在劳动力规模持续减少的背景下，面对人口红利渐失、劳动力成本上升，要素投入面临更大约束的严峻形势，如何提升人力资本水平并有效配置，有非常重要的意义。已有研究多从教育扩张、工资收入差距、生产率等角度研究人力资本配置问题，而从工作时间的角度看，失业、歇业或工时不足体现出人力资本的闲置，工时过长带来的生命健康危害、生产效率下降和工作家庭冲突造成人力资本不能取得最大回报，地区、职业、部门间的工时差异可能意味着闲暇福利的不平等和人力资本的错配，但国内相关研究由于数据来源的限制，仍十分缺乏。本文建立理论和实证模型，从增量配置角度讨论以受教育水平提高为主的人力资本提升如何影响工作时间；从职业配置角度讨论工时约束和工时过长如何影响不同受教育水平人群的劳动参与和职业选择；从空间配置角度探讨人力资本向城市的迁移和集聚如何影响工作时间的地区差异；最后从部门配置角度探讨劳动力市场外部负向冲击对不同所有制、产业部门劳动者短期失业和工时闲置的异质影响。主要研究结论如下：
       第一，我国劳动者受教育水平越高，工作时间越短，并适用于按性别、年龄、户籍、职业分组的各类人群。本科及以上教育程度劳动力周工作时间，比初中及以下平均低11.9%，比高中平均低10.3%，比大专平均低4.6%。分年龄组看，不同教育水平的工作时间差异在年轻组别（22—29岁）最小，在年老组别（40—59岁）最大。分性别看，女性接受高等教育后工作时间减少的幅度要明显大于男性。分户籍状况看，不同受教育水平间的工时差异，外来人口要大于本地人口，非农户籍要大于农业户籍。受教育水平较高的劳动者不仅从事了工作时间更短的岗位（以行业、职业、单位类型等界定），且其在相同的工作岗位上工作时间也更短；另外，更短的工作时间主要源自工时过长的占比下降，而非工时不足占比的提高。本科及以上周工时大于48小时的占比要比初中及以下低31.7%，比高中低20.0%，比大专低6.3%。在受教育水平与工作时间存在负相关关系的情况下，以月（年）工资收入衡量工资水平会明显低估教育回报率，教育的小时工资回报要明显高于月工资收入回报。
       第二，我国劳动力市场存在岗位工时约束现象，劳动者调整工作时间的主要方式是更换不同的工作，高教育水平已婚女性受家庭抚养责任影响，更可能选择工作时间更短的职业，或退出劳动力市场。工时过长的岗位占比对大专及以上已婚女性的劳动参与率以及在相关职业的就业份额均有显著的负向影响。与不需要照顾6岁及以下幼儿相比，已婚女性需要照顾的幼儿年龄越小，其劳动参与率受到的负面影响越大。工时过长的岗位占比提高，不会影响在相关职业就业的女性一年内退出劳动力市场的概率，但会导致大专及以上已婚女性更换一份工作时间相对较短职业的概率显著提高，对22—39岁年轻组的影响要大于40—59岁年老组，且家庭照料负担会放大其负面影响。
       第三，工作时间的城市间差异十分显著，并与城市化（urbanization，以城市人口密度衡量）和地方化（localization，以同行业就业人口密度衡量）经济密切相关，与工作分摊（work-spreading）和疯狂竞争（rat race）假说的预期相符。雇员工作时间对城市人口密度的弹性为负，对同行业就业密度的弹性为正；受教育程度越低，弹性系数的绝对值越大。年轻劳动力处于事业发展的初期，工作任务更容易被同质劳动者分摊，也更可能在晋升激励下参与疯狂竞争。对于高中学历组雇员，与疯狂竞争假说的预测一致，在同年龄组同职业的雇员间的工资不存在差异时，更可能分摊工作量，而不是互相竞争；另外，即使不存在同质人群间的竞争，工资差距也能激励劳动者工作更长的时间。但对于初中及以下组雇员，主要为获取更大的收入回报而提高工作时间，疯狂竞争机制似乎并不存在。
       第四，新冠肺炎疫情的负向冲击短期内也带来人力资本在产业间的再配置，居家办公可能和行业必要性在行业、职业、地区和人群间存在显著差异，导致保供必需或可以电子化办公的行业就业受影响较小，而非保供必需或接触性行业在短期内出现工时闲置，包括失业、在岗歇业和退出劳动力市场，工资水平更高的非接触性行业的人力资本占比提升，人力资本配置效率下降，因此可能扩大了产业部门间的福利差异。国有集体部门等更可能居家办公和从事必要行业的就业人员在疫情封控政策和配套的稳就业促就业保障措施下，更可能保留工作岗位，处于在职未上班或居家办公的状态，疫情期间工作时间投入和工资下降较少，而临时就业、自主就业人员则更可能因失去打零工等机会处于未就业或歇业状态。
       本文的研究可以得到以下几点理论和政策启示。第一，教育回报率的测算采用工资收入和小时工资会有较大差异，基于小时工资的收入回报能够更好的衡量个体福利的改善，但在我国薪酬多以月工资为支付方式，全日制工作缺少小时工资标准的情况下，分别测算工资收入和工作时间对教育回报的贡献有重要的现实意义。第二，与男性相比，女性工作与生活的平衡面临更大压力，“全面三孩”等生育政策优化后，不仅影响生育高峰期女性的劳动参与和职业选择行为，也因隔代照料负担变化影响非育龄期女性的劳动参与和职业选择。要注重女性劳动权益的保护，扩大养老托幼服务供给，鼓励企业实行更加灵活的工时制度，从而促进高素质女性就业。第三，城市间工时差异也意味着工资收入的地区差异并不必然代表地区间的福利差异，劳动力资源空间错配的程度要大于剩余工资差异所展现的水平，地区间教育回报率的差异也要考虑工作时间差异的贡献。劳动力往大城市迁移的原因可能不仅由于更高的工资收入，还因为有更多的闲暇。第四，人力资本配置要充分考虑到不同行业、职业、人群应对外部冲击的能力，数字技术如何为企业和劳动者赋能，尤其是人工智能等现代信息技术迅猛发展，不同群体间的“数字鸿沟”如何影响劳动者就业和应对供求冲击的能力，需要引起政策制定者高度关注。

