Economics Working Papers

8-5-2020 Working Paper Number 20010

Who Can Work and Study from Home in : Evidence from a 2018-19 Nationwide Household Survey

Syed Hasan University of Management Sciences

Attique Rehman Lahore University of Management Sciences

Wendong Zhang Iowa State University, [email protected]

Original Release Date: May 2, 2020 Revision: August 5, 2020 Follow this and additional works at: https://lib.dr.iastate.edu/econ_workingpapers

Part of the Health Economics Commons, Inequality and Stratification Commons, Labor Economics Commons, Regional Economics Commons, and the Work, Economy and Organizations Commons

Recommended Citation Hasan, Syed; Rehman, Attique; and Zhang, Wendong, "Who Can Work and Study from Home in Pakistan: Evidence from a 2018-19 Nationwide Household Survey" (2020). Economics Working Papers: Department of Economics, Iowa State University. 20010. https://lib.dr.iastate.edu/econ_workingpapers/105

Iowa State University does not discriminate on the basis of race, color, age, ethnicity, religion, national origin, pregnancy, sexual orientation, gender identity, genetic information, sex, marital status, disability, or status as a U.S. veteran. Inquiries regarding non-discrimination policies may be directed to Office ofqual E Opportunity, 3350 Beardshear Hall, 515 Morrill Road, Ames, Iowa 50011, Tel. 515 294-7612, Hotline: 515-294-1222, [email protected].

This Working Paper is brought to you for free and open access by the Iowa State University Digital Repository. For more information, please visit lib.dr.iastate.edu. Who Can Work and Study from Home in Pakistan: Evidence from a 2018-19 Nationwide Household Survey

Abstract In this article, we examine the feasibility of working and studying from home in Pakistan. We take advantage of the recently released 2018–19 Pakistan Social and Living Standards Measurement (PSLM) Survey. PSLM is a nationally representative household survey with extensive information on employment outcomes, children’s educational attainment, as well as and computer access and prior usage information. Modifying Dingel and Neiman [2020]’s approach, we define the easibilityf of jobs that can be done from home based on the percentage of tasks that can be switched online and accounting for internet accessibility. We also investigate the possibilities for students to study from home via TV or internet. We find that only 10% of jobs in Pakistan can be done from home; however, megacities have much higher rates and rural areas have lower rates. In addition, many of Pakistan’s male workers are in low-skill, low-paying service industries and cannot work from home, while occupations with a higher female employment share have a relatively higher work-from-home share despite lower percentage of prior internet use. Our results also highlight the homeschooling challenges Pakistan’s students face given the low rates of television and . The government’s outreach effort through the new Teleschool TV channel could help alleviate pre-existing gender inequalities in access to education.

Keywords COVID-19, Occupations, Tasks, Pakistan, Work from Home, Distance Learning

Disciplines Health Economics | Inequality and Stratification | Labor conomicsE | Regional Economics | Work, Economy and Organizations

This article is available at Iowa State University Digital Repository: https://lib.dr.iastate.edu/econ_workingpapers/105 Who Can Work and Study from Home in Pakistan: Evidence from a 2018-19

Nationwide Household Survey

Syed M. Hasan

Associate Professor, Department of Economics

Lahore University of Management Sciences

D.H.A, Lahore Cantt. 54792, Lahore, Pakistan

Email: [email protected]

Phone: +92-331-5036704

Attique Rehman

Adjunct Faculty, Department of Economics

Lahore University of Management Sciences

Email: [email protected]

Wendong Zhang

Assistant Professor, Department of Economics and Center for Agricultural and Rural

Development, Iowa State University

478C Heady Hall, 518 Farmhouse Lane, Ames, Iowa 50011

Email: [email protected]

Phone: 515-294-2536 / : 515-294-0221

Acknowledgement

This work was supported by the Department of Economics, LUMS, Lahore and the USDA

National Institute of Food and Agriculture Hatch Project [101,030]. The authors would like to thank Nathan Cook and Kate Epstein for their valuable comments and suggestions on an earlier draft. Who Can Work and Study from Home in Pakistan: Evidence from a 2018-19 Nationwide Household Survey

August 5, 2020

Abstract

In this article, we examine the feasibility of working and studying from home in Pakistan. We take advantage of the recently released 2018–19 Pakistan Social and Living Standards Measurement (PSLM) Survey. PSLM is a nationally representative household survey with extensive information on employment outcomes, children’s educational attainment, as well as internet and computer access and prior usage information. Modifying Dingel and Neiman [2020]’s approach, we define the feasibility of jobs that can be done from home based on the percentage of tasks that can be switched online and accounting for internet accessibility. We also investigate the possibilities for students to study from home via TV or internet. We find that only 10% of jobs in Pakistan can be done from home; however, megacities have much higher rates and rural areas have lower rates. In addition, many of Pakistan’s male workers are in low-skill, low-paying service industries and cannot work from home, while occupations with a higher female employment share have a relatively higher work-from-home share de- spite lower percentage of prior internet use. Our results also highlight the homeschooling challenges Pakistan’s students face given the low rates of television and internet access. The government’s outreach effort through the new Teleschool TV channel could help alleviate pre-existing gender inequalities in access to education.

