Development
Research
Policy
01.04.2021

Working from home in developing countries

Abstract
We use worker-level data on the task content of jobs to measure the ability to work-from-home (WFH) in developing countries. We show that the ability to WFH is low in developing countries and document significant heterogeneity across and within occupations, and across worker characteristics. Our measure suggests that educated workers, wage employees and women have a higher ability to WFH. Using data from Brazil, Costa Rica and Peru, we show that our measure is predictive of actual WFH both in terms of overall levels and variation with occupation and individual characteristics, as well as employment outcomes. Our measure can thus be used to predict WFH outcomes in developing countries.

The spread of Covid-19 has led to the widespread adoption of social-distancing measures in countries across the world, be it in response to government mandates or on a voluntary basis. Since social distancing frequently involves the closure of workplaces to limit interpersonal contact, the ability to work from home (WFH) is a critical factor for determining its economic consequences.

Early in the pandemic, Dingel and Neiman (2020) measured the ability to WFH by occupation using US task data, and found that 37% of workers can. Their findingscomplement the literature measuring actual WFH prior to the pandemic (Mas and Pallais 2020, Hansvik et al. 2020). Mongey et al. (2020) extend these measures, stressing the fact that low-income workers are substantially less able to WFH and thus more vulnerable to job losses from social distancing. Bick et al. (2020) analyse US data during the pandemic and show that actual WFH largely lines up with predictions. Similarly, Adams-Prassl et al. (2020a) find that during the spring of 2020, roughly 40% of survey respondents reported being able to carry out work remotely during the pandemic months in Germany, the UK, and the US. This is consistent with other studies on the potential to WFH in several European countries (Alipour et al. 2020, Boeri et al. 2020). Furthermore, Adams-Prassl et al. (2020b) emphasise the considerable variation in the ability to WFH within occupations and industries.

The above measures cannot be directly extrapolated to developing countries, as the task content of occupations may vary significantly across contexts (Dicarlo et al. 2016, Lo Bello et al. 2019, Saltiel 2019). Berg et al. (2020) consult labour-market experts to define the occupational ability to WFH in different countries. Garrote Sanchez et al. (2020) argue that the US-based occupational WFH ability rates of Dingel and Neiman (2020) should be adjusted downward in developing countries due to more limited internet access. Hatayama et al. (2020) go a step further by using data from developing countries on tasks to characterise jobs that are amenable to WFH. 

In this column, we discuss the findings of a recent paper that contains two contributions (Gottlieb et al. 2021). First, we provide a measure of the ability to WFH using task data in developing countries and analyse its predictions. Second, we validate that measure with data on actual WFH and employment outcomes.                                      

A task-based measure of the ability to work from home

We develop a measure of the ability to WFH based on tasks performed by workers from the Skills Toward Employability and Productivity (STEP) survey. This covers a representative sample of workers in urban areas in ten low- and middle-income countries, ranging from Kenya to Macedonia. 

Table 1 provides the fraction of workers able to WFH, broken down by occupation, educational attainment, self-employment status, and gender. Overall, 9.3% of urban employment can be done remotely. The ability to WFH varies significantly across broad occupation groups. While close to one-quarter of jobs in managerial, professional occupations, and clerical support occupations could be done from home, fewer than 3% of jobs in elementary occupations, crafts, or occupations involving plant or machine operation can be done remotely.

The ability to WFH also varies across personal and job characteristics. The second and third columns of Table 1 show that educational attainment is a strong predictor of ability to WFH. The share for those who complete high school surpasses dropouts by 13 percentage points. High-school graduates have higher WFH ability than their counterparts across all occupations. Similarly, the ability to WFH for wage employees (12.4%) is far higher than for self-employed workers (3.3%). Finally, women have a slightly higher WFH ability than men. 

Our statistical analysis indicates that while a worker’s occupation is the main driver of the ability to WFH, individual characteristics (education, self-employment status, and gender) matter as well. Once these observables are taken into account, WFH ability is similar across countries in the data sample. This means that our estimates of WFH ability by occupation and observed characteristics are likely to hold across a large spectrum of developing countries spanning the income per capita range of STEP countries ($3,000–14,000 PPP USD). 

The existing literature on the ability to WFH largely builds on worker task data from developed countries. To compare these with the findings from STEP countries, we use US data from O*NET to compute for each three-digit occupation the fraction of workers able to WFH using a definition equivalent to ours. Figure 1 shows the results relative to our estimates from STEP. For virtually all occupations, the O*NET data predict a substantially higher WFH ability. This cautions against extrapolating from O*NET to developing countries, because the task content within occupations differs between rich and poor countries. At the aggregate level, developing countries have a relatively low ability to WFH for two reasons: (1) their employment is concentrated in low-WFH ability occupations; and (2) within occupations, tasks are less amenable to being executed remotely.

