Productivity dynamics of work from home: Firm-level evidence

doi:10.21203/rs.3.rs-3299574/v1

Download PDF

Research Article

Productivity dynamics of work from home: Firm-level evidence

https://doi.org/10.21203/rs.3.rs-3299574/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 29 Feb, 2024

Read the published version in Journal of Evolutionary Economics →

You are reading this latest preprint version

This study documents the changes in work from home (WFH) practices since the onset of the COVID-19 pandemic using panel data from original firm surveys in Japan. Particular attention has been given to the productivity dynamics of WFH. The results indicate the following: First, compared to when the first state of emergency was declared in Spring 2020, at the end of 2021, the ratio of firms utilizing WFH and the intensity of WFH decreased substantially. Second, according to the firms’ evaluations, the mean productivity of WFH improved by more than 10 percentage points, although it was still approximately 20% lower than that of conventional workplaces. The selection effect arising from the exit from this practice among firms with low WFH productivity, and the improvement in productivity among WFH-continuing firms contributed almost equally to the average productivity growth. Third, the majority of firms are planning to discontinue WFH practices and revert to the conventional work style after the end of COVID-19, and the gap between firms’ plans and the desire of remote workers is widening.

JEL Classification: D24, J24, J81, M12, M54

COVID-19

work from home

productivity dynamics

amenity value

Following the onset of the COVID-19 pandemic, the number of firms and workers adopting work from home (WFH) practices increased substantially. Remote work (or telework) refers to work performed outside the usual workplace, including home, satellite office, or café. However, during the pandemic, governments have promoted WFH to ensure social distancing and reduce infections. Therefore, this study focused solely on WFH, including hybrid (partial) WFH as well as full-time WFH. Many studies have been conducted on WFH in parallel with its development. The trends in WFH practices during the COVID-19 pandemic, the characteristics of workers who perform WFH, and the effect of WFH on labor market outcomes have been well documented (e.g., Béland et al., 2022; Bick et al., 2020; Brynjolfsson et al., 2020; Janys et al., 2021; Petroulakis, 2023). In Japan, Kawaguchi and Motegi (2021), Kikuchi et al. (2021), Morikawa (2022, 2023), and Okubo et al. (2021) are examples of such studies. Overall, these studies indicate that high-skilled and high-wage white-collar workers tend to undertake WFH, which alleviates the negative impact of the pandemic on these workers.

Because WFH is an effective means of controlling the spread of infection and maintaining economic activity, many studies have found that its spread has mitigated its negative impact on GDP and employment (e.g., Eberly et al., 2021; Hoshi et al., 2021; Kawaguchi et al., 2022). Fujii and Nakata (2021) and Jones et al. (2021) used the susceptible–infectious–recovered (SIR) macro model to analyze the effects of WFH on the trade-off between output and infection. However, the effect of WFH in mitigating the trade-off between health and economic activity depends not only on the feasibility of WFH but also on its productivity.

Recently, the trade-off has become less severe as the impact of the pandemic has weakened and economic activity is normalizing around the world. However, WFH is still used at a substantially higher level than the pre-pandemic level. Against this backdrop, firms are exploring the appropriate use of WFH in normal times, with some firms trying to make this work style permanent, while others revert to traditional workplace work. Whether WFH will be widely used, even after the pandemic, depends heavily on the productivity of the work style.

However, as discussed in more detail in Section 2, studies on WFH productivity are still in their infancy and are far from reaching a consensus. The future course of WFH productivity depends on the mechanisms that improve productivity, but the productivity dynamics of WFH-decomposition of productivity change into “selection effects” and “learning effects”།are not well documented. Therefore, this study focuses on the productivity dynamics of the WFH using panel data from surveys of Japanese firms. In addition, by observing changes in firms’ plan to use WFH, we attempt to predict the prospects of WFH after the COVID-19 pandemic.

This study finds that, first, at the end of 2021, the ratio of WFH-utilizing firms and the intensity of WFH decreased substantially compared to when the first state of emergency was declared in 2020. Second, the mean productivity of WFH improved by approximately 11 percentage points, although it was still approximately 20% lower than that at usual workplaces. To decompose productivity improvement, the selection effect arising from the exit from this practice among firms with low WFH productivity (selection effects) and productivity growth among WFH-continuing firms (within-effects) contributed almost equally to improved productivity at the aggregate-level. Third, the majority of firms are planning to discontinue the WFH practice and revert to the conventional work style after the end of COVID-19, indicating that there is a large gap between firms’ intentions and the desires of remote workers. This result suggests a trade-off between the non-pecuniary benefit (or amenity value) of WFH for remote workers and the productivity of this work style.

The remainder of this paper is organized as follows: Section 2 overviews the literature related to WFH, focusing on its productivity. Section 3 explains the design of the survey used in this study. Section 4 reports the results on the prevalence, frequency, and productivity of WFH, as well as firms’ plans for WFH after the COVID-19 pandemic. Finally, section 5 summarizes the conclusions and discusses the implications of the study.

WFH was practiced even before the COVID-19 pandemic and was regarded as a flexible work style; however, with the exception of self-employed and family workers, very few workers engaged in WFH until recently. Following the sudden and rapid diffusion of WFH triggered by the pandemic, starting with the pioneering work of Dingel and Neiman (2020), many studies have examined the number of jobs that can be performed at home (e.g., Boeri et al., 2020; Brussevich et al., 2022). Dingel and Neiman (2020) estimated that 34 percent of jobs in the United States could be performed at home. Boeri et al. (2020) indicated that between 23% (Italy) and 32% (Germany) of the jobs could potentially be carried out at home in major European countries. Brussevich et al. (2022) estimated the WFH potential of 35 countries, covering developed and emerging economies, from 16% (Turkey) to 32% (Finland). They indicated that there are significant differences in WFH feasibility by job type and that countries with higher income levels have a larger percentage of WFH-feasible jobs.

