Abstract
This paper examines whether the COVID-19-induced employment shocks have increased suicides and safety net use in the second and third quarters of 2020. We exploit plausibly exogenous regional variation in the magnitude of the employment shocks in Japan and adopt a difference-in-differences research design to identify the employment-shock impact. Our preferred point estimates suggest that a one-percentage-point increase in the unemployment rate in the second quarter of 2020 resulted in, approximately, an additional 0.39 suicides, 89 and 82 recipients of two temporary loan programs, 28 recipients of the Housing Security Benefit, and 11 recipients of Public Assistance per 100,000 population per month. A simple back-of-the-envelope calculation based on the estimate suggests that if Japan experienced a one-percentage-point increase in the unemployment rate caused by the COVID-19 crisis in the second quarter (i.e., April to June) in 2020, which is roughly equivalent to the third-highest employment shock at the regional level, this could have led to an additional 495 suicides in July 2020.
1 Introduction
The COVID-19 pandemic has caused serious infections and deaths around the world, and it is becoming clearer that its economic and social consequences are also tremendous. The COVID-19 pandemic and subsequent social-distancing policies have caused a sharp contraction of economic and social activities. National governments around the world have been trying to mitigate socio-economic damages by introducing new emergency cash benefits and expanding existing safety net programs. There are daily news reports of job loss, poverty, mental disorders, and even suicides that have been directly or indirectly induced by the COVID-19 crisis. Numerous social-science studies of the impact of the COVID-19 crisis have also been published on a daily basis.
However, it is still not well understood how the COVID-19 economic shocks, such as a sharp increase in unemployment, have affected society. Previous social-science studies have tended to focus on how the COVID-19 crisis as a whole, including social-distancing policies, has affected economic and social outcomes, but fewer studies have directly examined the effects of COVID-19-induced employment shocks on social outcomes such as mental and financial distress.
In this paper, we provide an early-stage analysis on how the COVID-19-induced employment shocks affect suicide and financial distress in Japan. We exploit an increase in the regional unemployment rate as an indicator of the economic shocks because unemployment can directly harm the well-being of workers and their families. We focus on suicide and safety net use as social outcomes because the former can be interpreted as a devastating consequence of deteriorated well-being and the latter are administrative indices that reflect the number of people or households that suffer from distressed living and financial conditions.
As a research design, we exploit a prefecture-level (akin to state-level in the U.S. or province-level in Canada) variation in the magnitude of the COVID-19-induced increase in unemployment. That is, we adopt an even-study difference-in-differences (DID) research design and examine how the regional employment shocks caused by COVID-19 are correlated with the trends of suicide rates and safety net participation before and during the crisis. As for safety net programs, we examine five programs in Japan’s three-tier safety net system addressing unemployment and poverty: unemployment benefits (first tier), two temporary loans and temporary housing benefits (second tier), and a public assistance program (third tier).
Our preferred point estimates suggest that a one-percentage-point increase in the unemployment rate in the second quarter (i.e., April to June) of 2020 resulted in, approximately, an additional 0.393 suicides in the subsequent month. A simple back-of-the-envelope calculation based on this estimate suggests that if Japan experienced a one-percentage point increase in the unemployment rate caused by the COVID-19 crisis in the second quarter of 2020, which is roughly equivalent to the third-highest employment shock at the regional (i.e. prefecture) level, this could have led to an additional 495 suicides in July 2020.1 This effect size is not negligible given that the number of suicides (based on the estimated days of death) was 1637 in July 2019.
Our results also imply that the same one-percentage-point increase in the unemployment rate led to no statistically significant increase in unemployment benefits, but resulted in an additional 89 and 82 recipients for two temporary loan programs, 28 recipients for temporary Housing Security Benefit, and 11 recipients for Public Assistance per 100,000 population per month. There having been no effect on the first-tier unemployment benefit appears incomprehensible, but our further investigation suggests that this may be explained by the fact that our baseline treatment variable with the unemployment rate is closely related to the COVID-19-induced employment shocks for contingent workers who were not entitled to receive unemployment benefits.