Keywords: COVID-19, Occupations, Tasks, Pakistan, Work from Home, Distance Learning

JEL Codes: D24, J22, R12, R23, I24, O53

1 Who Can Work and Study from Home in Pakistan: Evidence from a 2018-19 Nationwide Household Survey

Word Count: 3,460

1 Introduction

COVID-19 poses heterogeneous global challenges. In many developing countries, disease containment and suppression through non-pharmaceutical interventions, such as lockdowns, have spurred massive layoffs and sustenance issues. Enforcing lockdowns leads to a paradox- ical situation—it is essential to prevent the infection’s spread and “flatten the curve,” but those at the lower end of the income distribution will struggle to survive due to economic impacts, as highlighted by Mobarak and Barnett-Howell [2020]. The absence of a social security mechanism to cover unemployment claims and the prevalence of informal workers in the economy makes COVID-19 management a daunting task for Pakistani government. Uncertainty has been high for the overall economy, and particularly acute for the education sector, as COVID-19’s highly contagious nature will limit traditional learning institutions’ roles in the near future, which demands specifically designed and targeted policy interven- tions to mitigate economic and learning losses. As lockdowns seem inevitable and pervasive, we need to analyze the feasibility of working and studying from home thoroughly. Previous research focusing on the developed economies, such as [Dingel and Neiman, 2020, Gottlieb et al., 2020], indicates a high capacity for significant transition to home-based online options for work and education. However, we need to determine how a similar transition will work for Pakistan, a developing country with the world’s sixth-largest population and significant regional and gender-based disparities.

The nature of jobs, the possession of digital devices, internet access, prior technology ex- posure, and requisite training provision are important factors we must analyze to determine if home-based online work and education options are attainable for a developing country like Pakistan. Pakistan is likely to suffer heavily due to lockdown [Nafees and Khan, 2020], as high-skilled services in finance and information technology are only around 10% of its total workforce [Sharma, 2019], and most sectors cannot switch easily to online service. E- commerce platforms use is extremely limited in Pakistan, and even such jobs often require

1 face-to-face interaction. Furthermore, distance learning is also constrained by the limited access to internet and computers, as well as teacher’s lack of prior exposure to them.

We use Pakistan’s 2018–19 Pakistan Social and Living Standards Measurement (PSLM) Survey data [Pakistan Bureau of Statistics, Government of Pakistan, 2020] to assess the extent to which workers and students could work or study from home. PSLM is a nation- ally representative household survey with extensive information on employment outcomes, children’s educational attainment, and internet, computer and TV access information. An- alyzing occupational codes at 4-digit level and internet availability across rural and urban regions, we adapt Dingel and Neiman [2020]’s approach to determine what percentage of Pakistan’s labor force can switch to online work, while explicitly accounting for limited access to mobile or broadband internet. We also examine differences in rural and urban workers’ and male and female workers’ abilities to switch to online work. Finally, we use the PSLM survey to assess the degree to which students could study from home using computers, TVs, or mobile phones during school closures. We specifically examine the constraints of TV availability for students to compensate for academic loss. Pakistan recently introduced Teleschool, a dedicated TV channel, for educational outreach [Zahra-Malik, 2020].

Our article contributes to a growing literature that examines the impacts of COVID-19 on work-from-home feasibility. Most existing literature focuses on developed countries, such as the [Dingel and Neiman, 2020, Baker, 2020, Mongey and Weinberg, 2020, Bick and Blandin, 2020], Italy [Barbieri et al., 2020], the United Kingdom [United Kingdom Office for National Statistics, 2020], and Germany [Fadinger and Schymik, 2020]. These studies find that even in developed countries, less-educated workers, women, and those in occupations with more need for physical proximity are affected more [Mongey and Wein- berg, 2020]. In particular, Mongey and Weinberg [2020] find that workers who cannot work from home are more likely to be non-white, lower-income, renters, less-educated, and lack employer-provided health insurance. Saltiel [2020] and Dingel and Neiman [2020] are the two exceptions to our knowledge that provide information on the share of jobs that can be done from home in developing countries. They use 2-digit ISCO (international standard classification of occupations) employment numbers, and find that 13% of Pakistan’s workers can do their job from home [Dingel and Neiman, 2020].

We find that 12% of Pakistan’s urban jobs and 9% of rural jobs can be done at home, which is slightly lower than and roughly aligns with the prior study’s findings. Additionally, we provide a heterogeneity analysis that indicates that megacities are considerably different

2 from urban and rural regions in terms of work-from-home options. Interestingly, we find that the top female occupations—the health and education sectors—have more work-from- home options than the top male occupations. Our examination reveals female students face substantial pre-existing inequalities in primary and secondary education opportunities. Less than half of eligible females enroll in primary schools, and this ratio drops to less than 30% for secondary and 20% for tertiary education. The proportion of households with access to TV and internet is very low, especially in rural areas, which suggests disproportionally larger pandemic impacts on students’ abilities to study from home in socially disadvantaged households.

We enrich the literature by validating and refining Dingel and Neiman [2020]’s findings by explicitly accounting for internet access. Additionally, our results reveal important additional challenges to workers effectively working from home in rural areas and in urban districts outside the ten megacities. Finally, we also provide the first systematic estimate on the feasibility of Pakistani students studying from home using a computer or TV.