It should be mentioned that the number of full-time remote workers is small and hybrid WFH that combine work at home and at workplace is prevalent. Adams-Prassl et al. (2022) and Alipour et al. (2023) make noteworthy contributions. Adams-Prassl et al. (2022), using survey data from the United States and the United Kingdom, examined the share of job tasks that can be performed from home on a continuous 0–100% scale. They show that the mean shares are 43% and 41% for the United States and United Kingdom, respectively. Alipour et al. (2023) estimate Germany’s overall capacity to work from home by including not only full-time WFH-feasible jobs but also partial WFH-feasible jobs, and find that 56% of jobs can be done from home, at least partially.

However, as Dingel and Neiman (2020) point out, productivity at home may differ significantly from that in regular workplaces. Therefore, the productivity of WFH needs to be assessed to estimate the quantitative effects of mitigating the trade-offs between infection risk and economic activity as well as to consider the outlook of WFH after the pandemic.

Theoretically, the effects of WFH on productivity can be both positive and negative (e.g., Deole, et al., 2023; Felstead and Reuschke, 2023; Van der Lippe and Lippényi, 2020). Worker productivity may increase owing to greater autonomy in allocating work time, the ability to concentrate on work without interruptions from colleagues, and reduced commute fatigue. In terms of overall firm productivity, the possibility of saving office space also contributes positively to TFP (Bloom et al., 2015). However, loss of face-to-face communication may negatively affect productivity, making it difficult to exchange informal tacit knowledge, build trust, and monitor workers. For employees who are suddenly forced to work from home by the COVID-19 pandemic, the limitations of their work environment and ICT infrastructure at home may also have a negative impact on their productivity. However, productivity may gradually improve through the diffusion of new ICT tools such as online meetings and through learning by experience.

Of course, WFH productivity differs by worker characteristics and depends on the nature of the occupation and the type of task, especially whether the work is self-contained or whether cooperation/coordination within a team is essential. Housing structure, family composition, and the personalities of workers also matter. Therefore, it is necessary to empirically clarify actual WFH productivity, how it is changing, and the factors that affect productivity.

Before the COVID-19 pandemic, Bloom et al. (2015), a representative study of WFH productivity, present evidence from a field experiment with call center operators in China that WFH enhanced the productivity of workers and organizations. However, since their study was based on a specific occupation in which WFH is relatively easy to implement, it is difficult to generalize their results to a wide variety of white-collar workers engaged in WFH during the pandemic. Van der Lippe and Lippényi (2020), using a survey in 2016 involving nine European countries, found that WFH reduced employees’ perceived efficiency. Dutcher (2012), based on a laboratory experimental approach, indicated that WFH may have a positive effect on productivity for creative tasks, but a negative impact on dull tasks.

Although not directly addressing WFH productivity, Battiston et al. (2021) and Atkin et al. (2022) demonstrate the importance of physical proximity and face-to-face communication. Battiston et al. (2021), exploiting a natural experiment with a public sector organization in the United Kingdom (the Greater Manchester Police), find that productivity was higher when teammates were in the same room, particularly for urgent and complex tasks, and interpreted teleworking as unsuitable for tasks requiring face-to-face communication. Atkin et al. (2022), using smartphone data to measure face-to-face interactions between workers in Silicon Valley, indicated that face-to-face meetings significantly contribute to knowledge flows between workers.

Studies dealing with the impact of WFH on coordination within organizations include those by Teodorovic et al. (2022) and Van der Lippe and Lippényi (2020). Van der Lippe and Lippényi (2020) analyzed the impact of WFH on workplace team performance based on a survey conducted in 2016 and found that teams work less efficiently when there are more colleagues working from home. Teodorovic et al. (2022), using time use survey, argue that the rapid shift to WFH associated with the COVID-19 pandemic increased coordination costs in the form of increased time devoted by managers to meetings.

Although studies on the productivity of WFH after the COVID-19 outbreak remain relatively scarce, Aksoy et al. (2022), Barrero et al. (2021), Etheridge et al. (2020), Felstead and Reuschke (2023), Kitagawa et al. (2021), and Morikawa (2022) are studies based on surveys of individual workers. Barrero et al. (2021), using survey data from the United States, documented that most respondents who have used WFH practices report productivity equal to or higher than that of business premises. Aksoy et al. (2022) extended a similar survey to workers in 25 countries and found that the productivity of WFH was, on average, 7% higher than expected. Etheridge et al. (2020), using survey data from the United Kingdom, reported that the mean productivity of WFH, on average, is not different from productivity in the usual workplace, although the productivity of WFH is quite heterogeneous by worker characteristics. Felstead and Reuschke (2023), based on a survey of workers in the United Kingdom, reported that about 70% of telecommuters reported no decrease in productivity as of June 2020 and about 85% as of September 2020.¹ Kitagawa et al. (2021), in a survey of employees from four large Japanese manufacturing firms, indicated that for the majority of employees engaged in WFH, productivity decreased relative to employees who did not use WFH. Morikawa (2022), using a survey of workers in Japan, reported that the mean WFH productivity relative to working in a usual workplace was approximately 60–70%.