The contributions of our paper are at least two-fold. First, we provide early-stage plausible evidence of the impact of the COVID-19-induced employment shocks on mental and financial distress such as suicide and safety net use, which has rarely been studied so far. Note that the impact of the COVID-19 pandemic on socio-economic outcomes including unemployment has been extensively studied (see, among others, Chetty et al. (2020), Adams-Prassl et al. (2020) and Kikuchi et al. (2021)). There are also several epidemiological studies that investigate suicide under the COVID-19 shocks (Tanaka and Okamoto 2021; Nomura et al. 2021; Sakamoto et al. 2021; Ueda et al. 2020; Isumi et al. 2020; Sueki and Ueda 2020). Some studies also point out that the rise in unemployment caused by the COVID-19 pandemic is expected to lead to an increase in the suicide rate (Kawohl and Nordt 2020; Gunnell et al. 2020). Arthi and Parman (2020) examines the possible long-run impacts of COVID-19 economic consequences on various socio-economic outcomes from a historical perspective. None of these studies, however, directly investigate the effects of COVID-19 employment shocks on suicide and other socio-economic outcomes as we do.
Second, from a more general viewpoint, our study also contributes to the literature on the impact of unemployment on suicide, poverty, and safety net use. The effect of recessions on mortality has been well studied and many studies have shown that increases in unemployment are associated with a rise in the suicide rate (Ruhm 2000; Chen et al. 2009; 2015; Ruhm 2015; Matsubayashi et al. 2020). For example, Ruhm (2000) show that a one-percentage-point increase in the state unemployment rate was associated with an approximate 1.3 percent increase in the suicide rate in the US, which is much lower than our counterpart estimated effect size of an additional 490 suicides in a month when the counterpart monthly suicide number is 1637 in the previous year. When it comes to poverty and safety net use, some studies examine the impact of unemployment-rate fluctuation on use of safety net programs such as unemployment insurance benefits and means-tested programs before the COVID-19 crisis (Bitler and Hoynes 2016; Ganong and Liebman 2018; Bitler et al. 2020). Our findings add new evidence to the literature by estimating and comparing the impact of unemployment on the use of multi-tier safety net programs.
The rest of the paper is organized as follows. Section 2 provides background information about the COVID-19 socio-economic crisis in Japan with time-series graphs. Section 3 describes our research design, empirical model, and data. In Section 4 we present baseline estimation results and in Section 5 we also provide the results of robustness checks and alternative estimations with a different employment-shock variable. Section 6 concludes.
2 Background
Japan has suffered from economic and social difficulties despite its relative success in mitigating COVID-19 infection spread without imposing a strict lockdown policy.2 As panels (a) and (b) in Figure 1 shows, in Japan, the COVID-19 employment shocks emerged as sharp drops in the employment rate and labor force participation (LFP) rate since April 2020, while the indices of the unemployment rate and the jobs-to-applicants ratio (i.e., the ratio of job offers to job applicants) gradually deteriorated.
In turn, panels (c) and (d) in Figure 1 suggest that suicide rates have been increasing since July 2020. It is not clear from this graph to what degree the COVID-19 crisis has led to the increase in suicides, but some specialists and reports in the mass media have asserted that the sharp deterioration in employment in the first few months of the COVID-19 crisis may have resulted in the slightly lagged increase in suicides.3 In appendix A, we present the numbers of suicides in 2019 and 2020 by sex, age and occupation and these data show that the increase in suicides from 2019 and 2020 stands out in both employed and non-employed women within all age cohorts and young employed men aged 20-29.
Furthermore, Figure 1 also shows that the use of several social safety net programs also increased in the COVID-19 crisis. Panel (e) shows that the number of unemployment benefits under the Unemployment Insurance, which is sometimes called the first-tier safety net for unemployment in Japan, clearly increased since July 2020. Panel (f) also shows that the numbers of accepted applications for public temporary loan programs (Emergency Small Amount Funds and General Support Funds) and temporary housing allowance called Housing Security Benefit, which are considered to be second-tier safety nets, also sharply increased from April 2020 after the relaxation of requirements. On the other hand, panel (g) shows that the number of recipients of the Public Assistance program, which can be regarded as a third-tier safety net and is often called a “last resort”, continued to slightly decrease even during the COVID-19 crisis. But when we examine the year-on-year numbers of recipient households by household types in panel (h), the number of “others”, which should include low-income households that are sensitive to employment shocks, clearly started increasing in April 2020.
