Will Schumpeter catch COVID-19? Evidence from France

Mathieu Cros, Anne Epaulard, Philippe Martin 04 March 2021



The COVID-19 crisis, a global shock ‘like no other’, has had dire consequences for several economic variables, including consumption, production, employment, trade, productivity, business and consumer confidence, and so on. However, one economic impact that was anticipated very early on (e.g. Gourinchas et al. 2020) has not materialised so far – namely, firm bankruptcies. Indeed, the number of bankruptcy filings has decreased significantly. As illustrated in Figure 1, the number of firms filing for bankruptcy in France, for example, is well below its normal level (-36% at the end of 2020 compared to 2019). Although international comparisons of bankruptcy filings are not easy, the UK and German situations appear similar. 

Figure 1 Cumulative number of firms filing for bankruptcy, 2008–2020

Note: At the end of 2020 the cumulative number of bankruptcy filings had reached 26,779, while at the end of 2019, the cumulative number of bankruptcy filings had reached 42,687. Source: BODACC data up to December 2020

The main explanation for this unexpected observation is that governments have provided ample liquidity and financial support to firms most affected by the pandemic. But have governments gone too far? Some concerns have emerged in the public debate that these policies may create ‘zombies’ by reducing the exit of non-productive firms (The Economist 2020, Financial Times 2020). If so, this may have dire consequences for productivity in the following years, as the exit of unproductive firms is a substantial contributor to aggregate productivity growth. Foster et al. (2001) find that entry and exit of plants accounted for around 25% of US manufacturing productivity growth over the period 1977–1992 and that the impact of net entry is probably larger in the services sector. This effect comes from exiting firms that are less productive and/or less innovative than both continuing and new firms (Syverson 2011). Furthermore, Adalet-McGowan et al. (2018) find that zombie firms reduce the growth of more productive firms and might also reduce entry. This further increases the potential burden of surviving low-productivity firms on aggregate productivity.   

The concern that public policies to support firms may impair the cleansing effect of the recession by saving unproductive firms from exit is therefore legitimate. But the opposite concern that productive firms may go bankrupt because of the COVID-19 crisis is also legitimate. The cleansing effect is based on the implicit assumption that markets efficiently select the most productive firms. However, several studies show that the probability of firm failure depends not only on their productivity but also on their access to credit. Barlevy (2002), for example, studies the consequences of credit frictions on resource allocation during recessions and shows that credit frictions can lead to the opposite of the cleansing effect during recessions. Laeven et al. (2020) argue that “the different nature of the crisis means that many firms that normally would be classified as zombie firms are in fact viable firms”. Gagnon (2020) also argues that the concern on zombies in the COVID-19 crisis is overblown.

In a recent paper (Cros at al. 2021), we examine whether there is early evidence that the selection process of firm bankruptcies is not only partially frozen but also distorted, putting into danger Schumpeterian creative destruction. We offer a preliminary answer to this question based on French data. Although, firm bankruptcies have been sharply reduced, we still observe some (more than 60% of the ‘normal’ level) and we can therefore analyse whether the determinants of the mechanism of firm destruction have been sharply distorted by the crisis. Our results, again at an early stage, are relatively reassuring and point to hibernation rather than zombification:

  • The risk of an increase in productive firms going bankrupt during the pandemic did not materialise. The firms filing for bankruptcy in 2020 were already less productive and/or had higher debt in 2018. A logit model shows that the same main predictors of bankruptcy were at work in 2020 as in 2019 – productivity, debt, and age are still associated with bankruptcy probability. Moreover, the coefficients of these variables are not statistically different from one year to another. Creative destruction has been partially frozen but not distorted.
  • Not surprisingly, the reduction in the number of bankruptcies comes from lower bankruptcy filing of less-productive firms. In the short run, however, the impact on the aggregate productivity gain is likely to be small. This is only true if the process of creative destruction is unfrozen once the crisis is over. 
  • The COVID-19 shock has been very heterogeneous across sectors. This is particularly true for the commercial sector (e.g. restaurants versus food stores). We measure the shock for these sectors using the change in credit card transactions. We find that firms operating in sectors more affected by the COVID-19 shock are more likely to file for bankruptcy. However, the predictive power of the sectoral COVID-19 shock on bankruptcy is much smaller than that of firm productivity or debt. This suggests that public policies did compensate, in the short term, a very large part of the sectoral nature of the COVID-19 shock. 

