Who can work from home? The roles of job tasks and HRM practices

https://doi.org/10.1016/j.jjie.2021.101162Get rights and content

Highlights

  • Examine the determinants of remote work using a unique Japanese survey dataset.

  • Remote work were prevalent among those engaged in non-routine tasks.

  • Remote work were prevalent among workers subject to HRM practices.

  • The expansion of remote work opportunities during COVID-19 was inequality enhancing.

  • Policy implications are discussed.

Abstract

This paper examines the characteristics of remote work using a unique Japanese survey dataset that provides information on engagement in remote work together with the specific job tasks and human resource management (HRM) characteristics workers face. We show that the opportunity to work remotely was more likely to be available to those engaged in non-routine tasks as well as to workers subject to HRM practices presupposing that worker performance is measurable. The implications of these findings for income transfer policies and management practices in light of the COVID-19 pandemic are also discussed.

Introduction

Throughout 2020, the spread of COVID-19 and the implementation of social distancing policies have confined millions of workers to their homes. While primarily a public health measure, social distancing has had a profound economic impact, but its effect on the aggregate labor supply and any distributional consequences will be determined largely by the extent to which remote work is possible. Against this backdrop, there has been a heightened interest in better understanding which jobs can and are being performed remotely. Studies using job task descriptions to estimate the proportion of jobs that can technically be accomplished from home have arrived at numbers ranging from 37% in the US to 56% in Germany.

This study takes a different approach by documenting who actually worked from home just prior to the COVID-19 crisis and identifying their specific job characteristics. In contrast to the literature, this approach allows us to understand the conditions facilitating a remote work arrangement and to articulate any expected challenges as remote work expands. To accomplish this, we utilize a unique panel data set from Japan that includes job task characteristics and human resource management conditions. As of December 2019, 8% of Japanese workers worked outside of their official workplace, and while the prevalence of remote work does not vary much across industries, it does vary considerably across occupations. In particular, workers involved in non-routine and non-manual tasks and those employed under performance-based human resource management (HRM) practices are significantly more likely to engage in remote work. Together, these results indicate that workers engaging in manual or interactive tasks are less likely to work remotely, as previous studies have found, and also that the extent to which employers can prevent “shirking from home” by, for example, quantifying work output is a critical determinant in the adoption of remote work. We show that job task and HRM characteristics, which vary considerably within a given 2-digit occupation or 1-digit industry category, largely explain the remote work experience, conditioned on industry and occupation.

In addition, as the determinants of earnings and remote work potential are positively related, the burden of social distancing policies disproportionately falls on low income earners. For instance, while the 50% of consultants who worked remotely earned 6 million yen annually, the 10% of care workers who worked remotely earned only 2.8 million yen annually. More generally, those who worked remotely earned 23.4% more than non-remote workers. However, we find that remote workers did not earn more than non-remote workers when conditioned on observed job and demographic characteristics, which suggests that income transfer policies conditioned on observable characteristics or socioeconomic status would mitigate the disadvantage experienced by non-remote workers under the hardships caused by the pandemic.

Remote work increased from 8% in December 2019 to 14% in December 2020, and we further examine how the determinants of remote work evolved during the COVID-19 pandemic by examining the job characteristics associated with the increase in remote work using the latest wave of the panel survey implemented in December 2020. Exploiting the panel structure, we examine the impact of job characteristics on December 2019 on the engagement in remote work in December 2020, and find that as remote work expanded, workers engaging in interactive tasks, holding a university degree, and belonging to larger firms caught the wave, while workers engaging in manual tasks were left behind. Thus, the increase in remote work during COVID-19 exacerbated the pre-existing inequality.

Since the outbreak of COVID-19, numerous studies have investigated the potential for working remotely based on occupation. Dingel and Neiman (2020), for example, determine whether a job can be performed remotely from responses to an O*NET questionnaire on “work context” and “generalized work activities”. By aggregating feasibility according to the distribution of the 6-digit standard occupational classifications published by the U.S. Bureau of Labor Statistics, they conclude that 37% of U.S. jobs can be performed from home.1 Using a similar mapping, Boeri et al. (2020) find that 24-31 % of jobs can be performed at home in major European countries, and Holgersen et al. (2021), who determine the potential for working remotely through detailed ISCO-08 job descriptions and the marginal distribution of occupations through online job advertisements, similarly find that 36% of jobs in Norway can be performed from home. Meanwhile, Alipour et al. (2020) find from employee surveys that 56% of jobs in Germany can be performed from home. Finally, Brussevich et al. (2020) and Hatayama et al. (2020) independently calculate the possibility of remote work for more than 50 countries based on task characteristics recorded in the OECD Survey of Adult Skills (PIAAC) and find that per capita GDP and prevalence of remote work are positively associated. They also report that women, college graduates, and salaried and regular-contract workers have jobs that are more amenable to working from home than the average worker.

In addition to these studies of the potential for remote work based on occupation, research appears on the proportion of the workforce that is actually working from home in the midst of the COVID-19 crisis. Based on a survey conducted from April 1-5, 2020, Brynjolfsson et al. (2020) find that 15% of U.S. workers had already been working remotely and that 38% of those workers who formerly commuted were now working from home. Bick et al. (2021) reports that the proportion of newly remote workers rose from 8% in February to 35% in May, but Adams et al. (2020) notes substantial heterogeneity in the roll-out of remote work within industries and occupations. Meanwhile, in Japan, Okubo (2020) indicates that the percentage of remote workers in 2020 was 6% in January, 10% in March and 17% in June, but Morikawa (2020) finds the proportion in June to be higher, at 32%, based on a June 2020 survey.