3 Research design and data
3.1 Employment shocks
Exploiting the sudden and plausibly exogenous variation in the COVID-19-induced employment shocks at the prefecture level, we adopt an event-study DID strategy to investigate the effect of the COVID-19 employment shocks on suicide and safety net use. Japanese prefectures, such as Tokyo, Kyoto, and Hokkaido, are comparable to states in the U.S. and provinces in Canada and some European countries, and number 47 in total.
We define the COVID-19-induced employment shock as the following de-trended increase in the unemployment rate in the second quarter (i.e. April to June) in 2020: where i indicates prefecture, X is the unemployment rate, and Q2 and Q4 are the second and fourth quarters respectively. The first term in equation (1) is a year-on-year difference in the unemployment rate in the second quarter of 2020, which consists of a COVID-19-induced employment shock and a prefecture-specific year trend. The second term is a corresponding year-on-year difference in the fourth quarter of 2019, which should reflect a prefecture-specific year trend just before the COVID-19 crisis. Assuming that the prefecture-specific year trends are similar just before and after the COVID-19 crisis, the difference between these two terms is thus expected to capture only the COVID-19-induced employment shock.
Note that what we use is the prefecture-level estimates of the unemployment rate provided in the Labour Force Survey. The prefecture-level estimates are estimated by the Statistics Bureau. The Bureau states that some imprecision is expected in these estimates, but we treat the estimated values as true values and do not consider their statistical uncertainty in our statistical analysis due to data limitation.
3.2 Outcome variables
For outcome variables, we focus on total, female, and male suicide rates and five safety net programs used in Figure 1. See also Ando et al. (2020) for more details about Japan’s fiscal measures and safety net programs addressing COVID-19 in the first six months of the COVID-19 crisis.
Suicide rates
Our primary outcome of interest is the suicide rate, which is the number of suicides per 100,000 population. We investigate the monthly panel data of three suicide-rate variables: total suicide rate, female suicide rate, and male suicide rate. We use the prefecture-level data between January 2019 and September 2020 and the number of suicides is aggregated based on the dates and places of suicide. Suicide statistics are originally provided as police statistics and aggregated and arranged by the Ministry of Health, Labour, and Welfare (MHLW) as the Statistics of Suicide. (Jisatsu no Tokei).
Unemployment benefits
The first outcome of a safety net program examined is the number of unemployment benefit recipients per 100,000 population under the Unemployment Insurance. The Unemployment Insurance is unarguably the first-tier safety net program for the unemployed in most developed countries, but there are some caveats about the Japanese unemployment insurance system. First of all, the coverage rate of the unemployment benefits among unemployed was less than 30 percent in Japan in the early 2010s and it is lower than other developed countries even after taking into account institutional differences (Kitazawa 2015). Second, the COVID-19-induced employment shocks are more concentrated on part-time or contingent workers (Kikuchi et al. 2021) and contingent workers are often not eligible for unemployment benefits. Hence there is a possibility that the Unemployment Insurance did not work well as a safety net against the COVID-19 employment shocks as pointed out by Sakai (2021).
Temporary loan programs
The second and third outcomes of safety net programs are the numbers of accepted applications per 100,000 population for two types of interest-free and guarantor-free emergency temporary loan programs: Emergency Small Amount Funds for those who urgently need cash (up to 200,000 JYP or about 1,900 USD) and General Support Funds for those who need cash for a certain period (up to 150,000-200,000 JYP per month). They are both existing means-tested temporary loan programs which can be utilized as second-tier safety net programs before applying for Public Assistance. They were rarely used before the COVID-19 crisis (see Figure 1). However, their requirements have also been relaxed since March 2020 and the number of loan recipients has increased dramatically. We use the monthly data between January 2019 and September 2020, but due to the limited data availability we do not have the data for February and March 2020.4 In addition, the monthly-level data aggregation from April to July 2020 is slightly irregular, although it should not cause serious estimation bias.5
Housing Security Benefit
The fourth safety net program outcome is the number of accepted applications per 100,000 population for the Housing Security Benefit, which is an existing scheme of short-term (i.e. three to nine months) housing allowance, which is also a second-tier safety net program in Japan. This benefit scheme, originally intended only for those who have lost their jobs, was rarely used before the COVID crisis as shown in Figure 1. But since April 2020, its eligibility has been extended to those who have not lost their jobs but have still experienced a large income reduction, resulting in a considerable increase in the number of house-holds receiving the benefits. Because of the limited data availability, we can use only January, February, and March 2019 in the pre-COVID-19 period.
