote work-sharing related to the minimum number of workers in the firm that is covered, since this does not constrain potential coverage.

Generosity (G) is expressed as the cost of hours not worked for governments, as a share of labour costs, evaluated at the average wage at the maximum permissible reduction in working time (usually a complete work stoppage) (0‑1). It uses information on (gross) policy replacement rates, caps and ceilings, and requirements for paying employer social security contributions for hours not worked. The methodology is fully consistent with the OECD Tax and Benefits Model (OECD, 2024[2]). It is assumed that suspensions do not affect worker entitlements to social security and that contributions are paid either by governments or firms.

The resulting indicator measures the generosity of job retention support from a scale from 0 to 100. Countries without job retention schemes necessarily are assigned a value of zero. For countries with job retention schemes, it provides an indication of the maximum share of the wage bill that can be subsidised by governments through job retention support. The indicator therefore naturally accommodates job retention schemes in the form of wage subsidies and job retention programmes.

The information on each of the three dimensions is collected at monthly frequency for the period from January 2020 to December 2022 for 27 OECD countries.¹ Policy information is obtained through two ad hoc questionnaires to OECD delegates from member countries: the OECD Policy Questionnaire on Working Time Regulation and Short-Time Work Schemes which collected information for January 2020 and May/June 2020 and the OECD Questionnaire on Policy Responses to the COVID‑19 Crisis, which collected information for November 2021. Information for the remaining months is obtained through desk research. Labour force surveys are used to translate eligibility rules into potential coverage rates.

The new synthetic indicator of the de jure generosity of job retention support builds on previous work by Calligaris et al. (2023[3]) who developed a similar indicator that takes account of generosity and eligibility for 12 OECD countries for 2020-2021. The present indicator extends coverage to all European OECD countries as well as the United States, and measures policy changes during the entire COVID‑19 pandemic and beyond. Moreover, it relies on information validated by OECD Member countries through official OECD questionnaires and the policy rules recorded in the OECD Tax and Benefits Model. Finally, it offers a tractable way to account for the effective generosity of job retention support by focussing on easily measurable factors, without the need for strong assumptions (e.g. weights, functional form) when translating eligibility into potential coverage.

1. This includes Belgium, Switzerland, Czechia, Germany, Denmark, Estonia, Greece, Spain, Finland, France, Hungary, Iceland, Ireland, Italy, Lithuania, Luxembourg, Latvia, the Netherlands, Norway, Poland, Portugal, Sweden, Slovenia, the Slovak Republic, the United Kingdom, and the United States.

3.2.2. The overall degree of job retention support

While Belgium scaled up support in response to the pandemic similarly quickly as many other OECD countries with pre‑existing schemes, support in Belgium was much more persistent. In Belgium as well as in other countries with pre‑existing schemes, the effective generosity of support increased from about 30% of normal labour costs before the crisis to about 60% in March-May 2020 (Figure 3.1 and Figure 3.2). Countries that introduced entirely new schemes in response to the crisis provided similar levels of generosity initially. However, while the level of support in Belgium remained broadly constant until the end of 2022, support was gradually reduced in other countries as economic activity was resuming. In countries with permanent schemes, support was gradually withdrawn from the middle of 2021 onwards, reaching 40% by the end of 2022. In countries with temporary schemes, support was withdrawn even more quickly, with a first stepdown after the first wave of the virus in the summer of 2020 and further declines starting from the summer of 2021.1 As a result of these adjustments, the average effective generosity of support across countries was only marginally higher by the end of 2022 than before the pandemic.

Figure 3.1. OECD countries quickly scaled up the generosity of job retention support

Effective generosity of job retention support (% of normal labour costs) and average monthly containment stringency

Note: Effective generosity measures generosity, eligibility and the maximum permissible reduction in working hours (see Box 3.1). Permanent schemes refer to those in place before the COVID‑19 pandemic and still operational in December 2022. The average monthly containment stringency refers to the average of the index for the stringency of containment and closure policies as recorded in the Oxford COVID‑19 Government Response Tracker (OxCGRT) (Hale et al., 2021[4]).

Source: OECD calculations based on national JRS regimes and the Oxford COVID‑19 Government Response Tracker (OxCGRT).

Figure 3.2. Many OECD countries scaled back their job retention scheme by the end of 2022

Effective generosity of job retention support, percentage of normal labour costs

Note: Effective generosity measures (generosity, eligibility and the maximum permissible reduction in working hours (see Box 3.1).

† Countries in which job retention support was temporary. Before refers to January 2020.

Source: OECD calculations based on national JRS regimes.

The remainder of this section discusses the different dimensions of the effective generosity of job retention support in terms of eligibility, work-sharing and generosity.

3.2.3. Eligibility

Restrictions on firm eligibility are unlikely to have played much of a role in explaining the use of job retention schemes during the COVID‑19 crisis (OECD, 2022[5]; 2021[6]).2 While job retention schemes in OECD typically exclude public-sector firms, as they are less susceptible to economic shocks and operate under different financial constraints compared to private enterprises, restrictions for private sector firms in terms of firm size or economic activity are rare. Before the COVID‑19 crisis, Italy was an exception, limiting job retention support eligibility to firms with a minimum number of 15 employees until 2015 and 5 employees until the start of the pandemic, but removed size restrictions in response to the pandemic. Indeed, throughout the period from January 2020 to December 2022, no OECD country restricted eligibility in terms of size or sector.3 Eligibility requirements related to the economic need for support tended to be modest and have been weakened further in response to the COVID‑19 crisis. When they existed, they typically took the form of reductions in sales or working hours and most likely played a limited role in explaining take‑up during the COVID‑19 crisis. Since the role of such restrictions for potential take‑up is not easily quantifiable and their economic impact is likely to have been small, this is not taken into account in the eligibility component of the synthetic indicator of job retention support.