2 Background

As of August 10, 2020, Pakistan confirmed about 285,000 COVID-19 cases and 6,100 deaths in Pakistan—fourteenth-highest in the world and fourth-highest in Asia, after India, Iran, and Saudi Arabia [Johns Hopkins University, Coronavirus Resource Center, 2020]. The COVID-19 pandemic has resulted in severe damages to Pakistan’s economy—the Interna- tional Monetary Fund (IMF) forecasted Pakistan’s GDP will drop by 0.4% in 2020, and the real GDP growth is expected to slow by 3% [International Monetary Fund (IMF), 2020]. Pakistan Ministry of Planning estimates that 12-18 million people will become jobless due to the pandemic. IMF also projects a sharp spike in Pakistan’s poverty rate, up to 40% [CGTN, 2020]. In addition, a new assessment by the United Nations Development Program finds that those most at-risk include women, poor households who rely on farm labor or daily wages, and people with disabilities. In addition, nearly 42 million children are out of school and over 300,000 schools are closed [United Nations Development Programme, 2020].

On April 1, Prime Minister Khan launched the Ehsaas Emergency Cash financial relief program, which includes a total of 144 billion rupees (US $0.86 billion) to approximately 12 million families. Since late March, Pakistan’s federal and provincial governments have im- plemented various measures to mitigate the spread of the virus, including travel restrictions, lockdowns, selective quarantines, and closing borders, schools, and universities, and imple-

3 menting social distancing measures. On May 9, Pakistan lifted its nationwide lockdown to reduce the devastating economic and livelihood impacts, and gradually eased mitigation re- quirements by allowing “low-risk industries” and “small retail shops” to reopen. A four-fold rise in the confirmed infections following the reopening [ur Rehman et al., 2020] prompted Pakistan’s government to enforce a selective “smart lockdown” strategy that targets 500 coronavirus hotspots across the country [Hashim, 2020].

The COVID-19 pandemic and the following governmental support has disproportionate impacts on the workforce and school-aged children. Nearly half of all Pakistani households rely on crop and livestock as their main or secondary source of livelihood, and they face signif- icant disruptions in the transportation, labor for harvest and transport, and access to inputs for next planting seasons [United Nations Development Programme, 2020]. In addition, the textile and garment factories ordered to shut down employ Pakistan’s largest industrial work- force. According to the Pakistan Workers’ Federation, as of March 28, at least one million textile and garment industry workers had been dismissed in Punjab province alone [Human Rights Watch (HRW), 2020]. Furthermore, the pandemic has a disproportionate effect on women workers, especially home-based or domestic workers, who are often invisible within the system and unaccounted for in wage provision and financial support programs. With over 300,000 schools closures impacting over 12 million students, Pakistan’s government pushed out a new distance learning program with a dedicated TV channel called Teleschool. Ac- cording to the Pakistan Authority, about 40 million Pakistani children have access to a television while only one million have regular access to broadband internet [Zahra-Malik, 2020]. According to the 2018–19 PSLM survey used in our study, only 27% of households own a computer and only 51% have access to internet in urban regions.

3 Data and Methodology

We link household occupation and internet/computer use information from the 2018-19 PSLM survey with the Pakistan Standard Classification of Occupations (PSCO) [Pakistan Bureau of Statistics, Government of Pakistan, 2015] to determine the number of tasks that can be performed online. Pakistan conducts PSLM surveys at the provincial level in alter- nating years, and data collection is based on stratified sampling of both urban and rural areas. The survey gives detailed information on occupation, income, age-wise school enroll- ment, and computer access and internet use.

We modify Dingel and Neiman [2020]’s approach to determine the work-from-home fea-

4 sibility across occupations by incorporating an internet accessibility factor in our analysis. Dingel and Neiman [2020] assume no constraints in the access to computer and internet in assessing the percentage of job tasks switchable to online. However, this could overestimate the work-from-home feasibility for Pakistan computer literacy and computer and internet availability are limited, especially in rural Pakistan. Thus, we incorporate household-level ac- cess information to mobile or broadband internet when determining the feasibility of working from home. Our measure of work-from-home share is lower than Dingel and Neiman [2020] because we assume workers could only do so if tasks can be switched online and they at least have access to mobile internet.

We list all 410 occupations at the PSCO 4-digit level to determine if they can be per- formed online. First, we make a binary decision of work-from-home feasibility based on the necessity of physical workplace presence. For occupations that can be done at online, we incorporate the share of access to mobile or broadband internet at home for all households within each occupation, resulting in the percent of tasks that can be performed remotely. For each occupation, we also validate this measure by adding home computer access and prior use of internet and computers. When making our decision, we assume that some farm- related occupations, especially marketing-related tasks, can continue but other operations cannot. We acknowledge that this is not measuring the percent of farm workers who can continue to work in the fields during lockdown.1

Next, we rank the 4-digit occupations on the basis of employment numbers and focus on the top 25 occupations. The top 25 occupations account for around 70% of employment in Pakistan, and we consider them as fairly representing the entire labor force. As a robustness check, we match occupation statistics with the most recent labor force survey to confirm oc- cupations’ distribution validity. To determine the feasibility of working online for the top 25 occupations, we add information on household-level internet availability and the proportion of tasks that can be switched online by reviewing the complete list of routine tasks stated in the occupation description.

1Determining the rural workers’ ability to work during the lockdowns is a tricky question, especially for the farm workers. Subsistence farmers who are working in fields adjacent to their residences can continue to work in the fields with few hurdles. According to Gottlieb et al. [2020], the farmers’ abilities to work during the lockdowns are critical in explaining the differences of working from home shares between developed and developing countries. Although it is an important debate, it is beyond the scope of this study since our focus is limited to working from home as opposed to the impacts or effectiveness of lockdowns. We acknowledge that rural Pakistan has remained less affected overall due to fewer confirmed cases and relatively fewer restrictions [Wikipedia, 2020].