This paper contributes to the literature by examining the relationship between job characteristics and the potential for remote work by exploiting the unique features of a Japanese worker panel data set which directly surveys whether a worker worked remotely. In addition, it records two measurable variables that are important determinants of the potential for remote work: the direct measurement of job task characteristics and the specific HRM characteristics of each employee’s work environment. The survey was implemented just before the outbreak of COVID-19 pandemic, and this makes this data set ideal for examining the natural determinants of the possibility for remote work. Further, identifying these natural determinants is useful for foreseeing the problems that might arise when employers are forced to adopt a remote work arrangement.

The high explanatory power of job characteristics variables has two important implications. First, on redistribution policy, the literature demonstrates that the negative impact of COVID-19 is heterogeneous across occupations and skill levels (Adams, Boneva, Golin, Rauh, 2020, Mongey, Pilossoph, Weinberg, 2020, Brussevich, Dabla-Norris, Khalid, 2020, Kikuchi, Kitao, Mikoshiba, 2021) and the likelihood of working remotely is strongly associated with a negative income shock (Mongey et al., 2020). Thus, our finding of a heterogeneous probability of engaging in remote work according to job tasks within a standard industry/occupation category implies that a heterogeneous negative income shock will exist even within the category. This therefore calls for a careful examination of the current income change and a much more fine-grained transfer policy to compensate for any earnings loss due to COVID-19.

Second, this study has important implications for the human resource management policies of private companies in terms of maintaining productivity. There is a strong relationship between HRM practices based on individual performance measures (such as pay for performance, management by key performance indicator, or management by objectives) and the potential for remote work, implying that remote work can occur only when output is observable and measurable (Allen, Golden, Shockley, 2015, Bloom, Liang, Roberts, Ying, 2015, Sewell, Taskin, 2015, Groen, van Triest, Coers, Wtenweerde, 2018).2 However, the output of certain jobs is inherently difficult to observe or quantify, and this potential for “shirking from home” is the major reason why some workers are not permitted to work remotely. Despite this concern, however, the sudden implementation of social distancing policies during the COVID-19 pandemic has forced firms to encourage remote work even for those workers in jobs deemed unsuitable for it. Thus, unless firms take measures to improve their observation of worker effort or output, the remote work phenomenon is likely to lead to reduced productivity in certain jobs, mainly because worker effort is difficult to sustain in the remote work environment.

Section snippets

Data

This paper uses the Japanese Panel Study of Employment Dynamics (JPSED), a panel survey with a standard set of demographic and labor market variables that has been conducted by the Recruit Works Institute every year since 2015. JPSED is a nationwide survey that is representative of all men and women over the ages of 15 years old and is conducted by an internet monitor registered to Intage Corporation.3

Job characteristics of remote workers

Before presenting our empirical specification for investigating the effects of task and HRM characteristics on the potential for remote work in Section 4, in this section, we discuss the relationship between remote work and specific job characteristics independently.

Determinants of remote work

The analysis thus far has discussed the relationship between remote work potential and various occupation and task characteristics independently. In this section, we analyze how task characteristics, basic demographic characteristics, and HRM practices together affect the status of working from home by estimating the following probit model:Yi*=Xiβ+Ziα+θind+ηocc+ϵiYi={1(Yi*0)0(Yi*<0)where Yi is a binary variable indicating whether the respondent engaged in remote work; Xi is a set of

Effect of remote work on earnings

To this point we have demonstrated that workers in high-earning occupations or with high human capital are more likely to engage in remote work. Now we ask if engagement in remote work affects earnings after conditioning on observable job and demographic characteristics. This exercise is important for deriving implications for government transfer policies, for if remote work and earnings are positively correlated through observed job and demographic characteristics, an income transfer policy

Concluding remarks

This paper analyses remote work in December 2019, the period just before the breakout of COVID-19. The opportunity to work remotely was more likely to be available to those in professional occupations characterized by non-routine, analytical and non-interactive tasks, and less likely to be available to service sector workers requiring face-to-face interactive tasks or manual laborers performing routine and manual tasks. Furthermore, workers subject to HRM practices that presume the

References (21)

There are more references available in the full text version of this article.

Cited by (25)

  • Working from home: Too much of a good thing?

    2024, Regional Science and Urban Economics
  • Expenditure responses to the COVID-19 pandemic

    2023, Japan and the World Economy
    Citation Excerpt :

    Hoshi et al. (2021) investigate the heterogeneous effect of a policy-induced decline in people’s mobility on the Japanese labor market outcome during the early COVID-19 period. Kawaguchi and Motegi (2021) examine the determinants of remote work using a unique Japanese survey dataset. Okubo (2022) studies demand-inducing policies targeting travel, and food and beverage industries in the background of COVID-19.

  • Telework in the spread of COVID-19

    2022, Information Economics and Policy
    Citation Excerpt :

    Using the survey by JILPT, Takami and Yamamoto (2021) found that workers increased telework use under the first state of emergency, but some occupations and tasks were not suited.8 Using JPSED,Kawaguchi and Motegi (2021) found that workers who engage in nonroutine tasks and human resource management tend to use telework, and then Ohtani (2021) found that telework did not greatly change task-sharing for housekeeping and childcare between wife and husband. These previous studies on Japan focus on which personal traits (gender, age, and occupation) affect telework and how telework changed work and lifestyle, using one or a few waves of surveys in the first state of emergency (April to May, 2020).

View all citing articles on Scopus

This paper does not represent an official position of Recruit Co., Ltd. We thank a referee for valuable comments and Akito Kamei, and also thank Philip C. MacLellan for editorial assistance. The authors are responsible for all remaining errors and interpretations.

View full text