Because these studies cover a wide range of occupations, productivity measures are based on workers’ self-assessments. Studies that use objective productivity measures include those of Bloom et al. (2022), Gibbs et al. (2021), Shen (2023), and Emanuel and Harrington (2023). Gibbs et al. (2021), using the achievement rate of assigned tasks divided by working hours as a measure of productivity, reported that measured productivity decreased by approximately 20% in a large IT firm in Asia. Bloom et al. (2022), based on a randomized control trial (2021–2022) for IT-related engineers of a large firm headquartered in China, indicated that physical productivity measured as the lines of computer code written increased by approximately 8% after the adoption of hybrid WFH (option to WFH on two days a week), although the increase arose mainly from the performance on non-WFH days. Using data from a large open-source software platform (GitHub), Shen (2023) finds a negative but almost negligible change in individual-level output during state-imposed workplace closures during the COVID-19 pandemic. Emanuel and Harrington (2023), using data for workers in a U.S. Fortune 500 firm’s call centers, find that the physical productivity of formerly on-site workers declined by 4% after the closure of the call centers due to COVID-19. However, these studies have the limitation of focusing only on IT-related workers or call center operators, whose output can be quantitatively measured, and whose tasks are suited to be conducted at home, making it difficult to generalize to workers in other occupations, such as clerical and managerial positions.

In summary, studies on WFH productivity after the onset of the COVID-19 pandemic using worker-level data have produced very different results.

Bartik et al. (2020) and Morikawa (2022) are examples of studies using firm surveys. Bartik et al. (2020) report that on average, WFH reduced productivity by approximately 20%, based on a survey of small and medium-sized businesses in the United States. Morikawa (2022) indicates that among Japanese firms, the mean productivity of WFH is approximately 68% of productivity in their usual workplace, and that the lack of face-to-face interactions, poor telecommunication environment at home, and tasks that must be conducted in the office are the major reasons of lower productivity at home. However, both surveys were conducted in the first half of 2020, and WFH productivity may have changed through learning by experience and WFH-related investments as the COVID-19 pandemic prolonged. In addition, firms with low WFH productivity may selectively exit WFH practices. In this respect, analyzing the change in productivity of this work style using panel data is important for evaluating the efficacy of WFH.

In the industrial organization literature, many studies use firm- or establishment-level panel data to analyze productivity dynamics (e.g., Baily et al., 1992; Foster et al., 2001, 2006; Griliches and Regev, 1995). ² These studies decompose productivity growth at the aggregate-level into “within-effects” and “selection/reallocation effects.” Within-effects reflect the productivity growth of individual firms, and selection/reallocation effects arise from, for example, the entry of productive firms and the exit of unproductive firms. Studies on productivity dynamics generally indicate that both mechanisms contribute to productivity growth at the aggregate-level.

However, to the best of our knowledge, no studies have used firm-level panel data to analyze the productivity dynamics of WFH. If there is large potential for increasing the productivity of WFH through within-effects, this work style is likely to prevail even after the pandemic. However, if selection effects dominate, the number of firms that continue to utilize WFH is likely to be limited. In this respect, unraveling productivity dynamics provides valuable information for drawing inferences about the future of this work style.

In addition to productivity, amenity value to workers also affects the prevalence of WFH. Amenity value is often measured by workers’ willingness to pay (WTP) or compensating wage differentials. In normal times, the amenity value of WFH is estimated to be 5–10% of wages (e.g., He et al. 2021; Mas and Pallais, 2017). Studies after the COVID-19 pandemic have generally indicated that the value is not significantly different from that during normal times (e.g., Aksoy et al., 2022; Lewandowski et al., 2022; Moens et al., 2022).³ When firms make decisions on WFH strategy after the pandemic, productivity is obviously important, but firms may also consider the amenity value of WFH to maintain good labor-management relations.

This study contributes to the research field by documenting the productivity dynamics of WFH since the onset of the COVID-19 pandemic using panel data constructed from original firm surveys in Japan. Another contribution is presenting evidence of the change in firms’ plans to utilize WFH after the pandemic to clarify the gap with employees’ desires.

This study uses data from the “Survey of Corporate Management and Economic Policy” (SCMEP), designed by the author of this paper, and conducted by the Research Institute of Economy, Trade, and Industry (RIETI) in 2020 and 2021. The implementation of the SCMEP was contracted from RIETI to Tokyo Shoko Research Ltd. The SCMEP sample firms were selected from the registered list of the Basic Survey of Japanese Business Structures and Activities (BSJBSA) conducted by the Ministry of Economy, Trade, and Industry (METI). The BSJBSA is a representative government survey of all Japanese firms with 50 or more regular employees and capital of at least 30 million yen engaged in the mining, manufacturing, electricity and gas, wholesale, retail, and selected service industries. Approximately 30,000 firms are annually surveyed. Because firms registered in the BSJBSA have at least 50 employees, the SCMEP does not include small firms.

The 2020 SCMEP, a follow-up survey of the 2019 SCMEP, was conducted between August and September 2020. The 2020 SCMEP was sent to 2,498 Japanese firms that responded to the SCMEP in 2019.⁴ The responses of these firms to the 2020 SCMEP totaled 1,579. The WFH questions asked about the situation in the early phase of the COVID-19 pandemic when the government declared its first statement of emergency. The 2021 SCMEP was conducted from October to December 2021 immediately after the fourth statement of emergency was lifted. The 2021 SCMEP was sent to 15,000 firms selected from the registered list of the 2019 BSJBSA, including those that responded to the 2020 SCMEP. The number of firms responded to the 2021 SCMEP is 3,194, of which 961 responded to both the 2020 and 2021 surveys (hereinafter referred to as “panel firms”). A comparison of SCMEP- and BSJBSA-registered firms is presented in Appendix Table A1. In the following, we mainly use data from all firms that responded to the SCMEP, unless otherwise noted; however, the analysis of the dynamics of WFH is conducted for a subsample of panel firms.