5 To analyze the within-sector heterogeneity and variations across male/female and ur- ban/rural jobs we aggregate the 4 -digit occupation classification to 2-digit sectors, based on the employment-weighted percent of tasks that can be performed online. In particular, we examine the work-from-home feasibility for the top 15 sectors with the highest female or rural employment shares. We make a separate megacities category for the top 10 cities with at least one million residents. We also aggregate and plot the weighted average work- from-home shares for every district to examine the spatial distribution of the feasibility of working online. Prior computer and internet use information provides another lower limit estimate that indicates the option of immediate switch to online work without any training.

Finally, to examine students’ abilities to study from home, we separately examine the percentage of students with access to internet, TV, and computers, for rural and urban areas. We divide school enrollment into primary (grades 1–5), secondary (grades 6–12), and tertiary (all undergraduate and graduate studies), and assess the pre-existing gender inequality in school access, especially in rural areas. For the vast majority of students, school closures are indefinite, only a very small proportion of students are enrolled in schools that have any online teaching capacity.

4 Results

Table 1 presents the percent of jobs that can be done online and the percent of workers with home internet access for the top 25 occupations, which collectively account for more than 70% of Pakistan’s workforce. Of these 25 occupations, only workers in eight occupations could possibly work online. We also analyze the proportion of tasks for each of the 25 oc- cupations to determine the extent of possible switch to online work. Table 1 shows a huge variation, with around 11%–90% task-switching possibility for different occupations. In this list, three occupations (primary school teachers, secondary school teachers, and office clerks) have a high percentage of switchable tasks (80%, assuming internet access). However, in 2018-19, the proportion of workers who have access to reliable internet was 60%–70%, which includes both mobile and broadband internet. Furthermore, it is difficult for many agri- cultural producers, which accounts for almost 40% of workers, to switch to working online. Mixed-crop growers or dairy producers could do some management and marketing tasks at home; however, over 70% of their production tasks are impossible to perform online, and less than one-quarter have internet access.

By aggregating the feasibility of working online for all 410 occupations, we find that over-

6 all, weighted on the percent of tasks performable home, 10% of Pakistan’s jobs can be done from home (12% for urban and 9% for rural). Our results are roughly comparable to Dingel and Neiman [2020]’s finding of 13%. For robustness checks against these estimates, we add percentage statistics for prior internet usage, home computer availability, and percentage of prior computer usage, respectively, against each occupation in the last three columns of Table 1. All three percentages are lower than the internet availability percentage when allowing for mobile internet. As a result, using computer usage history to constrain the work-from-home possibility yields even more conservative estimates.

Figure 1 depicts the geographical distribution of the employment-weighted measure of the share of Pakistan’s urban and rural jobs that can be done online, which reveals a significant variation across cities—the weighted average share ranges from 8% to 67%. Northern urban districts in Pakistan have a higher percent of jobs that can be performed online than southern urban districts. However, only 51% of all urban Pakistani households have internet access. Given that 73% of rural Pakistani households have no internet access, the few who may not have the means. The three largest cities in Pakistan, , Lahore, and , enjoy an average of 25%, 16%, and 18% of jobs, respectively, that could be performed online. In contrast, only 15% or fewer of jobs in rural districts in the Sindh province could be done online. This finding highlights that rural regions will suffer more gravely from lockdowns and work-from-home orders than their urban counterparts. The scale of pre-existing inequal- ities in public service provisions leaves rural residents fewer options for setting up a home office and effectively working from home [Human Rights Watch (HRW), 2018]. Our result is consistent with previous findings that lower income groups in the United States have more difficulty being able to work from home [Mongey and Weinberg, 2020, Baker, 2020].

Figure 2 highlights the district-level distribution of work-from-home feasibility as di- vided among rural and urban households. The right-skewed distribution in urban areas, as compared to the more normal distribution for rural households, indicates a regional divide. However, the differences could only be observed for the ten megacities with some of whose work-from-home feasibility extending beyond 20%. Figure 3 shows similar trends at the provincial level — more urbanized provinces like Sindh and Punjab — have, on average, a high percent of jobs that can be done from home compared to Balochistan and Khyber Pakhtunkhwa.

We aggregate the 410 4-digit occupations to 44 2-digit sectors to provide a picture of all occupations in Pakistan. To do so, we rely on the dichotomous categorization for each 4-digit

7 occupation with the top 25 occupations by employment (see Table 1). Table 2 presents the employment-weighting by percent of tasks that could be performed online and the percent of home internet access. Table 2 echoes the findings in Table 1—teaching professionals, busi- ness professionals, and some market-oriented agricultural workers could work online, but are severely limited by sub-optimal internet access. In addition, Table 2 also shows that many low-skill workers — food industry, drivers, and hospitality workers — cannot conduct their jobs effectively online.

Table 3 separates rural and urban workers, and presents the top and bottom ten occu- pations ranked by the share of employment. The top occupations are dominated by rural workers and the percentage of tasks switchable to online is low (0%–18%). In contrast, the urban regions (except for assemblers), top employment share belongs to occupations that have very high percentage of switchable tasks. Table 3 shows that many workers unable to switch tasks online provide low-paid services, including food processing, driving, and mining and construction, which reveals urgent need for the government to divert resources, med- ical protective gear, and financial support to workers in these sectors [Saltiel, 2020]. The Pakistani governments recently have announced economic packages for unemployed workers, but the monthly payments (less than US$30) are even less than the minimum wage in any Pakistani province and thus likely inadequate [Human Rights Watch (HRW), 2020].