Public Assistance
The fifth and sixth variables of safety net programs are the per-capita numbers of Public Assistance recipients and recipient households. The Public Assistance program in Japan, the third-tier safety net, is considered to be the “final safety net” or “last resort,” and the prerequisites for application are in general strict, such as having no savings and no assets. We can use only total recipient and total recipient household numbers for monthly prefecture-level data.
3.3 Empirical model
The baseline model specification takes the following event-study specification with time-varying DID coefficients: where Yit is the outcome variable such as a suicide rate for prefecture i at time t, πi and θt are prefecture and month fixed effects, respectively, ϕit is an individual (i.e. prefecture) linear trend, Empshocki is a continuous treatment variable of the COVID-19 employment shock defined as in (1), 1[t = τ] is a dummy variable that takes the value of one if t = τ and zero otherwise, and εit is an error term.
The coefficients of interest are the time-varying coefficients βτ, which capture the correlation between EmpShocki and the outcome trend from January 2020 to time τ. When time τ is before January 2020, βτ can be interpreted as a placebo estimate that is expected to be zero if no confounding trends existed before the COVID-19 crisis. When month τ is after January 2020, βτ can be seen as the average impact of EmpShocki given that the placebo estimates are around zero and the assumption of no differential trends in the COVID-19 period is plausible.
In the baseline regression analysis, we use the weighted least square (WLS) estimation method in which prefecture-level population sizes are used as weights. With this WLS model, the heterogeneous effect of the employment shock on an outcome for a larger prefecture is more highly weighted in a DID estimator. In the section of robustness checks, we also use the ordinary least square (OLS) estimation method because, as Solon et al. (2015) point out, it is not clear whether a WLS estimator is more appropriate than an OLS estimator as an estimator of a population average effect when effect heterogeneity exists.
Furthermore, for the outcomes of the two temporary loan programs and the Housing Security Benefit, we do not include an individual linear trend ϕit in the model (2) because the levels of these outcomes are negligible in the pre-COVID-crisis period compared with the COVID-crisis period (see Figure 1). As robustness checks, we also provide estimation results of all the other outcomes with the estimation model that does not control for the individual linear trends.
3.4 Data exploration
Table 1 shows the descriptive statistics of our treatment and outcome variables. We use the panel data of 47 prefectures from January 2019 to September 2020 (21 months) and the full sample size is 987. Because the employment-shock variable defined in equation (1) is a time-invariant and cross-sectional variable, we have 47 observations. The outcome variables such as suicide rates, unemployment benefits, and Public Assistance recipients include 21-months of data (from January 2019 to September 2020) while the variables of Emergency Small Amount Funds, General Support Funds, and Housing Security Benefit are more restricted due to data limitations as already explained.
Figure 2 shows the variation of the employment shock measured as equation (1) (panel (a)) and correlations of this employment shock and changes in total, female, and male suicide rates in July 2020 from the average of the three corresponding suicide rates in 2017-2019 (panel (b)-(d)). We use the monthly suicide rates in July because our employment-shock variable is calculated as a shock in the second quarter (i.e. April to June) 2020. Panel (a) shows that the most affected prefectures are Okinawa, Kanagawa, and Osaka and their employment shocks are more than one percentage point. Panel (b)-(d) present some positive correlations between the employment shock and the suicide rates.