The major survey questions related to WFH included the utilization of WFH practices, percentage of workers using WFH (coverage), mean frequency of WFH per week, mean productivity of WFH workers relative to their productivity at the usual workplace (office), and firms’ plans to continue WFH after the COVID-19 pandemic.⁵ The survey question simply refers to “work from home,” so interpretation is left to the respondent firms, but it does not include teleworking outside the home. Since the survey was conducted while the government recommended not attending workplaces to ensure social distancing to reduce infection, there is little room for misunderstanding the question. Since the frequency per week of WFH was also asked, it naturally included hybrid (partial) WFH as well as full-time WFH.

In addition, the SCMEP collects information on various firm characteristics such as industry (manufacturing, information, and communications (I&C), wholesale, retail, services, and other industries), firm size (number of employees), composition of employees (female ratio, ratio of non-standard employees, and ratio of employees with university education or higher), existence of labor unions, and location of headquarters. These firm characteristics were used in the analysis.

This study documents the overall changes in WFH practices during the COVID-19 pandemic by linking data from these two surveys and conducting simple regressions (OLS and probit estimations) to analyze the relationships between various firm characteristics and the utilization, frequency, and productivity of WFH practices. Appendix Table A2 lists the major variables used in the regressions, along with their summary statistics.

4.1 Utilization and intensity of WFH

The SCMEP asked whether the firm utilized the WFH practice. The percentages of firms using WFH in the 2020 and 2021 surveys are summarized in Table 1. For all firms that responded to the survey, this percentage decreased by approximately 15%, from 49.5% in 2020 to 34.5% in 2021. When the sample was limited to panel firms that responded to the two surveys, the WFH utilization rate decreased from 47.0–28.8%. The table also shows the WFH utilization rate of the panel firms by industry. The utilization rate is very different by industry but decreases between 2020 and 2021 in every industry.

We can calculate the transition rate between WFH utilization and non-utilization for the sample of panel firms (961 firms): 26.0% continued WFH practices, 50.4% did not utilize WFH continuously, 20.9% exited WFH practices, and 2.7% newly adopted WFH practices. Firms reporting lower WFH productivity in the 2020 survey tend to become non-utilizers in the 2021 survey as explained in the next subsection; the mean WFH productivity in 2020 of firms continue WFH and those discontinue WFH are 73.9 and 58.3, respectively, suggesting that a natural selection mechanism is functioning.

Table 2 presents the probit estimation results for the relationship between observable firm characteristics and WFH utilization for all firms responded to the SCMEP.⁶ Manufacturing is the reference category for the industry dummies. The characteristics of firms utilizing WFH were essentially the same in the 2020 and 2021 surveys, with the exception of labor unions.⁷ Large firms, firms belonging to the information and communications (I&C) industry, firms headquartered in densely populated prefectures, and firms with a high share of employees with university or higher education have a higher probability of utilizing WFH. However, small firms, firms in the retail industry, and firms with a high share of non-standard employees are less likely to utilize WFH. Interestingly, the absolute size of the coefficients was generally smaller in 2021, suggesting that selection by factors other than these observable firm characteristics occurred between 2020 and 2021. For example, even within the same industry, some firms have many operations that are feasible to conduct remotely, while others do not. During the prolonged period of the pandemic, only firms with many operations suitable for WFH continued to use this work style.

Even if a firm utilizes WFH practices, this work style is not necessarily applied to all employees, and the coverage of remote work differs by firm. The question regarding the coverage of WFH is “What percentage of your employees use WFH practice?” Row A of Table 3 presents the tabulation results. The mean percentage of employees engaged in WFH decreases from 30.7% in 2020 to 21.2% in 2021, although the change is small for the panel firm sample (from 27.1–23.4%). Looking at the subsample of firms continuously utilizing WFH, the coverage level is relatively high; however, it decreases from 32.9% in 2020 to 24.8% in 2021. These figures suggest that many firms increased the share of employees working in their usual workplaces following the reduced risk of infection and lifting of the state of emergency. When looking at firms exiting WFH in 2021, the share of employees engaged in WFH was 20.0% in the 2020 survey, which was lower than that of firms continuously utilizing WFH (32.9%). As previously stated, the number of firms newly utilizing WFH practices in 2021 is small, and the coverage of employees engaged in WFH in 2021 is only 9.1%.

Column (1) of Table 4 reports the OLS estimation results to explain the coverage of employees engaged in WFH in 2021 by firm characteristics for the sample of all WFH-utilizing firms (the estimation result of the 2020 survey is presented in column (1) of Appendix Table A3). Coefficients for firm size, industry, location, share of nonstandard employees, and share of employees with university education or higher, most of which are associated with the utilization of WFH practices reported in Table 2, are significant for the coverage of WFH. In other words, these firm characteristics relate to employees’ use of WFH through both extensive and intensive margins, although the absolute size of the coefficients was generally smaller in 2021.

Even if employees use WFH, they do not necessarily work at home every day. The question on the mean frequency of WFH is “For those employees who are working from home, what is the number of days per week on average that they do so?” Row B of Table 3 reports the tabulation results. The mean frequency decreased by 1.2 days, from 3.67 days in 2020 to 2.47 days in 2021. Since the answer is the average of the firm’s employees engaged in WFH, the frequency distribution within the firm is unknown, but the results suggest that hybrid WFH, in which employees commute to workplaces two or three days a week, was prevalent. Even when limiting the sample to firms continuously utilizing WFH practice, the mean frequency decreased about a day, from 3.87 in 2020 to 2.88 days in 2021. The results suggest that as the number of infections decreased, the state of emergency was lifted, and the government’s request for WFH weakened, there was more room for firms to adjust the frequency to the optimal level, and as a result, even employees who engaged in WFH increased the number of days working at their workplaces.