Table 4 shows the top employment options for males and females and the ability to work online and reveals that women have a higher percentage of switchable tasks. For the top five occupations for females, 9%–76% of tasks are switchable; however, for males in the top five occupations, 0% are switchable. Females are not better off than males in a lockdown — pre-existing inequality in prior internet use and accessibility inhibits their potential. The extremely low work-from-home feasibility for male-dominated industries are due to job tasks that require a factory or outdoor setting.

Table 5 shows Pakistani students’ abilities to study from home. Our statistics reveal that females have substantial pre-existing inequalities in access to school. On average, about 80% of school-aged females in urban Pakistan go to primary or secondary schools, and that percentage drops to less than 40% in Balochistan’s rural districts. The inequalities are more evident when examining tertiary school gender ratios — less than 10% of tertiary students in Sindh province are women. Table 5 shows internet, TV, and computer availability at the household level — rural areas face larger constraints than megacities. The high proportion of households with a TV in three out of four provinces indicates that there is considerable

8 margin to use this as a communication platform.

Table 5 shows that in urban Pakistan, only about one-half of households with primary- school-aged children have access to computers and the internet; however, Teleschool could help alleviate the gender gap in education access, especially for girls in rural Pakistan. Also note the geographical differences — one-third or less of primary school students in the Khyber Pakhtunkhwa and Balochistan provinces have a TV in their home, while 68% in urban areas of Punjab do.

5 Conclusions and Policy Implications

Segregating jobs that can be largely shifted online from those that cannot is useful for policy- makers designing interventions. Mapping geographical and sectoral vulnerabilities is useful to identify people and sectors that need relief payments and urgent support for online transi- tion. Our results reveal that limited access to internet, computers and lack of proficiency in using digital devices are significant barriers for workers and students in Pakistan — proper training and upgrading of worker skills is important to address the existing and ongoing challenges for workers posed by COVID-19.

Our analysis shows that internet access is a major constraint to transitioning education online. As rural students generally have low internet and TV access rates, their loss of learn- ing is larger. Our results also show that the impacts of COVID-19 are likely compounded by pre-existing inequalities particularly those in female students’ access to education, especially at secondary and tertiary levels [Van Lancker and Parolin, 2020]. The use of use and TV as a medium of education could help reduce these inequalities even in a post-COVID-19 era.

9 References

M. G. Baker. Who Cannot Work from Home? Characterizing Occupations Facing Increased Risk during the COVID-19 Pandemic using 2018 BLS Data. 2020. doi: https://doi.org/ 10.1101/2020.03.21.20031336.

T. Barbieri, G. Basso, and S. Scicchitano. Italian Workers at Risk during the COVID-19 Epidemic. 2020. URL https://dx.doi.org/10.2139/ssrn.3572065.

A. Bick and A. Blandin. Real Time Labor Market Estimates During the 2020 Coronavirus Outbreak. Working Paper, 2020.

CGTN. COVID-19 conundrum in Pakistan: To control virus or ig- nore poverty?, 2020. URL https://news.cgtn.com/news/2020-06-22/ COVID-19-conundrum-in-Pakistan-To-control-virus-or-ignore-poverty--Rx4gnyerRK/ index.html.

J. I. Dingel and B. Neiman. How Many Jobs Can be Done at Home? NBER Working Paper, (No. 26948), 2020.

H. Fadinger and J. Schymik. The Costs and Benefits of Home Office during the Covid-19 Pandemic: Evidence from Infections and an Input-Output Model for Germany. COVID Economics: Vetted and Real-Time Papers, (9):107–134, 2020.

C. Gottlieb, J. Grobovsek, and M. Poschke. Working from Home across Countries. Centre interuniversitaire de recherche en ´economie quantitative (CIREQ), (07-2020), 2020.

A. Hashim. ’smart lockdown’ in pakistan to target 500 coronavirus hotspots. Al Jazeera, 2020. URL https://www.nytimes.com/2020/06/15/world/asia/ pakistan-coronavirus-hospitals.html.

Human Rights Watch (HRW). “Shall I Feed My Daughter, or Educate Her?”: Barriers to Girls’ . Technical report, 2018. URL https: //www.hrw.org/report/2018/11/12/shall-i-feed-my-daughter-or-educate-her/ barriers-girls-education-pakistan.

Human Rights Watch (HRW). Pakistan: Workers Face Health, Economic Risks. Technical report, 2020. URL https://www.hrw.org/news/2020/04/01/ pakistan-workers-face-health-economic-risks.

International Monetary Fund (IMF). Five Charts That Illus- trate COVID-19’s Impact on the Middle East and Central Asia. 2020. URL https://www.imf.org/en/News/Articles/2020/07/14/ na071420-five-charts-that-illustrate-covid19s-impact-on-the-middle-east-and-central-asia.

Johns Hopkins University, Coronavirus Resource Center. COVID-19 Dashboard - Global Map, 2020. URL https://coronavirus.jhu.edu/map.html.

10 A. M. Mobarak and Z. Barnett-Howell. Poor Countries Need to Think Twice About Social Distancing. Foreign Policy, 2020.

S. Mongey and A. Weinberg. Characteristics of Workers in Low Work-From-Home and High Personal-Proximity Occupations. Becker Friedman Institute White Paper, 2020.

M. Nafees and F. Khan. Pakistan’s Response to COVID-19 Pandemic and Efficacy of Quar- antine and Partial Lockdown: A Review. Electronic Journal of General Medicine, (17(2)): emXXX, 2020. doi: DOI:10.29333/ejgm/7951.

