Alexander Hijzen
Egbert Jongen
Mateo Montenegro
Preparing ERTE for the Future
An Evaluation of Job Retention Support in Spain During the COVID-19 Pandemic
Abstract
This chapter provides a first discussion of the possible effects of ERTE during the COVID‑19 crisis in preventing job losses. It does so by reviewing the empirical literature on the employment effects of job retention schemes and by analysing key descriptive statistics related to the evolution of aggregate economic outcomes and take‑up during the COVID‑19 crisis.
The available empirical evidence suggests that the JR schemes tend to be highly effective tools in preventing excessive job losses in times of crisis. Efficiency losses tend to be small at the start of a sharp downturn but tend to increase as the crisis persists, particularly when JR support continues to be provided during the economic recovery. It should be stressed, however, that most empirical evidence relates to the global financial crisis which was very different in nature from the COVID‑19 crisis (e.g. financial system impaired, more long-lasting) and the provision of JR support tended to be more targeted. As a result, considerable uncertainty remains around the impact of JR retention support during the COVID‑19 crisis.
Descriptive evidence suggests that job retention support is likely to have played a major role in preventing a surge in unemployment during the COVID‑19 crisis in Spain. While the reduction in employment was larger in Spain than in the benchmark countries (France, Germany and Italy), this reflects in part the larger decline in economic activity in Spain. The employment reduction in response to the decline in economic activity was much smaller than during the global financial crisis when the use of JR support was negligible. Indeed, the Spanish labour market responded in a fundamentally different way during the COVID‑19 crisis. Whereas during the global financial crisis adjustment mainly took the form of labour shedding, during the COVID‑19 crisis adjustments in working time were much more important. This is likely to reflect at least in part the strong reliance on JR support during the COVID‑19 crisis.
Based on descriptive evidence, it is however difficult to assess how many jobs were saved thanks to the JR support in Spain. If the reduction in hours had fully translated into reductions in employment in the absence of JR schemes, the fall in the number of employees might have been as large as 12% instead of the 5% decline observed in the second quarter of 2020. The actual effect would have been smaller if some firms had been able to retain some workers even without the help of JR schemes, either by relying on private financial resources (savings, credit) or public ones provided by the government to provide liquidity to firms (e.g. tax deferrals, loan guarantees).
This chapter provides a first discussion of the possible role of ERTE during the COVID‑19 crisis in preventing job losses. It does so by reviewing the empirical literature on the employment effects of job retention schemes and by analysing key descriptive statistics related to the evolution of aggregate economic outcomes and take‑up during the COVID‑19 crisis. As in Chapters 2 and 3, the descriptive analysis focuses on Spain as well as the three benchmark countries (France, Italy and Germany).
This section discusses the existing empirical evidence for the global financial crisis and the COVID‑19 crisis drawing on the review in Hijzen et al (2024, forthcoming[1]) Table 5.1 provides a summary of selected micro-level studies.
5.2.1. Studies on the global financial crisis and previous crisis episodes
Macro-level evidence
Several early studies have looked at the effects of job retention schemes using cross-country data, studying the recession of the 1980s and the global financial crisis which started in 2008 and was followed by the Eurocrisis in Europe. They typically find that employment declines less during economic crises in countries where the use of job retention schemes is high, while employment is similar in the medium term.
Abraham and Houseman (1994[2]) and Van Audenrode (1994[3]) show that during the 1970s and 1980s employment reacted more strongly and hours per person less strongly to similarly sized economic shocks in the United States than in European countries. They attribute these differences in labour input adjustment to the much stronger degree of employment protection in European countries greater and the reliance on short-time work schemes during economic downturns.
Hijzen and Venn (2011[4]) Boeri and Bruecker (2011[5]), Cahuc and Carcillo (2011[6]) and Hijzen and Martin (2013[7]) analyse the effects of short-time work schemes during the global financial crisis using country-level data. These studies typically show that short-time work schemes are effective in preserving employment during a temporary downturn, with limited deadweight effects (one‑third of take‑up in Hijzen and Venn (2011[4]), no deadweight in Cahuc and Carcillo (2011[6]). Moreover, their positive effects are largely limited to workers on permanent contracts, further increasing the labour market divide between workers on permanent and temporary contracts. Finally, the effects of short-time work schemes are short-lived as they slow job creation in the recovery if their use is maintained for too long (Hijzen and Martin, 2013[7]).