During the COVID‑19 crisis, eligibility for workers to job retention support was largely unrestricted as several countries temporarily lifted restrictions (OECD, 2022[5]). In normal times, worker eligibility tends to be restricted to insured workers, who meet the minimum contribution requirements for unemployment benefits, or those with open-ended contracts. The idea of such restrictions is to target support to workers with important firm-specific human capital that would be costly to rebuild after layoffs. For example, before the crisis, Germany limited eligibility for Kurzarbeit to insured workers with permanent contracts. In response to the COVID‑19 crisis, most OECD countries broadened worker eligibility to include temporary (e.g. Germany, Spain) and uninsured workers (e.g. Spain) or introduced new schemes without eligibility restrictions (e.g. Netherlands). In Belgium, the general applicability of the force majeure scheme effectively meant that worker eligibility was no longer restricted to insured workers as was the case under the regular scheme for economic difficulties. The loosening of eligibility requirements at the height of the pandemic, was a result of extending the rationale of job retention schemes from the classic “automatic stabilizer” to also contribute to the protection of public health by making a containment restrictions possible without incurring large or insurmountable economic pressures for workers and firms. Nevertheless, workers on short-term contracts, particularly many young people, often did not qualify for job retention support and frequently didn’t receive unemployment insurance benefits or minimum income support (OECD, 2023[7]).

Figure 3.3. Eligibility for job retention support was largely unrestricted during the pandemic

Potential coverage of job retention support, percentage of dependent private sector employment

Note: Potential coverage of job retention support takes account of restrictions on workers in terms of minimum contributions requirements for unemployment benefits (tenure) and contract status (temporary contracts) as well on firms in terms of size or sector. For the United States, it is assumed that entitlement to unemployment compensation, which is a pre‑requisite for participation in the Short-Time Compensation (STC) program, follows the covered employment rate (see FRED (2023[8])). This may overstate eligibility, due to specific state‑level eligibility requirements.

† Countries in which job retention support was temporary.

Source: OECD calculations based on national JRS regimes.

3.2.4. Work-sharing

Restrictions on the maximum permissible reduction in working time to promote work-sharing tend to be rare, but when they are present, they can considerably reduce the use of job retention support (OECD, 2021[6]). Work-sharing restrictions on the maximum permissible share of working hours were in place before the pandemic in a number of countries including Denmark, Ireland, Sweden and the United States. While most countries allowed for a full suspension at the height of the pandemic, the STC programme in the United States continued to restrict the programme to reduction of in working hours of at most 60% (Figure 3.4).4 For full suspensions, the United States instead mainly relied on temporary layoffs, that is layoffs with a high probability of recall without preservation of the contract, while providing support to such workers through unemployment benefits. Right after the first wave of the pandemic, Greece introduced the Syn-Ergasia mechanism, under which employers could reduce weekly working hours of their employees by at most 50%. Alternative ways of promoting work-sharing relate to the minimum share of workers or entire units that participate in job retention schemes. For example, Germany requires a minimum share of workers to participate in job retention, while France used to require firms to apply job retention support equally within units before introducing more flexibility in response to the crisis. These additional forms of work-sharing are not taken into account in the synthetic indicator of effective job retention support as they do not restrict potential coverage. Belgium did not impose any requirements to promote on work-sharing at any time.

Figure 3.4. Most countries permitted a full reduction in working hours at the height of the pandemic

Maximum permissible reduction in working hours, percentage of usual working hours

Note: † Countries in which job retention support was temporary.

Source: OECD calculations based on national JRS regimes.

3.2.5. Generosity

Most of the variation across countries and over time in the synthetic indicator of job retention support comes from differences in the generosity component (Figure 3.5). For the purposes of the indicator, the generosity of support for hours not worked is evaluated at the average wage for the maximum permissible reduction in working time. The costs of hours not worked that are not subsidised by the government are either borne by firms (sometimes referred to as co-financing) or workers, resulting in lower earnings. On average across OECD countries, the share of labour costs for hours not worked subsidised by governments increased from about 30% before the pandemic to almost 70% at the peak of the crisis before reverting back to 30% by the end of 2022. There is considerable variation in the generosity of job retention support across countries, ranging from 114% of labour costs in the United States, where job retention support was temporarily increased through lump-sum subsidies (more below), to 25% in Poland. With 57% of labour costs for hours not worked financed by the government, the generosity of job retention support in Belgium is among the lower third of OECD countries during the peak of the crisis, but well above average before and after the COVID‑19 pandemic.5

Figure 3.5. The generosity of subsidies for hours not worked varies greatly across countries

Costs of job retention support for hours not worked borne by the government, percentage of normal labour costs at average wage for the maximum permissible reduction in working time

Note: It is generally assumed that employer social security contributions for hours are not worked that are not paid by employers are by paid by the government so that work entitlements to social security (e.g. pensions) are not affected by the time spent on job retention support. † Countries in which job retention support was temporary.

Source: OECD calculations based on national JRS regimes.

Cost for workers

The replacement rate for hours not worked for workers is important for smoothing consumption over periods of reduced working time and alleviating financial hardship, particularly in low-income households. In most countries, the replacement rate for hours not worked is either similar or higher than for unemployment benefits (OECD, 2021[6]). This means that workers on job retention scheme tend to be better off than workers on regular benefits because they combine full earnings for hours worked and higher replacement earnings for hours not worked. Moreover, and like unemployment benefits, replacement rates for hours not worked tend to be targeted at low-wage workers due to the use of benefit caps.

Differences in benefit caps across countries generate important differences in the country ranking of effective replacement rates and this is particularly relevant for Belgium (Figure 3.6). On average across countries, the replacement rate for hours not worked of average‑wage workers was 71% at the peak of the crisis (about 34% before the crisis and at the end of 2022), while it was 81% for workers at 67% of the average wage. By comparison, in Belgium it was 45% for average‑wage workers and 74% for low-wage workers. The presence of a low benefit cap in Belgium drives a large difference between the replacement rate at 100% and 67% of the average wage. It also implies that the generosity of support for workers in Belgium is closer to the average among OECD countries when focussing on low-wage workers.