Column (2) of Table 4 shows the OLS estimation results of the 2021 survey to explain WFH frequency by firm characteristics (the estimation results of the 2020 survey are presented in Column (2) of Appendix Table A3). Most of the observable firm characteristics are insignificant, and the overall explanatory power is very limited; however, the coefficient of the I&C industry is positive and highly significant, indicating that this industry is exceptionally suitable for WFH.

By multiplying the coverage of WFH employees by the frequency of WFH (expressed as percentages), we can calculate “WFH intensity,” which is the ratio of WFH hours to total working hours.⁸ The aggregate results are presented in Row C of Table 3. The WFH intensity decreased significantly, from 23.7% in the 2020 survey to 10.8% in the 2021 survey. However, when limiting the sample to firms continuously utilizing WFH practices, the reduction in WFH intensity is relatively small (from 33.1–22.2%), and the level of WFH intensity is relatively high even in the fourth quarter of 2021. The OLS estimation results explaining WFH intensity in 2021 by firm characteristics are reported in Column (3) of Table 4 (the estimation result of the 2020 survey is presented in Column (3) of Appendix Table A3). The results are generally similar to the estimation of WFH coverage; the coefficients for industry, population density of the headquartered prefecture, and share of employees with university or higher education have statistically significant associations with WFH intensity, although the absolute size of the coefficients was generally smaller in 2021.

The contribution of WFH hours to the total labor input at the aggregate-level can be calculated using the number of firms’ employees as a weight and including WFH non-utilizers whose WFH intensity is regarded as zero. Table 5 presents the calculation results by industry. The results for all industries decreased from 18.1% in the 2020 survey to 6.1% in the 2021 survey.⁹ The figure for the I&C industry is the highest; however, even in this industry, the contribution of the WFH decreases from 45.9–22.5%. The figure for the retail industry was the lowest at 5.1% from the beginning; however, in the 2021 survey, it decreased to 1.2%. This confirms that industry characteristics strongly influence WFH use.

4.2 Productivity dynamics of WFH

The question regarding the productivity of WFH is the firms’ subjective evaluation of their remote workers’ mean productivity at home relative to productivity at the office. The specific wording of the question was, “If the productivity of your employees normally achieved in the workplace is 100, roughly how much is their productivity when working at home? Please respond with the average number of all tasks specified to be performed from home.” It is noted that “if your employees are more productive at home than at the workplace, please write a number greater than 100.”

Table 6 presents the tabulation results, showing that the mean WFH productivity of firms utilizing WFH practices improved by 11.5 points from 68.3 in the 2020 survey to 79.7 in the 2021 survey (Row A). ¹⁰ However, there is a large dispersion in firm evaluations: the standard deviation is 23.5 in 2020 and 20.3 in 2021. This result is essentially the same for the subsample of panel firms (Row B). When limiting the sample to firms continuously utilizing WFH practices (Row C), productivity improved by 5.5 points from 73.9 in 2020 to 79.4 in 2021. This figure corresponds to the contribution of “within-effects” in the analysis of productivity dynamics.

Our interpretation is that the improvement in productivity of firms continuing WFH arises from the learning effect and reallocation of work within the firm, such as returning employees and/or tasks with relatively low productivity at home to their usual workplaces. The decrease in the share of employees engaged in WFH and the WFH frequency is consistent with such an internal reallocation mechanism. In other words, it is likely that during the COVID-19 pandemic, WFH exceeded optimal levels, and tasks that were efficiently carried out at the workplaces were also performed at home. Given that WFH is generally a hybrid type rather than full-time, the reshuffling of home and workplace use at the task level is likely to continue.

As the figure shows productivity at home relative to the office (= 100), the firms’ mean evaluation of remote workers’ productivity at home is still approximately 20% lower than productivity at usual workplaces. When dividing the responses into WFH > office, WFH = office, and WFH < office, the percentage of firms that evaluated productivity at home as lower than at the office was 92.3% in the 2020 survey and 77.8% in the 2021 survey. In the 2021 survey, the percentage of firms evaluating no difference in productivity between home and office work increased, but most firms rated WFH as less productive than office work. The results suggest that technical and institutional factors reduce the efficiency of WFH and that face-to-face information exchange is still important, even if various online communication tools have become available.

When looking at firms that utilized WFH practices in 2020 but exited from the practice in 2021, WFH productivity in 2020 was 58.3, which is far lower than that of firms continuously utilizing WFH (73.9). This result suggests that firms evaluating WFH productivity as lower selectively exit from the WFH practice.¹¹ Quantitatively, the contribution to the mean productivity of WFH arising from firms with low WFH productivity quitting WFH is 6.9 points, which is larger than the contribution of within-effects (5.5 points). The number of firms that did not utilize WFH practices in 2020 and started WFH in 2021 was small, with only 2.7% of firms responding to the two surveys. Their WFH productivity in 2021 was 64.8, which is lower than that of firms that continuously utilize WFH (79.4). Therefore, new entrants slightly reduced the mean WFH productivity in 2021; the quantitative contribution to the mean WFH productivity in 2021 was − 1.4 points. Therefore, the total contribution of the “selection effects,” the sum of exit from and entry into WFH, is 5.5 points. In short, the within and selection effects contributed almost equally to the improvement in the mean WFH productivity (11.1 points).