Pakistan Bureau of Statistics, Government of Pakistan. Pakistan Standard Classification of Occupations. Technical report, Islamabad, 2015.

Pakistan Bureau of Statistics, Government of Pakistan. Pakistan Social and Living Standards Measurements Survey, 2018-19. Technical report, Islamabad, 2020. URL http://www. pbs.gov.pk/content/pakistan-social-and-living-standards-measurement.

F. Saltiel. Who Can Work From Home in Developing Countries? COVID Economics: Vetted and Real-Time Papers, (6):104–119, 2020.

S. Sharma. Pakistan at 100 : Structural Transformation. Technical report, the World Bank, Washington, D.C., 2019. URL http://hdl.handle.net/10986/31410.

United Kingdom Office for National Statistics. Coronavirus and Homeworking in the UK Labour Market: 2019. Working Paper, 2020.

United Nations Development Programme. COVID-19 - Pakistan Socio-Economic Impact Assessment Response Plan. Technical report, 2020. URL https: //www.undp.org/content/undp/en/home/news-centre/news/2020/Pakistan_ pandemic_could_push_millions_into_poverty.html.

Z. ur Rehman, S. Masood, and M. Abi-Habib. Pakistan’s Lockdown Ended a Month Ago. Now Hospital Signs Read ‘Full.’. the New York Times, 2020. URL https://www.nytimes. com/2020/06/15/world/asia/pakistan-coronavirus-hospitals.html.

W. Van Lancker and Z. Parolin. COVID-19, School Closures, and Child Poverty: A Social Crisis in the Making. The Lancet Public Health, (5(5)):E243–244, 2020. doi: https: //doi.org/10.1016/S2468-2667(20)30084-0.

Wikipedia. COVID-19 pandemic in Pakistan, 2020. URL https://en.wikipedia.org/ wiki/COVID-19_pandemic_in_Pakistan.

M. Zahra-Malik. The Coronavirus Effect on Pakistan’s Digital Divide. BBC - Worklife, 2020. URL https://www.bbc.com/worklife/article/ 20200713-the-coronavirus-effect-on--digital-divide.

11 Table 1: Top 25 4-digit occupations in Pakistan by employment and workers’ abilities to work from home, 2018–19

% tasks % has % used Can % has % used PSCO Occupation description Employment switchable com- com- WFH internet internet to online puter puter 6130 Mixed crop and animal producers 7363321 Y 20% 23% 8% 4% 2% 9313 Bricklayers’ assistants 4851436 N 0% 15% 7% 2% 1% 9211 Vegetable pickers 4534370 N 0% 8% 2% 1% 0% 5221 Grocers 4125614 N 0% 49% 29% 11% 7% 6121 Dairy farmers 3582761 Y 15% 17% 5% 3% 1% 6114 Skilled farmers 2170272 Y 18% 16% 5% 2% 1% 8322 Taxi or cab drivers 1953390 N 0% 38% 21% 6% 3% 6330 Subsistence mixed crop and livestock farmers 1633260 Y 11% 22% 10% 2% 2% 7531 Tailors, dressmakers, furriers and hatters 1609803 N 0% 36% 20% 9% 5% 5223 Shop sales assistants 1598114 N 0% 47% 32% 10% 8%

12 7533 Sewing, embroidery and related workers 1482006 N 0% 21% 7% 4% 2% 6320 Subsistence livestock farmers 1348021 Y 11% 14% 5% 2% 0% Manufacturing labourers not elsewhere clas- 9329 1214677 N 0% 21% 11% 3% 2% sified 2341 Primary school teachers 1166808 Y 80% 66% 50% 17% 20% 6310 Subsistence farm laborers 1032974 N 0% 25% 12% 3% 1% 6111 Field crop and vegetable growers 813794 Y 18% 22% 9% 3% 2% 2330 Secondary education teachers 757999 Y 90% 77% 61% 22% 33% 7231 Motor vehicle mechanics and repairers 693215 N 0% 36% 18% 6% 4% 7111 House builders 647078 N 0% 32% 14% 6% 3% 7112 Bricklayers and related workers 616642 N 0% 9% 5% 2% 1% 5414 Security guards 582919 N 0% 37% 18% 7% 3% 9213 Mixed crop and livestock farm labourers 532071 N 0% 12% 3% 1% 1% 9111 Domestic cleaners and helpers 529834 N 0% 23% 5% 6% 1% 8321 Motorcycle drivers 519593 N 0% 27% 13% 6% 3% 9333 Freight handlers 507025 N 0% 23% 9% 3% 2%

Note: Access to and usage of the internet includes both mobile and broadband internet. Table 2: Share of jobs that can be done at home and share of workers with internet in Pakistan by industry, 2018–19

% tasks % used % has OCC % used Occupation description Employment switchable com- com- Code internet to online puter puter 61 Market-oriented skilled agricultural workers 14337016 18% 7% 2% 3% 52 Sales workers 7509117 15% 29% 7% 10% Labourers in mining, construction, manufactur- 93 7175610 0% 8% 1% 2% ing and transport 92 Agricultural, forestry and fishery labourers 5769997 0% 2% 0% 1% Subsistence farmers, fishers, hunters and gath- 63 4020797 4% 9% 1% 2% erers Food processing, wood working, garment and 75 3991577 0% 15% 4% 7% other craft and related trades workers 83 Drivers and mobile plant operators 3362436 0% 17% 3% 6%