Figure 3.6. Support for workers on reduced working hours is targeted to low-wage workers, particularly in Belgium

Effective replacement rates of hours not worked in the case of the full suspension*

Note: The effective replacement rate refers to wage replacement for average wage earners as a percentage of previous earnings, after accounting for caps and ceilings on the disbursed benefit.

† Countries in which job retention support was temporary.

Source: OECD calculations based on national JRS regimes.

The time variation in effective replacement rates for hours not worked mainly reflects the introduction (and termination) of temporary schemes in response to the pandemic, put in place either as a complement to pre‑existing schemes (e.g. Denmark, Ireland or the Netherlands) or because no pre‑existing scheme was available (e.g. Australia, Slovenia, the United Kingdom, New Zealand). Adjustments in replacement rates over time within a given scheme were relatively rare and typically quite small. The most notable change was in the United States, where a weekly lump-sum of initially USD 600 was paid temporarily irrespective of the reduction in working time, resulting in a replacement rate of more than 100% during its operation.

Cost to firms

The cost to firms for hours not worked determines the extent to which firms can reduce labour costs in line with declining business activity by reducing working time and hence the number of jobs at risk of being terminated as a result of acute liquidity problems in firms. The cost for firms is also an important parameter for modulating the level of support to firms over the course of a crisis to jobs that are temporarily at risk but remain viable in the medium term.

During the early stage of the COVID‑19 crisis, most countries set the cost of contractual hours not actually worked to zero, allowing firms to adjust labour costs in line with the decline in working time (Figure 3.7). For example, several countries eased the burden on firms, by reducing (e.g. Norway, Spain, Sweden) or removing (e.g. Austria, France, Germany, Italy) direct employer contributions for hours not worked. In some countries, employers have continued to bear some of the cost of idle workers. In Denmark and the Netherlands, employers are required to contribute respectively 25% and 10% of regular labour costs to ensure no change in income for workers. The schemes in Estonia, Portugal and Poland do not fully protect worker’s income, but still require employers to pay part of the income of workers on zero hours, i.e. who are temporarily not working. However, even in these countries JR schemes allowed for significant adjustments of labour costs during the crisis.

Figure 3.7. Belgium was not alone without employer contributions during the COVID‑19 pandemic

The cost to employers of hours not worked in the case of the full suspension of a worker at the average wage

Note: † Countries in which job retention support was temporary.

Source: Calculations based on national JRS regimes.

As lockdown restrictions were withdrawn and economic activity could resume several countries increased or reintroduced requirement for firms to share in the cost of not ours not worked subject to job retention support. Such co-financing requirements help ensure that support is used for jobs temporarily at risk that are likely to return after the crisis instead of jobs that have become permanently unviable. The re‑emergence of co-financing reflected concerns that as economic activity was allowed to resume job retention support was increasingly used to support jobs that had become unviable in firms with structural difficulties. This not only adds to the fiscal burden of these schemes but may also impede the economic recovery by delaying the reallocation of workers from less productive to more productive firms and worsening labour shortages (OECD, 2021[6]).

In Belgium, direct co-financing requirements play only a limited role, and if they exist, they tend to be regulated by collective bargaining rather than by the government. For example, in the case of temporary unemployment for economic reasons for white‑collar workers collective agreements typically require firms to pay additional supplements to workers on reduced working time. Since these supplements are not government-imposed and vary by collective agreement, they are not taken into account in the synthetic indicator. However, in the case of temporary unemployment for economic reasons for blue‑collar workers, employers are required to pay a “responsibility contribution” that is triggered when a worker is placed on job retention support for an extended period of time (Box 3.2). The main intention of this contribution is to promote work-sharing after prolonged use rather than to target support to jobs temporarily at risk. Since the contribution only kicks in after 110 days it is not taken into account in the synthetic indicator.

Box 3.2. The responsibility contribution

Temporary unemployment for economic reasons for blue‑collar workers includes a “responsibility contribution” that is triggered when a worker is on temporary unemployment for more than 110 days over the preceding three‑quarters. The precise contribution depends on the total number of days of temporary unemployment over the past three‑quarters as well as during the current quarter. When the total number of days ranges between 110 and 130 days, it is EUR 20 per day, while it increases to EUR 100 per day if it is more than 200 days. This system allows firms to reduce their responsibility contributions by distributing the days of temporary unemployment more evenly over different quarters and workers through work-sharing. The responsibility contribution may be reduced or waived in specific circumstances.^1,2

To illustrate how the responsibility contributions promote work-sharing, Figure 3.8 compares the contribution that is due under different scenarios of using temporary unemployment for economic reasons in a hypothetical firm with two blue‑collar workers. In a first scenario, only one worker is placed on temporary unemployment for 240 days, costing the firm EUR 8 800 of responsibility contribution in total. If the 240 days are instead shared between two workers, the firm pays as little as EUR 400 (i.e. EUR 200 per worker). In another scenario, one worker is placed on temporary unemployment for 360 days, costing the firm EUR 20 800. If these 360 days are equally shared between the two workers, the firm’s contribution is reduced to EUR 6 400 (i.e. EUR 3 200 for each worker). These illustrative simulations highlight that there are strong incentives for work sharing if blue‑collar workers are placed on temporary unemployment for longer periods of time.

Figure 3.8. The responsibility contribution provides strong incentives for work sharing

Scenarios of the effective responsibility contribution for blue‑collar workers placed on temporary unemployment for economic reasons

1. Firms in particular difficulties can apply for a 50% reduction on the contribution upon a request to the General Directorate of Collective Labor Relations at the FPS Employment, Labour, and Social Dialogue. A royal decree may also provide a temporary exemption of the contributions for entire sectors under particular risks, or for all firms when the economy faces exceptional circumstances.