Figure 2 illustrates these findings. It shows the contributions of different firm-level characteristics (supplier debt and other debt mostly to social contributions, bank debt, labour productivity, firm size, firm age) and the sectoral credit card transaction shock to the firm-level probability of failure. As in 2019, debt and productivity have the largest explanatory power in 2020. Although firms in sectors with larger credit card transaction reductions in 2020 have experienced higher risk of bankruptcy, the explanatory effect for firm-level risk of bankruptcy is quantitatively small.

Figure 2 Contributions of different predictors to bankruptcy risk in 2019 and 2020

Note: In 2019, including the ratio of bank debt to corporate assets amongst the explanatory variables for default increases the explanatory performance of the econometric model by 25% compared to a model where all the other variables are present, as well as sector fixed effects. 

The legacy of the pandemic on firms’ balance sheets will likely be large. The reduction in the number of bankruptcies due to generous liquidity measures comes at the cost of an increase in corporate debt, especially in those sectors that are most affected by the pandemic. For firms in these sectors, a return to ‘normal’ bankruptcy processes would see a large increase in bankruptcies from 1.1% in 2019 to 1.8% in 2021 (and after 0.7% in 2020). Although this is large, most of the increase comes from a catching-up process of bankruptcies that did not take place in 2020. One political economy issue for governments is that this return to normal through catch-up may be interpreted as a policy failure.

Our work is the first, to our knowledge, to estimate the factors predicting firm failures in the COVID-19 crisis based on actual data in 2020. At this stage, Schumpeter does not appear to have caught COVID-19 in the sense that the normal selection process in firm failure was not distorted in 2020. The policy challenge is therefore to continue to support productive and viable firms (but with potentially high debt due to the COVID-19 shock) while progressively discontinuing support to firms that are not viable.


Adalet-McGowan, M, D Andrews, and V Millot (2018), “The Walking Dead? Zombie Firms and Productivity Performance in OECD Countries”, Economic Policy 96: 685–736.

Barlevy, G (2002), “The sullying effect of recessions”, Review of Economic Studies 69(1).

Cros, M, A Epaulard and P Martin (2021), “Will Schumpeter Catch Covid-19?”, CEPR Discussion Paper 15834.

Economist (2020), “Why COVID-19 will make killing zombie firms off harder”, 26 September.

Financial Times (2020), “European Zombification becomes even scarier”, 3 December.

Foster, L, J C Haltiwanger and C J Krizan (2001), “Aggregate Productivity Growth: Lessons from Microeconomic Evidence”, In C. R. Hulten, E. R. Dean, & M. J. Harper (Eds.), New Developments in Productivity Analysis, University of Chicago Press.

Gagnon, J (2020), “Who’s afraid of zombie firms?”, PIIE.com, 22 October.

Gourinchas, P O, S Kalemli-Ozcan, V Penciakova and N Sander (2020), “COVID-19 and SME Failures”, NBER Working Paper, 27877.

Laeven, L, G Schepens and I Schnabel (2020), “Zombification in Europe in times of pandemic”, VoxEU.org, 11 October. 

Syverson, C (2011), “What Determines Productivity?”, Journal of Economic Literature 49(2): 326-65.



Topics:  Covid-19 Macroeconomic policy

Tags:  Schumpeter, creative destruction, COVID-19, Bankruptcy, zombification, zombie firms

Professor of Economics, Dauphine University - PSL

Professor of Economics, Sciences Po and Chair French Council of Economic Analysis; CEPR Vice President and Research Fellow


CEPR Policy Research