Predicted ability to work from home and actual outcomes

We then analyse the share of workers in Brazil and Costa Rica who actually worked from home during the pandemic. The estimated share of workers who actually worked from home in the second quarter of 2020 equals 10.6% in Costa Rica and 13.3% in Brazil, far below the corresponding shares in developed countries. Unsurprisingly, remote working varies enormously by occupation in both countries, remaining below 1% for plant operators and exceeding 40% for professionals. More educated workers, women, and wage employees are more likely to have worked from home during the pandemic, in line with our WFH ability predictions.

We propose various methods that validate our STEP-based measure of WFH ability on the Brazilian and Costa Rican data. Figure 2 visualises one such exercise. The horizontal axis portrays 20 equally sized bins of workers in the Brazilian (Costa Rican) sample whose likelihood of WFH is predicted using their occupation and observed characteristics following the methodology based on STEP. The vertical axis plots the corresponding share of workers who actually worked from home. The regression line is close to the 45-degree line, implying that our measure of the ability to WFH is a good predictor of actual WFH.

We also illustrate the usefulness of our measure of WFH ability for the case of Peru, a country with no information on actual WFH. We find that individuals with a low predicted ability to WFH are more likely to suffer employment losses during the Covid pandemic, especially if they have been employed in non-essential sectors. 

Conclusion                                          

We propose a measure of the ability to WFH for developing countries. Our findings indicate that fewer than 10% of urban jobs in developing countries can be done remotely. Various vulnerable groups are less likely to work remotely, including workers in low-wage occupations, high school dropouts, and the self-employed. Importantly, our results indicate that the low WFH ability in developing countries is driven not only by occupational composition, but also by the nature of tasks within occupations. 

We validate our measure with data on actual WFH from Brazil and Costa Rica, two countries that stand out among developing nations in gathering such information. For a broad range of other developing countries, the WFH ability can be computed using our measure of WFH ability by subgroups coupled with country-specific employment shares. Users can do so using our Work from Home simulator, which can be accessed at https://work-in-data.shinyapps.io/work_in_data/.

References

Adams-Prassl, A, T Boneva, M Golin and C Rauh (2020a), “Inequality in the Impact of the Coronavirus Shock: Evidence from Real Time Surveys”, Journal of Public Economics 189. 

Adams-Prassl, A, T Boneva, M Golin and C Rauh (2020b), “Work Tasks That Can be Done from Home: Evidence on the Variation Within and Across Occupations and Industries”, Cambridge Working Papers in Economics 2040, May.                                                                                               

Alipour, J-V, O Falck and S Schller (2020), “Germany’s Capacities to Work from Home”, IZA Discussion Papers 13152, Institute of Labor Economics (IZA).   

Berg, J, F Bonnet and S Soares (2020), “Working from Home: Estimating the Worldwide Potential”, VoxEU.org, 11 May.                                    

Bick, A, A Blandin and K Mertens (2020), “Work from Home After the COVID-19 Outbreak”, CEPR Discussion Paper 15000.                   

Boeri, T, A Caiumi and M Paccagnella (2020), “Mitigating the Work-Security Trade-Off while Rebooting the Economy”, VoxEU.org, 9 April.                              

Dicarlo, E, S L Bello, S Monroy-Taborda, A M Oviedo, M L Sanchez-Puerta and I Santos (2016), “The Skill Content of Occupations Across Low- and Middle-Income Countries: Evidence from Harmonized Data”, IZA Discussion Papers 10224.

Dingel, J I and B Neiman (2020), “How Many Jobs Can be Done at home?”, VoxEU.org, 7 April.

Gottlieb, C, J Grobovšek, M Poschke and F Saltiel (2021), “Working from Home in Developing Countries”, European Economic Review 133, 103679.

Garrote Sanchez, D, N Gomez Parra, C Ozden, B Rijkers, M Viollaz and H J Winkler (2020), “Who on Earth Can Work from Home?”, World Bank Policy Research Working Paper 9347.

Hatayama, M, M Viollaz and H Winkler (2020), “Jobs’ Amenability to Working from Home: Evidence from Skills Surveys for 53 Countries”, World Bank Policy Research Working Paper 9241.                                       

Hensvik, L, T Le Barbanchon and R Rathelot (2020), “Which Jobs Are Done from Home? Evidence from the American Time Use Survey”, IZA Discussion Papers 13138.                            

Lewandowski, P, A Park, W Hardy and Y Du (2019), “Technology, Skills, and Globalization: Explaining International Differences in Routine and Nonroutine Work Using Survey Data”, IZA Discussion Papers 12339.                          

Lo Bello, S, M L Sanchez Puerta and H Winkler (2019), “From Ghana to America: The Skill Content of Jobs and Economic Development”, IZA Discussion Papers 12259.        

Mas, A and A Pallais (2020), “Alternative Work Arrangements”, Annual Review of Economics 12, 14 August.      

Mongey, S, L Pilossoph, and A Weinberg (2020), “Which Workers Bear the Burden of Social Distancing Policies?”, NBER Working Papers 27085.        

Saltiel, F (2019), “Comparative Evidence on the Returns of Tasks in Developing Countries”, Mimeo.

#wfh
#remotework
#tasks
#covid-19
#occupations