2. Temporary unemployment for economic reasons in construction prescribes a contribution of EUR 46.31 for each day of temporary unemployment that exceeds 110 days in the previous calendar year.

Source: OECD calculations.

3.3. Determinants of job retention scheme take‑up

The previous section has shown that the extent of containment measures as well as the generosity of job retention schemes evolved rapidly through the COVID‑19 pandemic and its aftermath. To identify to what extent both factors drove the actual take‑up of job retention schemes, this section presents an empirical analysis of take‑up across countries.

3.3.1. Take up during the COVID‑19 crisis has been unprecedented

During the COVID‑19 pandemic, take‑up of job retention schemes reached unprecedented levels, even though it differed widely across OECD countries (Figure 3.9). From negligible levels, OECD-wide take‑up increased to a peak of almost 20% in May 2020, supporting approximately 60 million jobs – significantly more than during the global financial crisis (OECD, 2021[6]) (see also Box 3.3). Countries with general short time work and wage subsidy schemes saw higher take‑up rates, often exceeding 30%, while those only allowing for partial reductions in working time, such as the United States, had lower rates and relied more on unemployment benefits (OECD, 2021[6]). In Belgium, take‑up increased from about 2% before the crisis to 26% in May 2020. By the end of the COVID‑19 in December 2022, take‑up of job retention schemes had largely returned to its pre‑COVID level in all countries. In Belgium, take‑up was highest, at 2%, similar to its level before the crisis, compared with 0.2% for the OECD as a whole.

Figure 3.9. Take‑up of job retention schemes differed widely across the OECD

Take‑up of job retention schemes, percentage of dependent employment

Source: OECD calculations based on national sources.

Box 3.3. The use of job retention support during the global financial crisis and the COVID‑19 crisis in Belgium and selected countries

This box documents the evolution in the use of job retention schemes since before the start of the global financial crisis to the most recent period for which data are available (early 2025) for four selected countries with pre‑existing job retention schemes, i.e. Belgium, France, Germany and Italy (Figure 3.10). This provides the following insights:

• Job retention support saw a significantly higher peak during the COVID‑19 pandemic than during the GFC. On average, across the four countries examined (France, Germany, Belgium, and Italy), the peak was seven times greater during the pandemic. France is a notable example, as its programme played a minor role during the GFC but was greatly simplified and expanded for the COVID‑19 crisis.

• Following the GFC, the use of job retention support persisted longer in some countries than others. For instance, France and Germany saw a return to pre‑crisis levels within two years, while Belgium and Italy experienced a more protracted use. After the COVID‑19 pandemic, take‑up had returned to or was below pre‑crisis levels in all countries by December 2022, except for Belgium.

• In Belgium, take‑up of job retention support remained elevated through early 2025, with strong upticks in 2023 and 2024. This extended use may be due to the prolonged maintenance of simplified job retention provisions related to COVID‑19, the use of schemes for energy-related reasons in late 2022 as well as the renewed use of job retention support in manufacturing due to declining economic activity (ONEM, 2024[9]).

Figure 3.10. Take‑up of job retention schemes in selected OECD countries

Percentage of dependent employment

Note: Italy: Data before 2018 are based on the number of authorised hours (estimated number of employees using the ratio of total hours authorised under the quarterly average hours worked by employee) and spliced using the actual number of participants from January 2018. Data from mid-2023 is not available. The absence of a persistent pattern following the COVID‑19 in Italy reflects the role of reforms to its job retention schemes. The Jobs Act of 2015 shortened the maximum duration of support, particularly for workers on zero hours, and increased the use of co-financing by firms for the costs of hours not worked.

Source: OECD calculations based on national sources.

3.3.2. Take‑up mainly depends on economic conditions but also on the design of job retention support

As laid out in the previous sections, both the institutional design and the level of take‑up of job retention schemes differ widely across OECD countries. Figure 3.11 shows that there was a modest correlation between the effective generosity and the take‑up of job retention schemes across countries, especially in 2020. Indeed, countries with more generous schemes, saw higher take‑up rates throughout the pandemic- and immediate post-pandemic period. However, the relationship became weaker as the years progressed and take‑up reduced considerably, effectively vanishing in 2022. Nevertheless, the precise relationship between effective generosity and take‑up may also be tainted by the stringency of containment measures as well as the prevailing economic conditions, both of which may influence effective generosity and take‑up of job retention schemes.

A natural question is therefore whether the institutional design of job retention schemes is related to the use of take‑up and its persistence, after controlling for the stringency of COVID‑19 containment measures, and which of the design features are the most critical ones. To this end, this section presents results from cross-country panel regressions, relating job retention scheme take‑up to various dimensions of policy generosity and its components as well as the stringency of COVID‑19 containment measures (see Hale et al. (2021[4])). Box 3.4 lays out the details of the methodological approach. The results based on equation Equation 3.2 are presented in Table 3.1 for the various models and visually in Figure 3.12.

Figure 3.11. Take‑up was strongly correlated with effective generosity during the pandemic

Take‑up (% of dependent employment) and effective generosity of job retention schemes (%) across countries, annual average

Note: The 2020 average refers to the March-December period only. Data includes Belgium, Switzerland, Czechia, Germany, Denmark, Estonia, Greece, Spain, Finland, France, Hungary, Ireland, Italy, Lithuania, Luxembourg, Latvia, the Netherlands, Norway, Poland, Portugal, Sweden, Slovenia, the Slovak Republic, the United Kingdom and the United States.

Source: OECD calculations based on national sources.