4 Results
4.1 Suicide rates
Figure 3 shows the estimation results for the suicide rates based on (2) using WLS. First, DID estimates in the pre-COVID-19 period are more or less zero, implying that there is no clear statistically significant correlation between the employment shock and all the outcome trends before January 2020. Second, when it comes to the post-COVID-19 estimates after January 2020, the estimates for total, female, and male suicide rates are positive from May to July and the lower limits of the 90% confidence intervals are often above or near zero. The sizes of the estimates suggest that the one-percentage-point increase in the unemployment rate in the second quarter in 2020 led to around 0.39, 0.33, and 0.46-point increases in the total, female, and male suicide rates in July 2020 respectively. In August and September, estimates are not significantly different from zero, implying that suicides in this period are not directly related to the second-quarter employment shock in 2020.
Overall, the DID estimates suggest that the COVID-19-induced employment shocks have led to increases in both female and male suicides. The magnitude of the estimated impact is not small. Because the average monthly suicide rate (i.e. the average number of suicides per 100,000 population in a month) during the sample period was around 1.33 in 2019 (based on the annual suicide rate of 16.0 given by police statistics), the increase by 0.39 is not negligible; as of January 2021 the population in Japan is about 12,5570,000, so a simple back-of-the-envelope calculation based on the estimate suggests that if Japan experienced a one-percentage point increase in the unemployment rate caused by the COVID-19 crisis in the second quarter (April-June) in 2020, which is roughly equivalent to the third-highest employment shock on Osaka (see Figure 1), this could have led to an additional 495 suicides in July 2020. Given the fact that the number of suicides per month in Japan in July 2019 was 1637, this is about a 30 % increase.
4.2 Safety net use
Figure 4 provides estimation results for safety net participation. To begin with, in the first row panels (a) and (b) correspond to estimates for the first-tier safety net program, that is Unemployment Insurance. DID estimates for the number of unemployment benefits for total, female, and male unemployed are unstable and not significantly different from zero in the COVID-19 crisis period. This surprising result may be explained by the fact that the unemployment-benefit coverage rate is low in Japan and that COVID-19-induced employment shocks are more concentrated on contingent workers who are often not eligible for unemployment benefits (Section 3.2).
The second row, panels (c) and (d), of Figure 4 provides estimates for the second-tier safety net programs: the two temporary loan programs and the Housing Security Benefit. They are mostly significantly different from zero in the COVID-crisis period, in particular in June and July, suggesting the positive effects of the second-quarter employment shocks on these programs. Note that the numbers of accepted applications for these programs are discontinuously smaller in the pre-COVID-19 period, so the pre-COVID-19 estimates are negligible and we cannot evaluate differential trends in these outcomes.
The third row, panels (e) and (f), of Figure 4 provide estimation results for the numbers of persons and households that receive Public Assistance, the third-tier safety net. Results show that estimates in the COVID-19 period are significantly different from zero while estimates in the pre-COVID-19 period are mostly not. While the estimates for the temporary loan programs and the Housing Security Benefit peaked in June or July 2020, the estimates for Public Assistance continued to increase through the third quarter of 2020.
The point estimates for the temporary loan programs and Housing Security Benefit suggest that a one-percent-point increase in the unemployment rate in the second quarter of 2020 resulted in an additional 88.8 accepted applications for Emergency Small Amount Funds in June, an additional 82.4 accepted applications for General Support Funds in July, and an additional 27.6 accepted applications for the Housing Security Benefit in June (all per 100,000 population). When it comes to Public Assistance, the one-percentage-point increase in the unemployment rate resulted in about an additional 11.2 recipients and 10.5 recipient households per 100,000 population. Similar effect sizes for recipients and recipient households suggest that the majority of COVID-19-induced Public Assistance recipients may consist of single-person households.
The effect sizes for these safety net programs are far from negligible. Just before the COVID-19 crisis, the number of Public Assistance recipient households excluding the households of the aged, the disabled, and the sick was about 255 per 100,000 population.6 On the other hand, the estimated increases per 100,000 population per one-percentage-point increase in the unemployment rate are around 80-90 applications for the temporary loan programs, about 28 applications for the Housing Security Benefit, and about 11 Public Assistant recipient households (per 100,000 population) in a month.
