The Covid-19 pandemic has highlighted that firm survivals and exits are directly tied to livelihoods. There is a long-standing view within existing research that the exits of unproductive firms are welfare-enhancing due to the so-called ‘cleansing effects’ (Foster et al. 2014). The overall welfare implication of firm exits, however, needs to be considered in a holistic manner, as firm exits may generate negative externalities with long-term employment losses if it takes time for new productive firms to enter the market. As firms are part of intricate interconnected networks, firm exits may have knock-on effects on remaining healthy firms.1 Further, firm exits could disrupt the inheritance of intangible assets, central to the value of firms.
In this column, we analyse the exit patterns of Japanese firms using a detailed firm-level dataset from Tokyo Shoko Research (TSR), covering around one million firms each year.2 This dataset provides valuable information regarding the type of firm exits by distinguishing bankruptcy, voluntary exit, or merger exit. Using this information, we document exit patterns during normal times as well as in severe downturns, identifying characteristics that affect how a firm exits. Existing research does not distinguish how a firm exits (and if exits are associated with bankruptcy or defaults). However, we find that not accounting for the type of firm exit is detrimental to understanding the life cycle of firms in Japan. While there is a small number of firms that exit due to bankruptcy, most firm exits are voluntary – where owners decide to discontinue the business (Hong et al. 2020a). Policies that focus on improving solvency and liquidity issues are unlikely to affect firm exit rates related to voluntary exits or mergers. As we explain below, firm characteristics also play an important role in determining how firms exit.
Firm exit patterns over time
The overall firm exit rate in Japan is very low (below 2%) and has been on a downward trend since 2008 (Figure 1). This is much lower compared to other countries such as the US (10%) or European countries (7% on average), suggesting that business dynamism is quite dormant in Japan.3 The most dominant form of firm exits in Japan are voluntary exits, with the owners of firms deciding to discontinue their businesses even when they are not forced to close for financial reasons. The relatively high share of voluntary exits in Japan is correlated with the aging of owners, where old owners cannot secure business successors (Hong et al. 2020a, 2020b). On the other hand, firm default rates – denoted by bankruptcy – are at low levels and appear to be decreasing over time. They also appear to be a-cyclical. Firm exits through merger are at very low levels (at around 0.1%). In terms of the effects on the labour market, however, mergers explain the largest share of the number of employees affected by firm exits.4 In 2020, the total number of employees affected by mergers was more than twice the total number of employees affected by voluntary exits, even though voluntary exit rates are more than seven times larger than merger exit rates.
Figure 1 Firm exit rates and total number of employees affecte,d by exit type
Extensive and intensive margin adjustment in the episodes of large economic shocks
What are the effects of Covid-19 on firm exit rate? A simple comparison of exit rates between 2019 and 2020 may not give a convincing answer, since other factors may be at play.5 Therefore, in this column, we focus on the characteristics of exiting firms and analyse if correlations between firm characteristics and exit rates have changed.
We first examine monthly firm exits together with the changes in a monthly economic output measure. In particular, we focus on three major economic events associated with significant economic downturns in Japan: the Lehman Brothers Shock (or global crisis), the 2011 Tohoku earthquake, and the Covid-19 pandemic. We use as a measure the IIP (indices of industrial production) – a commonly used indicator collected by METI (ministry of economy, trade and industry) based on a survey of production of (relatively large) manufacturing establishments.
Figure 2 Firm exits and output changes: GFC vs. Tohoku Earthquake vs. COVID-19
Notes: X-axis represents the number of months before and after the time of the major event. Time t=1 denotes the month when the major event broke out and refers to September 2008, March 2011 and February 2020 for the GFC, the Tohoku Earthquake and the COVID-19, respectively. Y-axis in the left chart represents the Indices of Industrial Production (IIP) for each month, normalized by IIP at time t=0. Y-axis in the right chart represents the seasonally adjusted number of firms exited for each month, normalized by the number of firms exited at time t=0.
The left panel of Figure 2 shows the change in the indices of industrial production in the months before and after each shock. We see a decrease in production and recovery in each episode, albeit of varying shapes. Similarly, we compare the monthly total number of firms exiting in the right panel of Figure 2. Interestingly, overall firm exits declined in most months following the Covid-19 outbreak.6 For all three episodes, firm exits remain relatively subdued in the months following the shocks. A similar pattern can be observed if we restrict the sample to the manufacturing sector, which contrasts sharply with the large swings in production. This shows that Japanese firms tend to respond to economic shocks predominantly through intensive margin adjustment (production/output), rather than extensive margin adjustment (exits).