Box 3.4. Empirical model to estimate the determinants of job retention support take‑up

This analysis investigates the factors driving the uptake of job retention schemes during the COVID‑19 pandemic across countries. It utilises cross-country data on job retention scheme participation alongside key policy determinants, as measured by the effective generosity indicator detailed in Box 3.1.

The core regression model is run over the period between 2020 and 2022 and specified as:

Equation 3.2

${\mathrm{J}\mathrm{R}\mathrm{S}}_{\mathrm{c}\mathrm{t}}\mathrm{ }\mathrm{}=\mathrm{\alpha }\mathrm{}+{\theta X}_{ct}+\mathrm{\mu }\mathrm{}{\mathrm{P}}_{\mathrm{c}\mathrm{t}}+{\mathrm{\gamma }}_{\mathrm{t}}\mathrm{}+{\mathrm{ϵ}}_{\mathrm{c}\mathrm{t}}$

where, c denotes countries and t denotes months. The dependent variable, ${\mathrm{J}\mathrm{R}\mathrm{S}}_{\mathrm{c}\mathrm{t}}$, represents the percentage of dependent employees on job retention support in a given month. $X_{ct}$​ is the stringency of health-related economic restrictions during the COVID‑19 pandemic, obtained from the Oxford COVID‑19 Government Response Tracker (Hale et al., 2021[4]), including mandated workplace closures and limitations on travel, amongst others. The effective generosity of job retention support, or its sub-components from Box 3.1, is captured by ${\mathrm{P}}_{\mathrm{c}\mathrm{t}}$​. Common trends across countries over time are accounted for through the time fixed effects in ${\mathrm{\gamma }}_{\mathrm{t}}\mathrm{}$. Robust standard errors are calculated and clustered by country.

The analysis is conducted by running a series of eight different models on data for 25 European OECD countries1 to explore the influence of various policy combinations on job retention scheme take‑up. Iceland is excluded from the analysis due to a lack of data on job retention scheme take‑up. The United States are excluded as data refer to short-time compensation benefits only, while many workers were instead covered by the Paycheck Protection Program (PPP).

1. This includes Austria, Belgium, Switzerland, Czechia, Germany, Denmark, Estonia, Greece, Spain, Finland, France, Hungary, Ireland, Italy, Lithuania, Luxembourg, Latvia, the Netherlands, Norway, Poland, Portugal, Sweden, Slovenia, the Slovak Republic, and the United Kingdom.

Take‑up of job retention support depends mainly on the ability to continue working as measured by the stringency of containment measures but also to some extent on the design of job retention. A 10‑percentage points (p.p.) increase in effective generosity is associated with a 0.2 p.p. increase in take‑up (Model 1). Looking at the different components of the synthetic indicator suggests that the financial generosity of support is particularly important, while worker eligibility also plays a role. The absence of restrictions on work-sharing (maximum permissible reduction in working hours) either plays no role for take‑up or may even reduce it (Model 2‑5). Looking at the generosity of support for firms and workers, both the replacement rate for workers and the cost to firms for hours not worked play a role. Differences in the costs to firms for hours not worked appear to be particularly important, as a 10‑p.p. increase in these costs is associated with a 1 p.p. decrease in take‑up (Models 6‑8). The latter is consistent with the focus of policymakers on co-financing for the modulation of support over the course of a crisis. Overall, Figure 3.12 shows that over the COVID‑19 crisis, containment measures contributed considerably more to take‑up than the generosity of job retention support.

Table 3.1. Regression analysis of job retention scheme take‑up

Regression of monthly JRS take‑up on JRS policies and COVID‑19 containment stringency, 2020 M1‑ 2022 M12

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Containment stringency	0.132*** (0.018)	0.123*** (0.018)	0.136*** (0.018)	0.135*** (0.018)	0.126*** (0.018)	0.134*** (0.018)	0.131*** (0.018)	0.127*** (0.018)
Effective generosity (%)	0.022*** (0.006)
Costs borne by government (%)		0.028*** (0.005)			0.048*** (0.009)
Workers covered (%)			0.008* (0.004)		0.033*** (0.007)			0.033*** (0.007)
Maximum reduction in hours (%)				0.005 (0.003)	‑0.046*** (0.008)			‑0.033*** (0.008)
Replacement rate (%)						0.012** (0.005)		0.038*** (0.009)
Costs borne by employers (%)							‑0.042*** (0.012)	‑0.101*** (0.015)
Constant	‑0.017* (0.007)	‑0.018** (0.007)	‑0.016* (0.007)	‑0.013 (0.007)	‑0.016* (0.007)	‑0.016* (0.007)	‑0.005 (0.007)	‑0.015* (0.007)
Observations Adjusted R-squared	900 0.553	900 0.559	900 0.548	900 0.547	900 0.567	900 0.549	900 0.551	900 0.568

Note: Each regression controls for year by quarter fixed effects. The average monthly containment stringency refers to the average of the index for the stringency of containment and closure policies as recorded in the Oxford COVID‑19 Government Response Tracker (OxCGRT) (Hale et al., 2021[4]). Regressions are run over data on Austria, Belgium, Switzerland, Czechia, Germany, Denmark, Estonia, Greece, Spain, Finland, France, Hungary, Ireland, Italy, Lithuania, Luxembourg, Latvia, the Netherlands, Norway, Poland, Portugal, Sweden, Slovenia, the Slovak Republic, and the United Kingdom. The United States and Iceland are excluded from this exercise.

Source: OECD calculations based on national sources.

Figure 3.12. Cross-country differences in take‑up are mainly explained by the stringency of lockdown measures

Components of take‑up explained by containment measures and the overall generosity of support

Note: The figure presents the contribution of the policy and containment components in Model 8 of Table 2. Each of the two components is based on a partial prediction over the 2020 to 2022 period. The figure also shows average actual take‑up over the same period as a reference. The difference between actual and predicted take‑up is accounted for by the time‑fixed effects and the residual in Model 8 of Table 2.