5 Further analyses
5.1 Robustness checks
As robustness checks, we present estimation results based on two model specifications (with or without individual linear trends ϕit in equation (2)), two regression schemes (OLS or WLS), and two types of outcome variables (monthly or monthly year-on-year difference). The rationales for these robustness checks are (a) to examine whether estimation results change if we use a simpler model without individual linear trends, (b) to compare OLS and WLS estimates and discuss how they differ and why as Solon et al. (2015) recommend, and (c) to control for prefecture-specific monthly fixed effects in addition to standard prefecture and monthly fixed effects.
Appendix C provides the estimation results of robustness checks for the suicide rates (Figure C.1), unemployment benefits (Figure C.2) and other safety net programs (Figure C.3) respectively. Overall, our primary findings for suicide and all the safety net programs seem plausible based on these different estimation settings. However, estimates for suicide are less robust and more imprecise when year-on-year differences are used as outcome variables. We also discuss these results in more detail in Appendix C.
5.2 An alternative treatment
In order to examine the characteristics of our main treatment variable based on the unemployment rate, we construct another employment-shock treatment variable based on a more limited employment indicator. That is, for the numerator of the unemployment rate Xit in the employment-shock equation (1), we use the number of the unemployed people who are registered as full-time-job seekers at public employment security offices. Because it is expected that these registered full-time-job seekers are often the breadwinners of households and eligible for unemployment benefits, this alternative employment-shock variable may reflect a specific part of the entire employment shock.
In Appendix D, we present DID estimation results using this alternative treatment variable in the same manner as the above robustness checks. Contrary to our main findings, estimates for suicide are not significantly different from zero (Figures D.2), but estimates for unemployment benefits are robustly positive and often significantly different from zero from July 2020 (Figure D.3). Estimates for the two temporary loan programs, the Housing Security Benefit, and Public Assistance are positive and in most cases significant in the COVID-19 period. (Figure D.4).
These results that are partly different from the baseline results may be explained by the fact that the alternative employment shocks reflect an unemployment increase in manufacturing industries whereas the baseline employment shocks are characterized as tertiary-industry shocks in metropolitan and service-sector regions. See further discussions in Appendix D.
6 Conclusion
Exploiting regional variations in the employment shocks caused by the COVID-19 crisis, this paper examines whether the COVID-19-induced employment shocks in the second quarter of 2020 have affected rates of suicide and safety net use in the second and third quarters of 2020.
Our estimation results suggest that the COVID-19-induced increase in unemployment has increased both the rate of suicide and the rate of safety net participation. Per 100,000 population and per month at peaks, the estimated effect sizes for a one-percentage-point increase in the unemployment rate are, approximately, an additional 0.39 suicides, 89 and 82 recipients for the two temporary loan programs, 28 recipients for the Housing Security Benefit, and 11 recipients for Public Assistance.
All of these values are not socio-economically negligible. Let’s consider a hypothetical region with a population of 10 million, which was roughly close to the populations of London (9.0 million), New York City (8.3 million), or Sweden (10.4 million) in 2020.7 Suppose that the COVID-19 crisis or a similar pandemic increases the unemployment rate of this region by one percentage point in a quarter in its early stage. In this region, this employment shock would then result in an additional 39 suicides, 8,900 and 9,200 recipients for the two temporary loan programs, 2,800 recipients of the Housing Security Benefits, and 1,100 recipients of Public Assistance in the last month or the next month of this quarter.
Although attitudes toward suicide and social safety net systems vary across countries and regions, our findings show the non-negligible magnitudes of mental and financial distress caused by the COVID-19-induced employment shocks. More precise interplay among unemployment, mental and financial distress, and safety net participation should be examined in future studies.
Data Availability
All the data and graphs presented in the paper will be available at https://github.com/michihito-ando.
Appendix
Appendices
A Suicides by age and occupation
B Variable definitions and Data sources
C Estimation results for robustness checks
As robustness checks, we present estimation results based on two model specifications (with or without individual linear trends ϕit in eqaution (2)), two regression schemes (OLS or WLS), and two types of outcome variables (monthly or monthly year-on-year difference). Note that our baseline estimation is based on WLS estimation with individual linear trends using a monthly outcome variable.