Micro-level evidence
A number of recent studies analyse the effects of job retention schemes using micro data (firm- or plant-level data) in combination with credible research designs that allow identifying causal effects by effectively addressing firm selection in programme participation. Note that while these firm-level studies typically score more favourably when it comes to selection bias and endogeneity than cross-country studies, they do not account for general equilibrium effects related to spillover and feedback effects (through consumption or wage‑setting). These firm-level studies also tend to find that job retention schemes save jobs in the short run, while results are more mixed thereafter.
Cahuc et al. (2021[8]) study a short-time work scheme in France during the global financial crisis. They stratify firms across regions according to the predicted shock in business activity and administrative efficiency. For the bottom quintile of firms in terms of the shock, i.e. the firms hit most severely during the global financial crisis, they find that one additional worker on short-time work saved 0.6 jobs in 2009. However, there is no effect for the other four quintiles of firms, which are hit less severely by the global financial crisis. The results are similar in the short and medium term.
Kopp and Siegenthaler (2021[9]) study a short-time work scheme in Switzerland. They use establishment-level data and exogenous variation in discretionary approval practices across Swiss cantons. They find that short-time work increased full-time equivalent employment by 9 to 17% which corresponds to between 0.19 and 0.36 full-time jobs for every worker enrolled in the program. Importantly, the effect persists in the medium term, suggesting that jobs were saved permanently. Moreover, savings on unemployment benefits may have compensated for the spending on short-time work benefits and hence that short-time work effectively paid for itself. This partly reflects limited deadweight and displacement effects, and partly the finding that workers whose job was preserved would have become unemployed for a relatively long time (and longer than their participation in short-time work).
Giupponi and Landais (2023[10]) consider a short-time work scheme in Italy. They use firm-level data along with exogenous variation in eligibility by firm size (15 or more employees depending on the industry). They find that firms that used the short-time work scheme had a smaller drop in employment (45% less) and a larger drop in hours worked (40% more). This suggests that in the short-run deadweight and displacement effects were limited. However, the positive effects of STW do not persist and disappear after one year. This may be related to the fact that mostly low-productivity firms used the job retentions scheme in Italy.
Source: Hijzen et al. (2024, forthcoming[1]), “The effectiveness of job retention schemes during economic crises”, Handbook of Labour Markets in Transition.
5.2.2. Studies on the COVID‑19 crisis
Macro-level evidence
Assessing the role of job retention schemes for employment during the COVID‑19 pandemic using country-level data is even more challenging than during the global financial crisis due to the unprecedented and synchronised nature of the pandemic and the variety of different support measures used by governments to protect firms and workers against the fallout of the pandemic (e.g. tax deferrals, debt moratoria, subsidised loans and direct income‑support to workers and households). These factors greatly complicate the task of defining a meaningful counterfactual. Based on the country-level evidence, OECD (2021[17]) suggests that job retention schemes played a significant role in limiting job losses, but also that there is considerable uncertainty about the magnitude of its impact. It estimates that employment would have been between 6‑11% lower during the first six months of the COVID‑19 crisis in the absence of job retention schemes instead of the actual reduction of employment by 4%.
The positive impact of JR schemes on employment during the COVID‑19 crisis is also found in a more recent OECD study by Calligaris et al. (2023[18]). This study conducts a differences-in-differences analysis that simultaneously exploits variation in the availability and generosity of JR schemes across countries and the size of the shock as measured by the “teleworkability” of an industry. The results show that JR schemes were successful in their purpose of cushioning the effect of the crisis on employment and firm survival. A particularly interesting feature of the study is that it also looks at the effects on different segments of the firm productivity distribution (by industry, country and month). They find that while JR schemes initially reduced firm exit rates, the latter recovered in 2021. Consequently, the study concludes that JR schemes did not result in significant distortions of productivity-enhancing reallocation.
Micro-level evidence
Several micro‑econometric studies have looked at the short-run impact of short-time work/job retention schemes during the COVID‑19 crisis. However, the near universal roll-out of the schemes during COVID‑19 makes, complicate the identification of causal effects, as there might not be a valid control group. Again, it should be noted that these studies focus on the differential effect between treatment and control firms, whereas the schemes may also have affected both types of firms via e.g. effects on consumer spending business investments and labour market tightness. Indeed, these type of spillovers effects may have been much more important during the COVID‑19 crisis given the mass use of JR support. Overall, studies looking at wage subsidy schemes typically find relatively small positive employment effects, which reflect their limited targeting, while studies looking into the short-time work schemes tend to find larger positive effects on employment.