Source: OECD calculations based on national sources.

3.4. Employment effects of job retention support

The COVID‑19 pandemic had potentially devastating effects on employment given widespread restrictions on economic activity and travel. Job retention schemes were the instrument of choice of governments to protect workers and firms against the fall out of the pandemic by allowing firms to adjust labour costs in line with economic activity, while supporting the earnings of workers on reduced work schedules. The objective of this section is to assess the effectiveness of job retention schemes during the COVID‑19 pandemic in supporting employment. More precisely, it analyses how employment evolved in response to the stringency of lockdown measures and how this depended on the degree of job retention support. Moreover, the analysis is conducted separately for teleworkable and non-teleworkable occupations to assess whether the role of job retention schemes was more important in occupations where the scope for teleworking was more limited. The analysis is implemented using quarterly data from the European Labour Force Surveys with the information on employment dynamics by detailed occupation. Box 3.5 lays out the empirical approach used to estimate the role of job retention schemes for employment dynamics during the COVID‑19 pandemic.

Box 3.5. Estimating the effect of job retention schemes on employment during the COVID‑19 pandemic

The impact of job retention schemes on employment is analysed using a triple difference‑in-differences approach that compares the employment response to a change in lockdown restrictions across countries that differ in the effective generosity of job retention support and across occupations that differ in their teleworkability. The employment response to a change in lockdown restrictions is expected to be larger in countries with more generous job retention support in occupations where telework is not feasible. Calligaris et al. (2023[3]) use a similar approach but instead of focussing on occupations directly, they focus on differences in the share of teleworkable occupations across 2‑digit industries.

The employment response to changes in containment stringency is characterised by means of an impulse response function estimated using the local projection method following Jordà (2005[10]). This allows for the robust estimation of the impulse response function by estimating its coefficients directly for each period rather than from a specific dynamic model, which can be sensitive to misspecification. The role of job retention support is analysed by interacting changes in the stringency of lockdown measures with the use of job retention support instrumented by effective generosity. The impulse response functions are estimated separately for teleworkable and non-teleworkable occupations.

Formally, the empirical model can be represented as follows:

Equation 3.3

$y_{oct+k}-y_{oct}=$

$\alpha + {{\beta }_k∆X}_{ct}+{\gamma }_k{\widehat{S}}_{ct}+{{\theta }_k∆X}_{ct}{\widehat{S}}_{ct}+\rho ∆y_{oct}+{\pi }_o+ {\mu }_c+{\vartheta }_t+{\epsilon }_{oct}$

$k=\{<!--\; -->1, 2, 3, 4\}$

where $y_{oct+k}-y_{oct}$is the change log employment in occupation o, which may be teleworkable or non-teleworkable, in country c between quarters t+k and t‑1. The change in the stringency of health-related economic restrictions is captured by ${∆X}_{ct}$. ${\widehat{S}}_{ct}$captures predicted take‑up of job retention support in dependent employment in the private sector as a function of its generosity following an extended and quarterly version of Model 8 in.1 Occupation fixed-effects enter as ${\pi }_o,$ country fixed-effects as ${\mu }_c$, and year times quarter fixed effects are captured by ${\vartheta }_t$. Finally, $∆y_{oct}$ is lagged employment growth. The impulse response function of employment growth with respect to changes in the stringency of COVID‑19‑related containment measures is evaluated for a scenario where job retention scheme take‑up is zero in the quarter of the change in severity as well as a scenario where job retention scheme take‑up was at the average level during the COVID‑19 pandemic. The impulse response functions are estimated over four quarters and weighted by occupation size.

The empirical model is estimated using data for 22 European OECD countries, with information for the period from 2020 to 2022.2 Quarterly employment growth by country and 3‑digit ISCO occupation is obtained from the EU Labour Force Survey. Information on the stringency of health-related economic restrictions is obtained from the Oxford COVID‑19 Government Response Tracker (Hale et al., 2021[4]). Occupations are ranked according to the intensity of telework in the period after the pandemic and key occupations are omitted (following the classification of Fasani and Mazza (2020[11]). The threshold for teleworkability is set to match the observed intensity of telework during the pandemic in 2020 on average across countries (almost 50% teleworked at least once). Figure A A.7 shows that the pre‑trends are reasonably flat and the differences between employment growth in teleworkable and non-teleworkable occupations prior to the COVID‑19 pandemic was mostly not statistically different.

1. To improve the take‑up prediction, the first stage model adds country fixed-effects and additional interactions of stringency with each generosity measure to Model 8 of Table 3.1. Further, while Table 3.1is based on monthly data, the analysis of employment dynamics is limited to quarterly level. As such, the first stage prediction is also estimated on the quarterly level. The model has an adjusted-R² of 0.84.

2. This includes Austria, Belgium, Switzerland, Czechia, Denmark, Estonia, Spain, Finland, France, Hungary, Ireland, Italy, Lithuania, Luxembourg, Latvia, the Netherlands, Norway, Poland, Portugal, Sweden, Slovenia, and the Slovak Republic. While part of the effective generosity indicators, Iceland, Germany, the United Kingdom and Greece are excluded from the analysis. For Iceland, information on JRS take‑up is lacking. For the other countries, employment levels are missing for a considerable amount of occupation-year pairs.