The rationales for these robustness checks are as follows. First, while incorporating individual linear trends may be effective for controlling for observed and unobserved differential trends across prefectures, it is useful to examine whether estimation results change if we use a simpler model without individual linear trends. Second, as Solon et al. (2015) argue, it is not clear a priori whether OLS or WLS regression is more suitable for estimating a population average effect, so we compare OLS and WLS estimates and discuss how they differ and why. Third, using a year-on-year outcome difference as an outcome variable is meant to control for prefecture-specific monthly fixed effects in addition to standard prefecture and monthly fixed effects in model (2).
Figures C.1, C.2 and C.3 provide additional estimation results using the two estimation methods (OLS or WLS), two model specifications (with or without prefecture linear trends), two outcome specifications (a monthly outcome or its year-on-year difference).
First, Figure C.1 shows the estimation results for the suicide rates. The left-hand graphs show estimates for monthly outcomes and the right-hand graphs present estimates for monthly year-on-year outcome differences. Results show that estimates are positive and often significantly different from zero for June and July 2020, but estimates with OLS regression or estimates for year-on-year outcomes tend to be smaller and sometimes statistically insignificant.
We interpret the smaller OLS estimates as reflecting heterogeneous employment-shock effects across prefectures: as Figure 2 indicates, prefectures with larger populations may be more clearly affected by the employment shocks and the WLS estimation put more weights on these prefectures. Smaller estimates for year-on-year outcomes suggest that COVID-19 employment shocks on suicide are not estimated robustly enough, but another interpretation is that controlling for prefecture-specific monthly fixed effects as well as prefecture and monthly fixed effects makes estimation more imprecise.
Second, Figure C.2 provides counterpart estimation results for unemployment benefits. It shows that the estimates for unemployment benefits are robustly not significantly different from zero in the COVID-19 period. However, gradually increasing OLS estimates in the COVID-19 period suggest the possibility that some positive impact of the employment shocks on unemployment benefits is not precisely captured by this data or this research design.
Third, Figure C.3 firstly shows that the baseline estimation results for the two temporary loan programs and Housing Security benefit are robust to different weighting schemes. Figure C.3 also indicates that the implications of the baseline results for Public Assistance recipients/recipient households do not change once individual linear trends are controlled for. However, pre-COVID-19 estimates for Public Assistance outcomes indicate the existence of differential trends if individual linear trends are not controlled for.
D An alternative treatment variable
In this appendix, we provide estimation results using an alternative employment-shock variable. More specifically, for the numerator of Xit in the employment-shock equation (1), we use the number of unemployed people who are registered as full-time-job seekers at public employment security offices called Hello Work, instead of the total number of the unemployed.
There are also some advantages of using this alternative employment-shock variable. First, registered unemployed people include all the unemployed who receive or will receive unemployment benefits, implying that this alternative employment-shock variable strongly reflects the COVID-19 shocks for those who are entitled to receive unemployment benefits.8 Second, it is expected that the registered unemployed who seek full-time jobs are often the breadwinners of households, not part-time or contingent workers. We thus may be able to capture the COVID-19 shocks for these core workers by using this alternative treatment variable.
Note that although Xit is quarterly in the baseline definition (1), we use the number of the registered full-time-job seekers in June for the second quarter and December for the fourth quarter. This is because the year-on-year total numbers of registered job seekers sharply dropped in April and May 2020 and increased in June 2020, probably due to the declaration of the State of Emergency in April and May 2020.9
As Figure D.1 shows, there is no clear correlation between the variations in the baseline employment shock and the alternative employment shock (R squared is 0.04 and the estimated slope is not significantly different from zero at the 10 % significance level). This is counter-intuitive given that registered unemployed people at Hello Work should be, at least conceptually, counted as unemployed people in the unemployment rate.
A plausible explanation of this unclear correlation is that a group of prefectures that have large manufacturing areas experienced the largest alternative employment shocks but relatively small baseline employment shocks, making the correlation obscure. In fact, in Figure D.1, the alternative employment shocks (Y axis) for some prefectures with major manufacturing regions are among the largest (e.g. Gifu, Aichi, Shizuoka, Shiga, Hiroshima, Toyama,and Mie). But the baseline employment shocks (X axis) for them are much smaller than the corresponding shocks for prefectures in metropolitan areas or with popular sightseeing spots that have high tertiary industry ratios (Okinawa, Kanagawa, Osaka, Nara, Hokkaido, Hyogo, Tokyo, Saitama, Chiba, and Kyoto).