Nowadays, with the continuous reduction of the scale of labor force, facing the gradual loss of demographic dividend, the rise of labor cost, and the severe situation of greater factor constraints, how to improve the level of human capital and effectively allocate it is of great significance. Existing research related to human capital allocation, are from the perspective of education expansion, wage income gap and productivity. From the perspective of working time, unemployment, closed or lack of hours reflects the idle human capital; long hours of work results in life and health hazards, productivity reduction and working family conflict, which means human capital cannot achieve the maximum return; regional, occupational, sectoral differences in hours of work means that leisure inequality and the mismatch of human capital; but due to the limitation of data sources, the domestic research is still very lack. This paper establishes both the theoretical and the empirical model, from the perspective of increment allocation, discuss how the improvement of human capital mainly with the improvement of education level affects the working hours; from the perspective of occupational allocation, discuss how time constraints and long hours affect the labor participation and career choice of people with different education levels; from the perspective of spatial allocation, discuss how the migration and agglomeration of human capital to cities affect the regional differences of working hours; finally, the heterogeneous impact of the policy of lockdown and social isolation under the impact of COVID-19 on the short-term unemployment and idle working hours of the workers of different ownership and industrial sectors is discussed. The main study conclusions are summarized as follows:

Firstly, the higher the education level of Chinese workers, the shorter their working hours, which applies to various groups of people grouped by gender, age, household registration, and occupation. The weekly working hours of laborers with a bachelor's degree or above are on average 11.9% lower than those with a junior high school education or below, 10.3% lower than those with a high school education, and 4.6% lower than those with a junior college education. In terms of age groups, the difference in working hours between different education levels is smallest in the younger group (22-29 years old) and largest in the older group (40-59 years old). In terms of gender, the reduction in working hours after receiving higher education is significantly greater for women than for men. In terms of household registration status, the difference in working hours between different education levels is greater for non-local residents than for local residents and greater for non-agricultural household registration than for agricultural household registration. Workers with higher education levels not only engage in positions with shorter working hours (defined by industry, occupation, and unit type), but also have shorter working hours in the same job position. In addition, the shorter working hours are due to a decrease in the proportion of excessive working hours rather than an increase in the proportion of insufficient working hours. The proportion of workers with a bachelor's degree or above who work more than 48 hours per week is 31.7% lower than that of workers with a junior high school education or below, 20.0% lower than that of workers with a high school education, and 6.3% lower than that of workers with a junior college education. Under the condition that there is a negative correlation between education level and working hours, measuring wage levels in terms of monthly (annual) income significantly underestimates the rate of return on education, and the hourly wage return on education is significantly higher than the return on monthly income.