While most of the population was affected by the strict containment measures during the COVID‑19 pandemic, labour markets simultaneously saw a strong shift to telework, with many employees performing their work from home. However, in many occupations telework was not possible, so job retention schemes were necessary to shield their workers from losing their employment. As such, one would expect different employment dynamics for occupations with activities that are, in principle, able to be performed at home (i.e. teleworkable occupations) and those that require the presence of workers at the employer’s premises (i.e. non-teleworkable occupations), while job retention schemes should only have a discernible effect on non-teleworkable occupations. These expectations follow also directly from the aims of job retention support in preserving employment in exposed jobs. A simple event study in Figure A.A.7 also shows that employment growth was not only developing relatively similarly for teleworkable and non-teleworkable occupations before the COVID‑19 pandemic, but teleworkable occupations were initially also much less effected by the COVID‑19 shock than non-teleworkable ones. To illustrate the role job retention schemes had in cushioning this effect on the most exposed occupations, Figure 3.13 shows employment dynamics following changes in COVID‑19 containment stringency. These are separately computed for teleworkable and non-teleworkable occupations and linked to the prevalence of job retention support.

The results of this exercise show that job retention schemes can indeed cushion the impact of COVID‑19 containment measures on employment in occupations that are not teleworkable, and therefore more exposed to the containment restrictions (Figure 3.13, Panel A). In the absence of job retention scheme take‑up, employment decreases by about 2 p.p. in the quarter the stringency of COVID‑19 containment measures increased by one standard deviation, reaching a cumulative employment decline of just above 3.2 p.p. 3 quarters after the change in containment stringency. A one standard deviation increase in the stringency of containment measures across included countries does not reduce employment when there is job retention scheme take‑up in the initial quarters following a one standard deviation change in containment stringency.6

As expected, changes in the stringency of pandemic containment measures had only minor effects on occupations that are able to be performed from home (Figure 3.13, Panel B). This also only differed marginally if job retention schemes were used or not. For example, in the absence of job retention support, teleworkable occupations saw no statistically significant employment effects of a one standard deviation change in the stringency of COVID‑19 containment measures. With job retention support, there just a small positive as well as a small negative change in employment at different points over the observed horizon, without any clear meaningful patterns. Overall, these results suggest that job retention scheme were indeed successful in preserving employment in jobs exposed to COVID‑19 containment measures and had little to no effect on barely exposed occupations.

Figure 3.13. Job retention reduced the negative employment effects of the COVID‑19 pandemic

Effect of a tightening in the COVID‑19 containment stringency index on cumulative employment growth, 22 OECD-EU countries

Note: The lines represent the effect of a one standard deviation increase in the containment stringency index after the change for two JRS take‑up scenarios. The no take‑up scenario models partial predications of equation 2, where $S_{ct}$ is set to zero, while the average take‑up scenario sets $S_{ct}$ to the average predicted take‑up during the COVID‑19 pandemic. The dotted lines represent the 90% confidence interval around the estimates based on standard errors clustered at the country-occupation and country-time level. Panel A presents the effect of a change in the severity of the pandemic after the change estimated through Equation 2 for 2‑digit occupations that tend not to be teleworkable. Panel B presents the same effects for 2‑digit occupations that tend to be teleworkable. Teleworkability for 2‑digit ISCO occupations is defined based on the share of workers occasionally teleworking. The countries included in the regression are Austria, Belgium, Switzerland, Czechia, Denmark, Estonia, Spain, Finland, France, Hungary, Ireland, Italy, Lithuania, Luxembourg, Latvia, the Netherlands, Norway, Poland, Portugal, Sweden, Slovenia, and the Slovak Republic.

Source: OECD calculations based on national sources, the quarterly EU-LFS data, and the Oxford COVID‑19 Government Response Tracker (Hale et al., 2021[4]).

The estimated effect of job retention schemes in preserving employment when containment measures tightened suggests that these schemes were able to save jobs that would have otherwise been destroyed. In order to get a sense of the number of jobs saved during the peak periods of the COVID‑19 pandemic, Figure 3.14 presents estimates on the difference in employment levels in counterfactual situations with and without job retention schemes given the containment measures in the country and the use of job retention schemes based on the estimated model coefficients of the model laid out in Box 2.4 and presented in Figure 3.13. The results suggest that the use of job retention schemes at the peak of the COVID‑19 pandemic in the second quarter of 2020 averted the loss of 9.5% of employment in countries with permanent job retention schemes and 6.2% in countries with newly established and temporary schemes (8% on average across countries). These effects subsided quickly with a less drastic pandemic situation and lower containment stringency, turning even marginally negative in some countries. Overall, the findings suggest that the presence of established job retention schemes can help in allowing for a quick scaling up of job retention support use among affected firms, who may already have the relevant administrative familiarity with said schemes.7

In Belgium, where take‑up was particularly high, the use of job retention schemes averted losses of about 12.9% of employment at the peak of the pandemic. This is considerably higher than in many other countries, in line with the high take‑up of job retention schemes at the peak of the COVID‑19 pandemic. As the scheme saw no major adjustments through the end of 2021, the take‑up of job retention schemes was still considerably higher in Belgium well after containment measures were relaxed, which meant that their use had small negative effects relative to an absence of job retention scheme use.

Taking account of the level of take‑up in Belgium, for every 100 workers placed on job retention schemes at the peak of the crisis, 55 jobs were preserved (52 jobs across all countries). Abstracting from working time and assuming fulltime for each worker, this implies that upper bound estimates of the deadweight effects associated with jobs being supported that would have been retained anyway or could not be retained even with support, amount to about 45% (48% across all countries). This is relatively modest, especially as these estimates do not take intensity of work into account, and is comparable to earlier estimates that placed deadweight losses at about one‑third across countries during the Great Recession (Hijzen and Venn, 2011[12]; Hijzen and Martin, 2013[13]). However, these deadweight costs are higher than estimates of 25% for Spain during the COVID‑19 pandemic (OECD, 2024[14]).