13 23 Teaching professionals 2295315 82% 56% 28% 20% Building and related trades workers, excluding 71 2080614 0% 15% 3% 6% electricians 51 Personal Services 1642331 0% 15% 3% 5% 72 Metal, machinery and related trades workers 1458438 0% 20% 5% 8% 91 Cleaners and helpers 1022560 0% 12% 3% 7% 54 Protective services workers 996775 0% 27% 5% 8% 81 Stationary plant and machine operators 973819 0% 20% 7% 8% 96 Refuse workers and other elementary workers 818798 0% 16% 4% 7% Business and administration associate profes- 33 765439 49% 54% 29% 21% sionals 74 Electrical and electronic trades workers 738134 0% 39% 13% 12% 73 Handicraft and printing workers 617650 0% 22% 8% 10% 22 Health professionals 563735 5% 59% 31% 21% 41 General and keyboard clerks 531337 84% 56% 35% 19% 12 Administrative and commercial managers 366778 88% 83% 60% 30% 26 Legal, social and cultural professionals 357386 6% 49% 24% 17% Table 2, continued

% tasks % used % has OCC % used Occupation description Employment switchable com- com- Code internet to online puter puter 14 Hospitality, retail and other services managers 348200 6% 38% 10% 7% 32 Health associate professionals 336780 0% 48% 20% 15% Legal, social, cultural and related associate pro- 34 325905 2% 31% 12% 12% fessionals 13 Production and specialised services managers 312534 23% 67% 39% 22% 43 Numerical and material recording clerks 298524 0% 50% 27% 20% 31 Science and engineering associate professionals 262347 1% 58% 27% 20% 24 Business and administration professionals 248268 77% 78% 57% 26% 42 Customer services clerks 224372 0% 68% 44% 24% 14 11 Chief executives, senior officials and legislators 164534 90% 83% 58% 25% 44 Other clerical support workers 158841 0% 43% 26% 20% 21 Science and engineering professionals 146508 19% 73% 53% 27% 3 Armed forces occupations, other ranks 141214 0% 39% 8% 7% Market-oriented skilled forestry, fishery and 62 117000 0% 8% 0% 4% hunting workers Software and applications developers and ana- 25 109434 83% 90% 77% 27% lysts 95 Street and related sales and service workers 90461 0% 9% 0% 2% 94 Food preparation assistants 82114 0% 20% 6% 0% 53 Personal care workers 80082 3% 13% 5% 1% 35 Information and communications technicians 66722 36% 81% 57% 24% 82 Assemblers 40488 0% 38% 11% 19% 2 Non-commissioned armed forces officers 21183 0% 59% 12% 24% 1 Commissioned armed forces officers 3243 0% 67% 33% 0% Table 3: Share of jobs that can be done at home for industries dominated by rural and urban workers, respectively

Rural em- % tasks # Rural % used in- % used % has com- OCC Code Occupation description ployment switchable workers ternet computer puter share to online Subsistence farmers, fishers, hunters 63 95% 3826k 5% 9% 1% 2% and gatherers Market-oriented skilled agricultural 61 93% 13390k 18% 7% 1% 3% workers Agricultural, forestry and fishery 92 93% 5373k 0% 2% 1% 1% labourers Labourers in mining, construction, 93 72% 5187k 0% 8% 1% 2% manufacturing and transport 3 Armed forces occupations, other ranks 64% 91k 0% 39% 8% 7% 15 Market-oriented skilled forestry, fishery 62 63% 74k 0% 8% 0% 3% and hunting workers Building and related trades workers, 71 63% 1305k 0% 14% 3% 5% excluding electricians 83 Drivers and mobile plant operators 58% 1954k 0% 16% 3% 6% 54 Protective services workers 54% 542k 0% 27% 5% 8% Non-commissioned armed forces offi- 2 53% 11k 0% 59% 12% 24% cers 51 Personal Services 51% 846k 0% 14% 2% 5% 32 Health associate professionals 49% 166k 0% 47% 20% 16% Legal, social, cultural and related asso- 34 49% 162k 1% 20% 7% 9% ciate professionals 23 Teaching professionals 48% 1091k 82% 55% 25% 19% Stationary plant and machine opera- 81 47% 461k 0% 19% 6% 8% tors Table 3, continued

Urban em- % tasks # Urban % used in- % used % has com- OCC Code Occupation description ployment switchable workers ternet computer puter share to online 82 Assemblers 85% 34k 0% 42% 12% 20% Administrative and commercial man- 12 84% 309k 89% 83% 60% 31% agers Software and applications developers 25 81% 89k 87% 91% 78% 29% and analysts Business and administration profes- 24 80% 198k 78% 78% 58% 26% sionals Chief executives, senior officials and 11 79% 129k 91% 84% 59% 26% legislators 21 Science and engineering professionals 76% 111k 22% 74% 54% 28% 16 94 Food preparation assistants 75% 62k 0% 19% 6% 0% 41 General and keyboard clerks 71% 375k 84% 57% 36% 19% Information and communications tech- 35 69% 46k 40% 78% 56% 23% nicians 42 Customer services clerks 69% 155k 0% 69% 47% 26% Business and administration associate 33 68% 524k 53% 55% 30% 22% professionals Numerical and material recording 43 68% 202k 0% 51% 27% 20% clerks Production and specialised services 13 67% 210k 20% 68% 40% 21% managers Science and engineering associate pro- 31 63% 167k 1% 61% 29% 21% fessionals Street and related sales and service 95 62% 57k 0% 8% 0% 1% workers Table 4: Share of jobs that can be done at home for industries dominated by female and male workers, respectively