The OLS estimation results explaining WFH productivity based on observable firm characteristics for the sample of all firms utilizing WFH is reported in Appendix Table A5. According to the results of the 2021 survey, firms in the I&C industry and those with a high share of employees with university or higher education tend to evaluate WFH productivity as higher. However, the coefficients of firm size, retail trade, and population density, which are significant in 2020, are insignificant in 2021. The results of not only the higher WFH intensity of I&C firms and firms with a large share of highly educated employees but also the significantly higher WFH productivity of such firms suggest that tasks that require intensive computer use and cognitive skills are compatible with working remotely.

4.3 Firms’ plan to utilize WFH after the COVID-19 pandemic

Finally, the survey asked about the firms’ plans to use WFH after the pandemic. The specific question was “How do you think the WFH practice after the COVID-19 pandemic subsides?” The three choices are “We will utilize WFH the same or more than when there is the impact of COVID-19,” “We will continue to utilize WFH even after the end of COVID-19, but for fewer employees and/or fewer days,” and “As a rule, we will return to working at the usual workplace as before COVID-19.”

Table 7 presents the tabulation results; the percentage of firms that responded “the same or more” increased slightly from 12.9% in the 2020 survey to 15.6% in the 2021 survey (Row A column (1)).¹² For the subsample of firms that continue WFH (Row C), the percentage is relatively high, but the change is small (from 20.8–22.8%). In both the 2020 and 2021 surveys, the majority of firms chose “return to working at the usual workplace,” and more than 30% of firms intend to reduce the coverage of employees and/or the WFH frequency (columns (2) and (3)). These figures indicate that a large majority of Japanese firms intended to reduce their WFH substantially once the COVID-19 pandemic subsided.

This is in sharp contrast to the findings of the employee survey reported by Morikawa (2023), which indicates that the percentage of remote workers who want to continue frequent WFH substantially increased from 38.1% in 2020 to 62.6% in 2021, suggesting a non-pecuniary benefit or amenity value of WFH for remote workers.¹³ There is a large gap between employers and employees regarding their intentions to use WFH practices after the COVID-19 pandemic.¹⁴ As the share of employees who want to continue frequent remote work after the end of the COVID-19 pandemic increases, the gap widens between 2020 and 2021.

As indicated in the previous subsection, many firms have evaluated that workplace productivity is higher than WFH productivity. Therefore, firms’ plan to reduce WFH intensity is unsurprising, but remote workers’ own productivity may not be the only one reason. As referred in Section 2, some studies suggest that WFH has a negative impact on team performance. In this respect, it is possible that firms are planning WFH policies based not only on the productivity of individual workers, but also on the spillover effects on coworkers and the performance of workplace teams.

From the viewpoint of the balance between productivity and wages as well as the theory of compensating wage differentials, a decline in the relative wages of remote workers is expected. As mentioned in Section 2, many studies have examined workers’ WTP for remote work (e.g., He et al. 2021; Mas and Pallais, 2017). In addition, some studies indicate that remote work has a negative impact on wages in the United States (e.g., Barrero et al., 2022; Golden and Eddleston, 2020; Kouki and Sauer, 2022; Oettinger, 2011). However, in practice, it is extremely difficult to accurately evaluate the productivity and amenity value of remote workers. Therefore, there will be serious conflicts between employers and employees regarding the use of WFH after the pandemic.

This study uses panel data from original firm surveys to document changes in the utilization, intensity, and productivity of WFH since the onset of the COVID-19 pandemic in Japan. Particular attention was paid to the productivity dynamics of WFH. Additionally, this study presents evidence of the change in firms’ plans to use WFH after the pandemic to clarify the differences with employees’ desires.

The major findings are summarized as follows: First, at the end of 2021, both the ratio of WFH-utilizing firms and WFH intensity decreased substantially compared with when the first state of emergency was declared in Spring 2020. Second, the mean productivity of WFH improved by approximately 11 percentage points through the learning effect, reallocation of tasks within firms, and the exit of low WFH productivity firms from this practice. However, firms’ evaluation of productivity at home was still approximately 20% lower than that at the usual workplace, which is quite similar to the results obtained from employee surveys. Third, the majority of firms are planning to discontinue WFH practices and revert to the conventional work style after the end of the COVID-19 pandemic, indicating that there is a large gap between firms’ intentions and the desire of remote workers.

These results indicate that a natural selection mechanism based on productivity is functioning. We expect that firms with low WFH productivity will continue to return to traditional workplace work, select WFH-eligible workers, and reallocate tasks at home and in the workplace to establish optimal hybrid WFH practices. As a result, although the percentage of workers engaged in WFH will continue to decline gradually, those who are productive at home and those who appreciate the amenity value of this work style will continue to work at home frequently. WFH, especially the hybrid type, is expected to continue at higher levels than before the COVID-19 pandemic. In this process, through the balance between productivity and wages as well as the mechanism of compensating wage differentials, wage adjustment could occur, which would lower the relative wages of remote workers.

Disclosure Statement

I declare that the research in this paper is supported by the Japan Society for the Promotion of Science (JSPS) Grants-in-Aid for Scientific Research (18H00858, 20H00071, 21H00720, 23K17548).

Data availability

The data used in this study are available from the author upon request.