Several studies have looked into the Paycheck Protection Program (PPP) in the United States, a preferential loan scheme for small and medium sized firms with less than 500 employees. Firms could draw PPP loans up to ten weeks of payroll costs, with a maximum of USD 10 million. These loans were forgiven if employment was maintained at a level comparable to that before COVID‑19. Autor et al. (2022[11]) show that in the first two tranches of the PPP in 2020, 94% of eligible firms took up PPP loans, suggesting that there was little targeting beyond firm size. Using a difference‑in-differences approach, with larger firms as the control group, they find that the programme increased employment by 4‑10% in mid-May of 2020, and just 0‑6% by the end of the year.1 They conclude that setting up a short-time work scheme that can be scaled up quickly during a major economic crisis would allow for a better targeting of job retention support in the United States.
Smart et al. (2023[13]) analyse the effects of the Canada Emergency Wage Subsidy scheme. They exploit discontinuities in subsidy rates by revenue loss to estimate the effects of the scheme on job losses and business closures. They find a net wage elasticity of employment of 0.1, implying a small aggregate employment effect and an estimated fiscal cost per job saved of nearly USD 200 000 per year. Subsidy payments caused a small but persistent reduction in business closure rates and increased the earnings of existing employees. The authors argue that the scheme functioned more as an income insurance scheme than as a job saving scheme.
Bennedsen et al. (2023[15]) consider the effects of a STW scheme in Denmark. The scheme covered 75% of wages when a worker was furloughed (i.e. job fully suspended). They use a selection-on-observables model together with a survey of managers on planned layoffs and furloughs at the start of the COVID‑19 crisis. They find substantial selection into STW, with firms that took support expecting to have a 16 percentage points higher layoff rate than firms that did not. Controlling for selection using planned furloughs and layoffs by managers, they find that the scheme reduced the layoff rate by 24 percentage points.
Finally, Benkovskis et al. (2023[16]) study a STW scheme in Latvia. The scheme initially operated as a furlough scheme, with eligibility conditional on a revenue loss of at least 30%. They use propensity score matching in combination with difference‑in-differences. Control firms exhibit a similarly sized predicted decline in hours worked as treated firms but did not take up the job retention support.2 They find that firms that took up job retention support experienced about 30% higher employment growth rates than control firms, with the effect becoming somewhat smaller towards October 2020 (the last month used in the regression analysis). A large part of the effect runs through less firm exits.
5.3.1. Job retention support is likely to have played a major role in preventing a surge in unemployment
To get a first impression of the possible role of ERTE during the COVID‑19 crisis, Figure 5.1 shows the evolution of key economic outcomes during the first eight quarters since the start of the crisis.3 The outcomes considered are GDP, employment, average hours worked and take‑up of job retention support. To put the developments for Spain in context, they are systematically compared to those observed during first eight quarters of the global financial crisis and those of selected benchmark countries (France, Germany and Italy).
The initial number of jobs lost during the COVID‑19 crisis in Spain was sizeable compared with other countries, but modest compared with the large reduction in economic activity. Two quarters after the start of the COVID‑19 crisis, employment reached its trough in Spain, 5% below its level before the crisis. In France, Italy and Germany, employment declined by 2‑3% over the same period. The larger employment decline in Spain to some extent reflects the larger decline in economic activity (22% in Spain versus 11‑19% in the other countries). However, the decline in employment during the COVID‑19 crisis in Spain was not as large as during the global financial crisis when employment fell by 8% after two years. This is even clearer when taking account of the decline in economic activity which was more than four times as large during the COVID‑19 crisis in Spain than during the global financial crisis. All in all, one can conclude that the employment response to the decline in economic activity in Spain was more normal during the COVID‑19 crisis, although still relatively large, while it stood out strongly for its tendency to shed jobs during the global financial crisis.
Indeed, the Spanish labour market seems to have adjusted in a fundamentally different way to the COVID‑19 shock than the shock that resulted from the global financial crisis. One clear indication of this was the massive reduction in hours worked per person during the COVID‑19 crisis. In Spain as well as in the benchmark countries, the sharp decline in economic activity translated in a very large reduction in hours worked per person. During the first two quarters of the COVID‑19 crisis, hours worked per person declined by 20% in Spain, similar to France and Italy, but substantially more than in Germany. In other words, labour market adjustment took the form of reductions in working time rather than reductions in employment. The importance of working-time reductions in turn is likely to reflect the unprecedented use of JR support and is likely to have contributed to the muted employment response to the decline in economic activity. The relative importance of working time reductions is not new for countries such as Germany and Italy, which have a long tradition of relying on short-time work schemes during economic downturns. However, this is a new phenomenon in Spain, and to a lesser extent France, two countries where short-time work remained underdeveloped until the advent of the COVID‑19 crisis.
