Shen et al. (2021) investigate the effect of COVID-19 on criminal behavior in Japan using a difference-in-differences approach. Their study covers a 3-year time period from 2018 to 2020, using a database of publicly available crime statistics (Hanzai Toukei) from all 47 prefectures. While there is a growing, global evidence base exploring how the COVID-19 pandemic has affected criminal behavior, Shen et al. is the first to investigate this important social issue in Japan to date. Their key contribution is the finding that Japan's voluntary lockdown in April and May 2020 reduced both violent and property crime victimization rates significantly. Additional findings show a significant decline in sexual assault victimization for those below the age of 30, and that the 30–59 age group saw the greatest fall in crime victimization due to the COVID-19 pandemic.

In this context, I would like to offer some specific comments on Shen et al.'s approach to regional differences in the data, its conclusions on mental health, and, finally, its policy recommendations regarding COVID-19 resource allocation.

First, it is not clear why Shen et al. did not consider the regional variations in the data. It is evident from the Hanzai Toukei figures that the decrease in the number of arrests for “property crime” varied by region between 2018 and 2020. Shen et al. closely follows Leslie and Wilson (2020), whose analysis of data on police calls for service from 14 large metropolitan US cities showed that COVID-19 increased domestic violence. Given that city characteristics are unlikely to vary, regional differences in the Japanese data are something that should be emphasized. It would therefore be helpful to explain and justify why these differences were not explored further.

Second, I would like to comment on one of Shen et al.'s conclusions pertaining to mental health. Shen et al. state that, “Japan has shown that the lockdown had a positive effect on mental health when proxied by suicide rates.” The statistics do indeed show a decline in the number of suicides during the lockdown. However, the number started to increase after the lockdown. Shen et al. refer to Ueda et al. (2020), and Tanaka and Okamoto (2021). In fact, both papers conclude that the number of deaths by suicide declined during the initial phase of the pandemic (February to June 2020), but from July 2020 actually began to accelerate, exceeding suicide numbers in the previous year. The source of the increase was suicides among females, children, and adolescents. Furthermore, and somewhat disturbingly, the number of suicides recorded in October 2020 was higher than the total number of deaths due to COVID-19 from the entire months of February 2020 until October 2020. It would therefore be prudent to state the limitations of using suicide rates as a proxy for mental health. Although Shen et al. only analyzed data from January to May in 2020, they should be clear that, while their analysis suggests an initially positive effect on mental health during the lockdown (when proxied by suicide rates, which showed a significant fall during the voluntary lockdown period, primarily in suicides stemming from economic and living conditions), previous studies have shown that suicides increased in the period following the COVID-19 lockdown, and, thus, its medium-to-long-term effects on mental health present a much more complicated picture.

Finally, I wish to address one of the policy implications that Shen et al. put forward in their discussion. Having found that the lockdown reduced both violent and property crime rates, Shen et al. suggest that limited public resources could be reallocated from some sectors – such as police and criminal justice – to health care, in order to alleviate the added stresses on the health sector during the pandemic. However, is it right to say that, overall, the Japanese health sector was under such levels of stress? Ii and Watanabe (2021) analyze the impact of the first, second, and third waves of the COVID-19 pandemic on the health care system in Japan, and find a drastic decline in outpatient visits and inpatient admissions. Only about 3% of the total general beds in Japan were used for COVID-19 patients. Those hospitals which accepted COVID-19 patients have been under additional stresses, but many other clinics and hospitals have, in fact, suffered from a loss in demand. While there may need to be resource reallocation within the health sector, a diversion of resources from other sectors may not be necessary.

沉等人。( 2021 ) 使用差异中的差异方法调查 COVID-19 对日本犯罪行为的影响。他们的研究涵盖了从 2018 年到 2020 年的 3 年时间段，使用了所有 47 个县的公开犯罪统计数据（Hanzai Toukei）数据库。虽然有越来越多的全球证据基础在探索 COVID-19 大流行如何影响犯罪行为，但 Shen等人. 是迄今为止日本第一个调查这一重要社会问题的人。他们的主要贡献是发现日本在 2020 年 4 月和 5 月的自愿封锁显着降低了暴力和财产犯罪的受害率。其他调查结果显示，30 岁以下人群的性侵犯受害率显着下降，并且由于 COVID-19 大流行，30-59 岁年龄组的犯罪受害率下降幅度最大。

在此背景下，我想对 Shen等人处理数据中区域差异的方法、其对心理健康的结论以及最后关于 COVID-19 资源分配的政策建议提出一些具体评论。

首先，不清楚为什么沉等人。没有考虑数据中的区域差异。它从明显Hanzai Toukei附图，在逮捕的数目减少为“属性罪行” 2018年到2020年之间沉按区域改变等。密切关注 Leslie 和 Wilson（2020 年），他们对来自美国 14 个大城市的警察呼叫服务数据的分析表明，COVID-19 增加了家庭暴力。鉴于城市特征不太可能发生变化，日本数据的地区差异是应该强调的。因此，解释和证明为什么没有进一步探讨这些差异将是有帮助的。

其次，我想评论一下沉等人关于心理健康的结论之一。沉等人。声明说，“日本已经表明，以自杀率为代表的封锁对心理健康产生了积极影响。” 统计数据确实显示封锁期间的自杀人数有所下降。然而，在封锁之后，这个数字开始增加。沉等人。参考 Ueda等人。（2020 年）、田中和冈本（2021 年））。事实上，两篇论文都得出结论，在大流行的初始阶段（2020 年 2 月至 6 月），自杀死亡人数有所下降，但实际上从 2020 年 7 月开始加速，超过了前一年的自杀人数。增加的原因是女性、儿童和青少年的自杀。此外，有点令人不安的是，2020 年 10 月记录的自杀人数高于 2020 年 2 月至 2020 年 10 月整个月份因 COVID-19 导致的死亡总数。因此，谨慎地说明使用自杀的局限性率作为心理健康的代表。虽然沉等人. 仅分析了 2020 年 1 月至 5 月的数据，但他们应该清楚，虽然他们的分析表明，在封锁期间对心理健康产生了最初的积极影响（以自杀率为代表，在自愿封锁期间显着下降，主要是在源于经济和生活条件的自杀），之前的研究表明，在 COVID-19 封锁之后的时期内，自杀人数有所增加，因此，其对心理健康的中长期影响呈现出更为复杂的局面。

最后，我想谈谈沉等人的政策含义之一。在他们的讨论中提出。沉等人发现封锁降低了暴力犯罪率和财产犯罪率。建议可以将有限的公共资源从某些部门（例如警察和刑事司法）重新分配给医疗保健，以减轻大流行期间卫生部门承受的额外压力。然而，总体而言，日本卫生部门承受着如此程度的压力是否正确？Ii 和渡边 ( 2021)) 分析了第一波、第二波和第三波 COVID-19 大流行对日本医疗保健系统的影响，发现门诊就诊和住院人数急剧下降。在日本，只有大约 3% 的普通病床用于 COVID-19 患者。那些接受 COVID-19 患者的医院承受了额外的压力，但实际上，许多其他诊所和医院都遭受了需求损失。虽然可能需要在卫生部门内重新分配资源，但可能没有必要从其他部门转移资源。