Adams-Prassl, A., Boneva, T., Golin, M., Rauh, C., 2022. Work that can be done from home: Evidence on variation within and across occupations and industries. Labour Economics 74, 102083.
Aksoy, C.G., Barrero, J.M., Bloom, N., Davis, S.J., Dolls, M., Zarate, P., 2022. Working from home around the world. Brookings Papers on Economic Activity, Fall 2022, 281–360.
Alipour, J.V., Falck, O., Schüller, S., 2023. Germany’s capacity to work from home. European Economic Review 151, 104354.
Atkin, D.M., Chen, K., Popov, A., 2022. The returns to face-to-face interactions: Knowledge spillovers in Silicon Valley. NBER Working Paper 30147.
Baily, M.N., Hulten, C., Campbell, D., 1992. Productivity dynamics in manufacturing plants. Brookings Papers: Microeconomics 1992, 187–249.
Barrero, J.M., Bloom, N., Davis, S.J., 2021. Why working from home will stick. NBER Working Paper 28731.
Barrero, J.M., Bloom, N., Davis, S.J., Meyer, B., Mihaylov, E., 2022. The shift to remote work lessens wage-growth pressures. NBER Working Paper 30197.
Bartelsman, E.J., Doms, M., 2000. Understanding productivity: Lessons from longitudinal microdata. Journal of Economic Literature 38(3), 569–594.
Bartik, A.W., Cullen, Z.B., Glaeser, E.L., Luca, M., Stanton, C.T., 2020. What jobs are being done at home during the Covid-19 crisis? Evidence from firm-level surveys. NBER Working Paper 27422.
Battiston, D., Vidal, J.B.I., Kirchmaier, T., 2021. Face-to-face communication in organizations. Review of Economic Studies 88(2), 574–609.
Béland, L.-P., Brodeur, A., Wright, T., 2022. The short-term economic consequences of COVID-19: Exposure to disease, remote work and government response. Canadian Journal of Economics 55(S1), 214–247.
Bick, A., Blandin, A., Mertens, K., 2020. Work from home after the Covid-19 outbreak. CEPR Discussion Papers 15000.
Bloom, N., Liang, J., Roberts, J., Ying, Z.J., 2015. Does working from home work? Evidence from a Chinese experiment. Quarterly Journal of Economics 130 (1), 165–218.
Bloom, N., Han, R., Liang, J., 2022. How hybrid working from home works out. NBER Working Paper 30292.
Boeri, T., Caiumi, A., Paccagnella, M., 2020. Mitigating the work-safety trade-off. Covid Economics 2, 60–66.
Brussevich, M., Dabla-Norris, E., Khalid, S., 2022. Who will bear the brunt of lockdown policies? Evidence from tele-workability measures across countries. IMF Economic Review 70(3), 560–589.
Brynjolfsson, E., Horton, J.J., Ozimek, A., Rock, D., Sharma, G., TuYe, H.-Y., 2020. COVID-19 and remote work: An early look at US data. NBER Working Paper 27344.
Deole, S.S., Deter, M., Huang, Y., 2023. Home sweet home: Working from home and employee performance during the COVID-19 pandemic in the UK. Labour Economics 80, 102295.
Dingel, J.I., Neiman, B., 2020. How many jobs can be done at home? Journal of Public Economics 189, 104235.
Dutcher, E.G., 2012. The effects of telecommuting on productivity: An experimental examination. The role of dull and creative tasks. Journal of Economic Behavior & Organization 84(1), 55–363.
Eberly, J.C., Haskel, J., Mizen, P., 2021. ‘Potential capital,’ working from home, and economic resilience. NBER Working Paper 29431.
Emanuel, N., Harrington, E., 2023. Working remotely? Selection, treatment, and the market for remote work. FRB New York Staff Report 1061.
Etheridge, B., Tang, L., Wang, Y., 2020. Worker productivity during lockdown and working from home: Evidence from self‑reports. Covid Economics 52, 118–151.
Felstead, A., Reuschke, D., 2023. A flash in the pan or a permanent change? The growth of homeworking during the pandemic and its effect on employee productivity in the UK. Information Technology and People 36(5), 1960–1981.
Foster, L., Haltiwanger, J., Krizan, C.J., 2001. Aggregate productivity growth: Lessons from microeconomic evidence. In C.R. Hulten, E.R. Dean, and M.J. Harper (Eds.), New Developments in Productivity Analysis (pp. 303–363). Chicago: University of Chicago Press.
Foster, L., Haltiwanger, J., Krizan, C.J., 2006. Market selection, reallocation, and restructuring in the U.S. retail trade sector in the 1990s. Review of Economics and Statistics 88(4), 748–758.
Fujii, D., Nakata, T., 2021. COVID-19 and output in Japan. Japanese Economic Review 72(4), 609–650.
Gibbs, M., Mengel, F., Siemroth, C., 2021. Work from home and productivity: Evidence from personnel & analytics data on IT professionals. IZA Discussion Paper 14336.
Golden, T.D., Eddleston, K.A., 2020. Is there a price telecommuters pay? Examining the relationship between telecommuting and objective career success. Journal of Vocational Behavior 116, February, 103348.
Griliches, Z., Regev, H., 1995. Firm productivity in Israeli industry 1979–1988. Journal of Econometrics 65(1), 175–203.
He, H., Neumark, D., Weng, Q., 2021. Do workers value flexible jobs? A field experiment. Journal of Labor Economics 39(3), 709–738.
Hoshi, K., Kasahara, H., Makioka, R., Suzuki, M., Tanaka, S., 2021. Trade-off between Job losses and the spread of COVID-19 in Japan. Japanese Economic Review 72(4), 683–716.
Janys, L., Zimpelmann, C., von Gaudecker, H.-M., Siflinger, B., Holler, R., 2021. Hours and income dynamics during the Covid-19 pandemic: The case of the Netherlands. Labour Economics 73, December, 102055.
Jones, C.J., Philippon, T., Venkateswaran, V., 2021. Optimal mitigation policies in a pandemic: Social distancing and working from home. Review of Financial Studies 34(11), 5188–5223.
Kawaguchi, D., Motegi, H., 2021. Who can work from home? The roles of job tasks and HRM practices. Journal of the Japanese and International Economies 62, 101162.
Kawaguchi, D., Kitao, S., Nose, M., 2022. The impact of COVID-19 on Japanese firms: Mobility and resilience via remote work. International Tax and Public Finance 29(6), 1419–1449.
Kikuchi, S., Kitao, S., Mikoshiba, M., 2021. Who suffers from the COVID-19 shocks? Labor market heterogeneity and welfare consequences in Japan. Journal of the Japanese and International Economies 59, 101117.
Kitagawa, N., Kuroda, S., Okudaira, H., Owan, H., 2021. Working from home: Its effects on productivity and mental health. RIETI Discussion Paper 21-E-024.
Kouki, A., Sauer, R.M., 2022. Remote work, children’s health and the racial gap in female wages. IZA Discussion Paper 15072.
Lewandowski, P., Lipowska, K., Smoter, M., 2022. Working from home during a pandemic: A discrete choice experiment in Poland. IZA Discussion Paper 15251.
Mas, A., Pallais, A., 2017. Valuing alternative work arrangements. American Economic Review 107(12), 3722–3759.
Moens, E., Verhofstadt, E., Ootegem, L.V., Baert, S., 2022. Disentangling the attractiveness of telework to employees: A factorial survey experiment. IZA Discussion Paper 15190.
Morikawa, M., 2022. Work-from-home productivity during the COVID-19 pandemic: Evidence from Japan. Economic Inquiry 60(2), 508–527.
Morikawa, M., 2023. Productivity dynamics of remote work during the COVID-19 pandemic. Industrial Relations 62(3), 317–331.
Oettinger, G.S., 2011. The incidence and wage consequences of home-based work in the United States, 1980-2000. Journal of Human Resources 46(2), 237–260.
Okubo, T., Inoue, A., Sekijima, K., 2021. Teleworker performance in the COVID-19 era in Japan. Asian Economic Papers 20(2), 175–192.
Petroulakis, F., 2023. Task content and job losses in the great lockdown. ILR Review 76(3), 586–613.
Shen, L., 2023. Does working from home work? A natural experiment from lockdowns. European Economic Review 151, 104323.
Teodorovic, T., Sadun, R., Kun, A.L., Shaer, O., 2022. How does working from home during COVID-19 affect what managers do? Evidence from time-use studies. Human–Computer Interaction 37(6), 532–557.
Van der Lippi, T., Lippényi, Z., 2020. Co-workers working from home and individual and team performance. New Technology, Work and Employment 35(1), 60–79.