Figure 3.14. Job retention schemes avoided a considerable amount of employment losses during the peak of the COVID‑19 pandemic, but became less efficient at later times

Jobs saved as a share of dependent employment, Q1 2020 to Q4 2021

Note: The jobs saved measure is the difference in counterfactual employment trajectories following country-specific changes in containment stringency under country-specific job retention scheme take‑up and in absence of such take‑up cumulated over three‑quarters following a change in containment stringency. Counterfactual employment trajectories are calculated using the estimated coefficients on job retention scheme take‑up, changes in containment stringency and their interaction, following the model laid out in Box 3.5. The countries included in these estimates are Austria, Belgium, Switzerland, Czechia, Denmark, Estonia, Spain, Finland, France, Hungary, Ireland, Italy, Lithuania, Luxembourg, Latvia, the Netherlands, Norway, Poland, Portugal, Sweden, Slovenia, and the Slovak Republic.

Source: OECD calculations based on national sources, the quarterly EU-LFS data, and the Oxford COVID‑19 Government Response Tracker (Hale et al., 2021[4]).

References

[3] Calligaris, S. et al. (2023), “Employment dynamics across firms during COVID-19: The role of job retention schemes”, OECD Economics Department Working Papers, No. 1788, OECD Publishing, Paris, https://doi.org/10.1787/33388537-en.

[1] DG EMPL (2025), Job Retention Schemes in Perspective: Lessons Learnt And, https://employment-social-affairs.ec.europa.eu/labour-market-and-wage-developments-europe_en.

[11] Fasani, F. and J. Mazza (2020), “Immigrant Key Workers: Their Contribution to Europe’s COVID-19 Response”, IZA Policy Paper, Vol. 155, https://www.iza.org/publications/pp/155/immigrant-key-workers-their-contribution-to-europes-covid-19-response.

[8] FRED (2023), The insured unemployment rate, https://fredblog.stlouisfed.org/2023/12/the-insured-unemployment-rate/.

[4] Hale, T. et al. (2021), “A global panel database of pandemic policies (Oxford COVID-19 Government Response Tracker”, Nature Human Behaviour, https://doi.org/10.1038/s41562-021-01079-8.

[13] 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, https://doi.org/10.1186/2193-9004-2-5.

[12] 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://doi.org/10.1787/5kgkd0bbwvxp-en.

[10] Jordà, Ò. (2005), “Estimation and Inference of Impulse Responses by Local Projections”, American Economic Review, Vol. 95/1, pp. 161-182, https://www.aeaweb.org/articles?id=10.1257/0002828053828518.

[14] OECD (2024), Preparing ERTE for the Future: An Evaluation of Job Retention Support in Spain During the COVID-19 Pandemic, OECD Publishing, Paris, https://doi.org/10.1787/a70bf8ec-en.

[2] OECD (2024), TaxBEN: The OECD tax - benefit simulation model - Methodology, user guide and policy applications, https://www.oecd.org/content/dam/oecd/en/topics/policy-sub-issues/income-support-redistribution-and-work-incentives/OECD-TaxBEN-methodology-and-manual.pdf.

[7] OECD (2023), Evaluation of Belgium’s COVID-19 Responses: Fostering Trust for a More Resilient Society, OECD Publishing, Paris, https://doi.org/10.1787/990b14aa-en.

[5] OECD (2022), Riding the waves: Adjusting job retention schemes through the COVID-19 crisis, https://www.oecd.org/content/dam/oecd/en/publications/reports/2022/03/riding-the-waves-adjusting-job-retention-schemes-through-the-covid-19-crisis_9c5e78fe/ae8f892f-en.pdf.

[6] OECD (2021), “Job retention schemes during the COVID‑19 crisis: Promoting job retention while supporting job creation”, in OECD Employment Outlook 2021: Navigating the COVID-19 Crisis and Recovery, OECD Publishing, Paris, https://doi.org/10.1787/c4c76f50-en.

[9] ONEM (2024), Le chômage temporaire après la crise du coronavirus, https://www.onem.be/file/cc73d96153bbd5448a56f19d925d05b1379c7f21/56b3ce93b30eb832bcabe25e6f5d83cdbf42ebd2/2024_11_21_etude_ct_post_corona_fr.pdf.

Notes

← 1.  For example, the job retention scheme in Estonia already ended before May 2021, while the schemes in Czech Republic, Denmark, Estonia, the United Kingdom, Greece, Hungary, Iceland, Lithuania, Latvia, the Netherlands, Poland, the Slovak Republic and Slovenia ended before June 2022.

← 2.  However, several countries effectively differentiated the degree of support across firms by restricting eligibility to additional support measures based on sector, firm size or firm performance. Belgium briefly restricted the use of the simplified coronavirus force majeure scheme to firms in severely impacted sectors in September 2020, while all other all firms continued to have access to temporary unemployment through other schemes (see OECD (2022[5]), for a discussion).

← 3.  While this report only considers the permanent Short-Time Compensation (STC) scheme for the Unites States, the US Government also introduced the temporary and restricted U.S. Paycheck Protection Program (PPP) between April 2020 and May 2021. The PPP was effectively restricted to small and medium-sized businesses with fewer than 500 employees and offered forgivable, low-interest loans so they could maintain their payrolls even without active work. Between April and June 2020, the PPP provided over USD 500 billion in support, amounting to about 2.5% of GDP (see OECD (2021[6])). Moreover, Canada Emergency Wage Subsidy (CEWS) programme was replaced in October 2021 by more targeted schemes focussed either on specific sectors such as tourisms and hospitality or the hardest hit firms.

← 4.  Under the Paycheck Protection Program (PPP), small- and medium-sized enterprises also had access to full temporary suspension (see OECD (2021[6])).

← 5.  Note that this includes the policy replacement rate for hours not worked subject to a cap and social security contributions for hours not worked usually paid by employers.

← 6.  In Belgium, containment stringency peaked at 81% during the period of the first lockdown, closely mirroring the average patterns across countries presented in Figure 3.13.

← 7.  At the same time, take‑up quickly increased to high levels even in some countries with newly established schemes in some countries, such as the Netherlands and Denmark.