Secondly, there is a phenomenon of hours constraint in China's labor market. The primary way for workers to adjust their working hours is to switch to different jobs. Married women with higher education levels are more likely to choose professions with shorter working hours or exit the labor market due to family caregiving responsibilities. The proportion of jobs with long working hours has a significant negative impact on the labor participation rate of married women with college degrees or above, as well as their employment share in related occupations. Compared with those who do not need to care for children aged 6 and below, the younger the children that married women need to care for, the greater the negative impact on their labor participation rate. An increase in the proportion of jobs with long working hours will not affect the probability of women employed in related occupations exiting the labor market within one year. However, it will significantly increase the probability of married women with college degrees or above switching to professions with relatively shorter working hours. The impact on the younger group aged 22-39 is greater than that on the older group aged 40-59, and family caregiving burden will amplify its negative impact

Thirdly, the difference in working hours between cities is significant and closely related to urbanization (measured by urban population density) and localization (measured by employment population density in the same industry). It is consistent with the expectations of the work-spreading and rat race hypotheses. The elasticity of employee working hours with respect to urban population density is negative, while the elasticity with respect to employment density in the same industry is positive. The lower the level of education, the greater the absolute value of the elasticity coefficient. The young labor force is in the early stage of career development, and work tasks are easier to be shared by homogeneous workers, and they are more likely to participate in rat race under the incentive of promotion. For high school employees, consistent with the prediction of the rat race hypothesis, employees are more likely to share work load with people in the same age group and occupation, and wage gap can motivate workers to work longer, even if there is no competition among homogeneous groups. But for employees in middle school and below, rat race seems to not exist, the reason for increasing working hours is mainly greater income returns.

Fourth is the COVID-19 epidemic has also brought about the reconfiguration of human capital between industries in the short term. There may be significant differences in the necessity of working from home among industries, occupations, regions and populations, resulting in less impact on employment in industries that are necessary for supply or can be electronically operated. In contrast, non-essential or contact-intensive industries have experienced idle working hours in the short term, including unemployment, on-the-job suspension and exit from the labor market. The proportion of human capital in non-contact-intensive industries with higher wages has increased, and the efficiency of human capital allocation has decreased, which may expand welfare differences between industrial sectors. Employees of state-owned collective departments and other jobs that are more likely to work from home and engage in necessary industries are more likely to retain their jobs under epidemic control policies and supporting employment protection measures. They are in a state of being employed but not working or working from home. During the epidemic period, their working time investment and wage decline less. Temporary and self-employed workers are more likely to be unemployed or suspended due to losing opportunities such as casual work.

The following theoretical and policy implications can be obtained from the research of this paper. One is the education yield measured using wage income and hourly wages will have big difference, based on hourly wage income returns can better measure the improvement of individual welfare, but in our country to pay monthly salary, full-time work lack of hourly wage standard, respectively measure wage income returns and leisure returns has important practical significance. Second, after the implementation of the "universal three-child" policy, the balance between work and life is under greater pressure, which not only affects women's labor participation and career choice behavior during childbirth peak period, but also affects the labor participation and career choice of women in non-childbearing age due to the changing burden of inter-generational care. We should pay attention to the protection of women's labor rights and interests, expand the supply of elderly care services, and encourage enterprises to implement more flexible working hour systems, so as to promote the employment of high-skilled women. Third, the difference in working hours between cities also means that the regional difference of wage income does not necessarily represent the difference in regional welfare. The degree of spatial mismatch of labor resources is greater than the level shown by the surplus wage difference, and the difference in educational return between regions should also consider the contribution of the difference in working hours. The migration of labor to big cities may be due not only to the higher wage income, but also to more leisure. Fourth, human capital allocation should fully consider the ability of different industries, occupations, and populations to respond to external shocks. How digital technology empowers enterprises and workers, especially the rapid development of modern information technologies such as artificial intelligence, and how the "digital divide" between different groups affects workers' employment and their ability to respond to supply-demand shocks need to be highly concerned by policy makers.