Deole et al. (2023), using the same survey data (the Understanding Society COVID-19 Study) extended to September 2021, find that the increased frequency of working from home is positively associated with employees’ self-reported productivity.
Bartelsman and Doms (2000) is a representative survey of the literature on productivity dynamics.
Barrero et al. (2022) present evidence that the amenity value gains associated with the recent rise of WFH moderate wage growth pressures in the United States.
The SCMEP in 2019 was sent to 15,000 firms that operate in the manufacturing and service industries, which were randomly selected from the registered list of the 2016 BSJBSA.
The survey questions were designed by the author based on interviews with employees and managers of an organization that adopted WFH practice just after the onset of the COVID-19 pandemic.
Since a relatively large number of firms did not provide the percentage of workers with a university or higher education, the inclusion of this variable reduces the sample size.
The results are almost the same when the sample is restricted to the panel firms.
The frequency of WFH expressed in percentage is calculated as dividing the mean weekly frequency of WFH by five. A small number of firms responded that the weekly frequency of WFH as six or seven days. In these cases, the frequency of WFH is treated as 100%.
The figures for the subsample of panel firms are not reported in the table, but the significant reduction is observed: from 19.7% in 2020 to 3.9% in 2021.
Morikawa (2023), using data obtained from employee survey, reports that WFH productivity relative to office productivity is 77.5 in 2021. The employees’ self-assessed productivity of WFH in 2021 is close to the firms’ evaluation presented in this study.
When we run probit estimation to explain the exit from WFH where productivity of WFH in 2020 is included as an explanatory variable, the coefficient for WFH productivity is negative and highly significant (see Appendix Table A4). The result confirms that after controlling for other firm characteristics, firms with lower WFH productivity selectively discontinued WFH.
According to an estimation of ordered-probit model to explain firms’ plan to continue WFH practice, high WFH intensity, and high WFH productivity firms, tend to have positive intention to continue frequent WFH after the COVID-19 pandemic.
Felstead and Reuschke (2023), based on a survey of workers in the United Kingdom during the COVID-19 pandemic, report that about 90% wanted to continue WFH.
Aksoy et al. (2022) pointed out the mismatch in employers’ and employees’ preferences for WFH frequency.

Tables 1 to 7 are available in the Supplementary Files section.

No competing interests reported.

Download PDF

Journal Publication

published 29 Feb, 2024

Read the published version in Journal of Evolutionary Economics →

Editorial decision: Revision requested
20 Nov, 2023
Reviews received at journal
09 Nov, 2023
Reviewers agreed at journal
26 Sep, 2023
Reviewers invited by journal
23 Sep, 2023
Editor assigned by journal
23 Sep, 2023
Submission checks completed at journal
29 Aug, 2023
First submitted to journal
27 Aug, 2023

You are reading this latest preprint version

Productivity dynamics of work from home: Firm-level evidence

Status:

Journal Publication

Version 1

Abstract

1 Introduction

2 Literature review

3 Design of the firm survey

4 Results

4.1 Utilization and intensity of WFH

4.2 Productivity dynamics of WFH

4.3 Firms’ plan to utilize WFH after the COVID-19 pandemic

5 Conclusion

Declarations

References

Footnotes

Tables

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1