While the discussion above provides some indication that short-time work is likely to have contributed to containing job losses in Spain, it is important to note that this tentative conclusion is partially based on comparisons in labour market adjustment during the COVID‑19 crisis and the global financial crisis even though these two episodes were very different in nature. While there are many differences, the fact that the COVID‑19 crisis was relatively short-lived, did not have any economic origins and the financial system was not impaired may have increased the incentives for firms to hoard workers during the COVID‑19 crisis and contributed to the shift in labour market adjustment from employment to working time. The comparison with Germany, France and Italy helps to overcome this challenge to some extent. Despite the very different nature of the crisis, the process of labour market adjustment in Germany and Italy is actually quite similar to that observed during the global financial crisis when the use of short-time work was also widespread (particularly when taking account of the smaller reduction in economic activity). This provides suggestive evidence that the unprecedented use of short-time work in Spain is likely to have contributed to the fundamentally different way the labour market adjusted to the decline in economic activity during the COVID‑19 crisis.
The evolution of GDP, employment, working time and job retention support since the start of the COVID‑19 (global financial) crisis in Spain as well as France, Germany and Italy
Source: GDP and hours: OECD National accounts database; Employment: OECD Labour Force Statistics database; Job retention schemes: National sources.
5.3.2. There is considerable uncertainty about the number of jobs saved thanks to JR support
While the analysis so far provides some indication that job retention support played a major role in preventing a surge in unemployment in Spain, it is not straightforward to provide an estimate of the number of jobs saved. This requires constructing a well-defined and credible counterfactual for what would have happened during the COVID‑19 crisis in the absence of JR schemes. However, this is far from obvious using country-level data due to the unprecedented and synchronised nature of the COVID‑19 crisis. Moreover, governments provided support to firms and workers through a variety of additional instruments such as tax deferrals, debt moratoria, subsidised loans and direct income‑support to workers and households. 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.
To nevertheless provide some further idea of the possible role of JR support in Spain, the approach adopted in Hijzen et al. (2021[17]) is applied here to provide a lower and an upper bound estimate of the number of jobs saved. The upper bound estimate of the number of jobs saved is derived from the cross-country correlation between take‑up of JR support and the change in average hours worked. This 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 terminated otherwise (i.e. no efficiency losses).4 The lower bound is based on the cross-country correlation between JR support and the change in the number of employees. It provides an indication of the number of jobs saved under the assumption that employment growth is not affected by any other factors that correlate with JR take‑up (i.e. other policy interventions, pre‑existing policies and institutions, the size and nature of the shock). While this in principle allows for efficiency losses, it does not take account of many important confounding factors. 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.
The estimates point to a broad range of plausible jobs effects during the COVID‑19 crisis. Their implications for the evolution of employment during the COVID‑19 crisis are visualised in Figure 5.2. 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 during the first year of the COVID‑19 crisis until the end of 2020. The counterfactual changes have been obtained by adjusting the actual change in employment for the estimated effect of JR use on hours and employment. As before, the analysis is conducted for Spain as well as for France, Germany and Italy.
In Spain, the number of employees fell sharply by about 5% between Q1 2020 and Q2 2020, compared with 4% for the OECD as a whole, and 2‑3% for the benchmark countries, 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 indicate 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 12%. 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 about 35% larger than the actual change in employment, resulting in a decline in employment of almost 7%. The pattern for France, Germany and Italy are similar by construction. The analysis only takes account of differences in the actual evolution of employment and the take‑up rate of JR support. It does not take account of any possible differences in effectiveness of JR support across countries.
Percentage change in the number of employees with and without JR support from Q1 2020 to Q4 2020
Note: Adjusted for the predicted effect of JR use on average hours = Actual employment – the average slope coefficient in the scatter plots relating average hours worked to the use of JR schemes * the actual use of JR schemes. Adjusted for the predicted effect of JR use on employment = Actual employment – the average slope coefficient in the scatter plots relating employment to the use of JR schemes * the actual use of JR schemes. Seasonally adjusted hours and employees’ data.
Source: OECD (2021[17]), “Job retention schemes during the COVID‑19 crisis: Promoting job retention while supporting job creation”.
The review of the empirical literature on the effects of job retention schemes and the descriptive evidence for Spain suggest that job retention schemes in general and ERTE in particular are likely to have played an important role in preventing a surge in unemployment. However, there is much uncertainty about the precise quantitative impact of ERTE on the number of jobs saved. In order to get a deeper understanding of the role played by ERTE a detailed evaluation is needed that allows assessing its causal effects based on a credible identification strategy. This will be the subject of the next chapter.