In other words, the baseline employment shocks are strongly related to the tertiary industry whereas the alternative employment shocks are associated with the manufacturing industry. In fact, simple regression analyses show that there is significant positive correlation between the tertiary industry ratio and the baseline employment shocks and significant positive correlation between the manufacture industry ratio and the alternative employment shocks.
We provide estimation results using the alternative treatment variable in the same manner as Appendix C. First, Figure D.2 shows the estimation results for the suicide rates. Although some estimates in the COVID-19 period are significantly different from zero, estimates tend to fluctuate both before and during the COVID-19 crisis and we do not observe robust estimation results. This means that a prefecture having a higher alternative employment shock does not result in a higher suicide rate.
Second, Figure D.3 shows that estimates for total, female, and male unemployment benefits are robustly significantly different from zero in the third quarter (July-September) of 2020. These results are in contrast with those of robustness checks with the baseline treatment variable (Figure C.2), but this is understandable given the fact that the alternative employment shock is more directly related to the COVID-19 shocks on unemployed people who are eligible for the unemployment benefit than the baseline unemployment rate as mentioned above.
Third, Figure D.4 shows that estimates for the two temporary loan programs, the Housing Security Benefit and Public Assistance outcomes are positive and mostly significantly different from zero in the COVID-19 period. Although some non-zero estimates for Public Assistance outcomes in the pre-COVID-19 period suggest that some confounding differential trends may not be properly controlled for, the overall trends of the estimates imply that the alternative employment shock should have some positive impact on the numbers of Public Assistance recipients and recipient households.
Overall, we observe that the alternative COVID-19 employment shock measured by full-time-job seekers clearly affects safety net participation, but not suicide. Note that these results do not change when we further control for the baseline treatment variable. We speculate that this absence of an effect on suicide is related to the socio-economic status of full-time-job seekers, including their eligibility for unemployment benefits, but further investigation is required.
Footnotes
* This paper has previously circulated under the title “The impact of COVID-19 employment shocks on suicide and poverty alleviation programs: An early-stage investigation”. This work was supported by JSPS KAKENHI Grant Number JP20K01733. All the data and graphs presented in the paper will be available at https://github.com/michihito-ando, where the estimation results and graphs are presented in more interactive forms. We thank Shintaro Yamaguchi and seminar participants for useful comments and Masaya Waki for excellent research assistance. We are also grateful to Ren Onishi for sharing his data on some safety net programs. All errors are our own.
↵1 Japanese population estimate in July 2020 is 125,960,000
↵2 As of February 2020, the infection and death rates of COVID-19 in Japan have been relatively low in comparison to other developed countries.
↵3 See, among others, articles in the Japan Times such as “Suicide spike in Japan shows mental health toll of COVID-19” (9th October 2020, https://www.japantimes.co.jp/news/2020/10/09/national/social-issues/suicide-mental-health-coronavirus) and “Japan suicides rise as economic impact of coronavirus hits home” (11th November 2020, https://www.japantimes.co.jp/news/2020/11/11/national/japan-suicide-rise-coronavirus).
↵4 Statistics in February and March 2020 were missing from the original data provided by the government.
↵5 The number of loan decisions between April and July 2020 in the statistics provided by the central government is aggregated as follows: April data is based on the numbers from 25th March to 2nd May, May data is from 3rd May to 30th May, June data is from 31st May to 27th June, July data is from 28th June to 1st August, August data is from 2nd August to 29th August, and September data is from 30th August to 3rd October.
↵6 The number for January 2020 given by National Survey on Public Assistance Recipients.
↵7 These population statistics were obtained from official government websites.
↵8 To receive unemployment benefits, an unemployed person needs to be registered at Hello Work.
↵9 We also note that this alternative “unemployment rate” with the registered unemployed is not commonly used because the numerator (the number of registered unemployed job seekers) and the denominator (total labor force) are based on different surveys. We nonetheless argue that it is useful to use this alternative variable for the further examination of the COVID-19 shocks.