Firm exit patterns and cleansing effects
Now, we focus on firm exit patterns focusing on a selection of firm characteristics that are often used to assess the cleansing effects (Foster et al. 2014). We look at a simple correlation between firm characteristics and firm exit rates. We do this by categorising firms based on firm size (the number of employees), labour productivity (sales per employee), and healthiness (a variable called ‘score’, defined by TSR reflecting financial health, management and other firm-related information).
Figure 3 Firm exit rate, by firm size
Notes: Firms are categorized into 5 groups: less than 10 employees, from 10 to 49, from 50 to 99, from 100 to 299, and over 300 employees. We compare the exit rates in 2020 with those in the preceding three years (from 2017 to 2019). ‘Variable_pre’ is the average of annual firm exit rates by exit rate from 2017 to 2019.
Figure 4 Firm exit rate by labour productivity (left) and health (right)
Notes: Firms are categorized in quantiles. Bin 1 represents firms in the lowest quartile of each variable. Firms are healthier with higher scores. We compare such effects of exit rates in 2020 with those in the preceding three years (from 2017 to 2019). ‘Variable_pre’ is the average of annual firm exit rates by exit rate from 2017 to 2019.
Overall, smaller firms are more likely to exit. Figure 3 shows that downward sloping curves are observed for the overall firm exit rate, which is driven by voluntary exits. On the other hand, merger exit rates are positively correlated with firm size, while the exit rate of bankruptcies do not vary across different firm size groups. Figure 4 shows the results for labour productivity (left chart) and healthiness (right chart). Less productive and less healthy firms have higher exit rates, as total exit rates and voluntary exit rates exhibit downward-sloping curves, pointing to cleansing effects of firm exits. The effect is not obvious for bankruptcy, while a positive correlation is observed for merger exit rates. The differences in exit rates across groups are clearer when we use the firm healthiness measure. Importantly, in all three figures, exit rates have increased very mildly in 2020. But overall the slope is stable for all types of exits.
The proponents of the cleansing effects argue that unproductive firms should exit during recessions to enhance total productivity in the long run (by keeping only productive firms in the market). On the other hand, there is a view that government interventions to support firms in recessions help avert a cascade of bankruptcies. Our evidence suggests that firm exit rates have been muted during the Covid-19 pandemic. This means that the cleansing effects – as measured by the slope of the correlations – have not, as it stands, increased in Japan.
Authors’ note: The views expressed herein are those of the authors and should not be attributed to the IMF or RIETI, its Executive Board, or its management.
Carvalho, V M, M Nirei and Y U Saito (2014), “Supply Chain Disruptions: Evidence from the Great East Japan Earthquake”, RIETI Discussion Paper Series 14-E-035
Carvalho, V M, M Nirei, Y U Saito and A Tahbaz-Salehi (forthcoming), “Supply Chain Disruptions: Evidence from the Great East Japan Earthquake,” Quarterly Journal of Economics.
Foster, L, C Grim and J Haltiwanger (2014), “Reallocation in the Great Recession: Cleansing or Not?”, NBER working paper 20427.
Hong, G, A Ito, Y U Saito and A Thi Ngoc Nguyen (2020a), “Structural Changes in Japanese Firms: Business Dynamism in an Aging Society”, IMF working paper series 20(182).
Hong, G, S Kikuchi and Y U Saito (2020b), “What are the Effects of the COVID-19 Crisis on Firm Exit in Japan”, RIETI column.
1 Shock propagation to non-affected firms through transaction network after 2011 Tohoku earthquake is examined (Carvalho et.al, 2014, forthcoming). Shocks are propagated not only to direct partners but also indirect partners.
2 Firm information, such as number of employees and sales, are collected by TSR and the timing of dataset provided is at the end of September for each year, from 2007 to 2019. Firm exit information is from October 2007 to September 2020. Exit rate of 2020, for example, is calculated number of firms exiting from October 2019 to September 2020 divided by number of total firms at the end of September 2019.
3 Source: Bureau of Labor Statistics for the US, OECD and Eurostat for the European countries
4 The deviation in total number of employees affected by voluntary exit in 2008 is caused by one big firm.
5 The changes in voluntary exits are difficult to assess empirically, as they are determined by owners’ individual expectations about future economic activity and firm business model. For instance, while the exit rate increased in 2018 while it decreased in 2019, it is difficult to tell a coherent story as to what caused these ups and downs.
6 An increase in month 5 (June 2020) is due to a temporary increase in mergers.