References
[2] Abraham, K. et al. (1994), “Does Employment Protection Inhibit Labor Market Flexibility? Lessons from Germany, France, and Belgium”, pp. 59-94, https://EconPapers.repec.org/RePEc:nbr:nberch:11255 (accessed on 18 December 2023).
[3] 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, https://www.jstor.org/stable/2138794 (accessed on 15 February 2021).
[11] Autor, D. et al. (2022), “The $800 Billion Paycheck Protection Program: Where Did the Money Go and Why Did It Go There?”, Journal of Economic Perspectives, Vol. 36/2, pp. 55-80, https://doi.org/10.1257/JEP.36.2.55.
[19] Autor, D. et al. (2020), An Evaluation of the Paycheck Protection Program Using Administrative Payroll Microdata *.
[21] Bargain, O. and T. Callan ((2010)), Analysing the effects of tax-benefit reforms on income distribution: a decomposition approach. , J Econ Inequal .
[16] Benkovskis, K., O. Tkacevs and K. Vilerts (2023), Did job retention schemes save jobs during the Covid-19 pandemic? Firm-level evidence from Latvia, Latvijas Banka Working Paper 3/2023.
[15] Bennedsen, M. et al. (2023), The effect of preserving job matches during a crisis, Labour Economics.
[5] 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, https://doi.org/10.1111/j.1468-0327.2011.271.x.
[6] 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).
[8] Cahuc, P., F. Kramarz and S. Nevoux (2021), “The Heterogeneous Impact of Short-Time Work: From Saved Jobs to Windfall Effects”, https://doi.org/10.34724/CASD.
[18] 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.
[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, https://doi.org/10.3386/w27431.
[10] Giupponi, G. and C. Landais (2023), “Subsidizing Labour Hoarding in Recessions: The Employment and Welfare Effects of Short-time Work”, The Review of Economic Studies, Vol. 90/4, pp. 1963-2005, https://doi.org/10.1093/RESTUD/RDAC069.
[20] Herzog-Stein, A., G. Horn and U. Stein (2013), “Macroeconomic Implications of the German Short-time Work Policy during the Great Recession”, Global Policy, Vol. 4, pp. 30-40, https://doi.org/10.1111/1758-5899.12054.
[1] Hijzen, A., E. Jongen and M. Montenegro (2024, forthcoming), The effectiveness of job retention schemes during economic crises, Handbook of Labour Markets in Transition, Publishing, Palgrave.
[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, https://doi.org/10.1186/2193-9004-2-5.
[17] Hijzen, A., A. Salvatori and A. Puymoyen (2021), “Job retention schemes during the COVID-19 crisis: Promoting job retention while supporting job creation”, in OECD Employment Outlook 2021, OECD Publishing, Paris, https://doi.org/10.1787/c4c76f50-en (accessed on 23 November 2021).
[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://doi.org/10.1787/5kgkd0bbwvxp-en.
[9] Kopp, D. and M. Siegenthaler (2021), “Short-Time Work and Unemployment in and after the Great Recession”, Journal of the European Economic Association, Vol. 19/4, pp. 2283-2321, https://doi.org/10.1093/JEEA/JVAB003.
[14] Montenegro, M. (2024, forthcoming), “Job Retention at Scale”, Working Paper.
[22] OECD (2023), Joining Forces for Gender Equality: What is Holding us Back?, OECD Publishing, Paris, https://doi.org/10.1787/67d48024-en.
[13] Smart, M.; Kronberg, M.; Lesica, J.; Liu, H.; D. Leung (2023), The employment effects of a pandemic wage subsidy, CESifo Working Paper No. 10218.
← 1. Using a different data source, Chetty et al. (2020[12]) find an even more modest employment effect of 2.5 percentage points for the PPP over the period April-August 2020, not statistically significantly different from zero.
← 2. Although the authors control for e.g. several observable characteristics of the firms, omitted variable bias might still be a concern in this study.
← 3. While in principle it is possible to extend the analysis using more recent data, this is not done here for two reasons. First, the main interest here to provide a first indication of the impact of the COVID‑10 crisis and job retention support on the evolution of employment. The focus on the first two years of the crisis here is broadly consistent with the remainder of the report. The link with the crisis and the use of job retention support beyond this period is likely much less conclusive. Second, extending the analysis to more recent periods would likely increase the sensitivity of the analysis to the data used related to GDP revisions related to the economic recovery and the diverging evolution of hours worked from 2022 onwards depending on the source used.
← 4. Note that this can be considered an upper bound of the partial equilibrium effect. When there are important general equilibrium effects, employment effects may be larger due to for example the role of labour market congestion effects. See Chapter 6 for a discussion.