Female % tasks # Female % used in- % used % has com- OCC Code Occupation description employ- switchable workers ternet computer puter ment share to online 53 Personal care workers 77% 61k 26% 11% 5% 0% 23 Teaching professionals 55% 1256k 76% 56% 28% 22% Food processing, wood working, gar- 75 ment and other craft and related trades 53% 2098k 23% 12% 3% 6% workers 91 Cleaners and helpers 50% 510k 22% 7% 0% 5% Agricultural, forestry and fishery 92 43% 2489k 9% 2% 0% 1% labourers 17 73 Handicraft and printing workers 39% 242k 14% 14% 1% 2% Subsistence farmers, fishers, hunters 63 37% 1498k 15% 8% 0% 1% and gatherers 22 Health professionals 36% 204k 72% 48% 23% 19% 32 Health associate professionals 35% 118k 60% 39% 8% 14% Market-oriented skilled agricultural 61 30% 4262k 15% 6% 0% 2% workers 94 Food preparation assistants 21% 18k 39% 21% 10% 0% Refuse workers and other elementary 96 20% 160k 17% 15% 0% 4% workers 51 Personal Services 16% 265k 28% 19% 3% 4% 42 Customer services clerks 13% 30k 82% 72% 65% 42% Software and applications developers 25 12% 13k 100% 87% 78% 21% and analysts Table 4, continued

Male em- % tasks # Male % used in- % used % has com- OCC Code Occupation description ployment switchable workers ternet computer puter share to online 74 Electrical and electronic trades workers 100% 738k 0% 67% 33% 0% Non-commissioned armed forces offi- 2 100% 21k 0% 59% 12% 24% cers 1 Commissioned armed forces officers 100% 3k 0% 39% 13% 12% 83 Drivers and mobile plant operators 100% 3361k 0% 17% 3% 6% Metal, machinery and related trades 72 99% 1450k 0% 20% 5% 8% workers 54 Protective services workers 99% 985k 0% 27% 5% 8% 3 Armed forces occupations, other ranks 99% 139k 0% 39% 8% 7%

18 Science and engineering associate pro- 31 99% 259k 1% 58% 28% 19% fessionals Business and administration associate 33 98% 755k 49% 54% 28% 21% professionals Street and related sales and service 95 98% 89k 0% 9% 0% 2% workers Building and related trades workers, 71 98% 2038k 0% 16% 3% 6% excluding electricians Numerical and material recording 43 98% 292k 0% 50% 28% 20% clerks Labourers in mining, construction, 93 97% 6974k 0% 8% 1% 2% manufacturing and transport 82 Assemblers 97% 39k 0% 40% 12% 19% Hospitality, retail and other services 14 97% 337k 6% 38% 10% 6% managers Figure 1: Share of jobs that could be done from home across Pakistan districts, 2018-19.

Note: Only four provinces of Pakistan are included in this Figure.

19 Figure 2: Distribution of work-from-home feasibility shares across urban and rural Pakistani districts, 2018-19. 20 Figure 3: Distribution of work-from-home feasibility shares across four Pakistani provinces, 2018-19. 21 Table 5: Percent of male and female students with home access to TV and internet in Pakistan, 2018–19

Education Total % Has In- % Has Province Region % Male % Female % Has TV level students ternet Computer Khyber Pakhtunkhwa Rural Primary 37883 64% 36% 33% 22% 4% Khyber Pakhtunkhwa Rural Secondary 22035 81% 19% 43% 30% 7% Khyber Pakhtunkhwa Rural Tertiary 718 81% 19% 78% 44% 20% Khyber Pakhtunkhwa Urban Primary 4460 57% 43% 46% 46% 13% Khyber Pakhtunkhwa Urban Secondary 3049 67% 33% 55% 52% 21% Khyber Pakhtunkhwa Urban Tertiary 163 62% 38% 83% 57% 41% Khyber Pakhtunkhwa MegaCities Primary 4270 61% 39% 57% 44% 11% Khyber Pakhtunkhwa MegaCities Secondary 2123 73% 27% 71% 50% 23% Khyber Pakhtunkhwa MegaCities Tertiary 219 78% 22% 97% 55% 43% Punjab Rural Primary 56421 56% 44% 19% 56% 3% Punjab Rural Secondary 28918 54% 46% 30% 63% 9% 22 Punjab Rural Tertiary 775 40% 60% 65% 71% 18% Punjab Urban Primary 13086 55% 45% 37% 68% 13% Punjab Urban Secondary 11282 46% 54% 53% 74% 26% Punjab Urban Tertiary 1109 25% 75% 82% 66% 34% Punjab MegaCities Primary 23154 53% 47% 41% 72% 10% Punjab MegaCities Secondary 18168 52% 48% 54% 76% 21% Punjab MegaCities Tertiary 1736 50% 50% 83% 69% 36% Sindh Rural Primary 18387 72% 28% 20% 39% 3% Sindh Rural Secondary 4925 82% 18% 38% 59% 7% Sindh Rural Tertiary 172 94% 6% 83% 76% 24% Sindh Urban Primary 9290 58% 42% 44% 70% 13% Sindh Urban Secondary 6242 67% 33% 59% 71% 22% Sindh Urban Tertiary 499 68% 32% 95% 63% 30% Sindh MegaCities Primary 6599 55% 45% 54% 74% 11% Sindh MegaCities Secondary 5599 57% 43% 70% 71% 23% Sindh MegaCities Tertiary 548 63% 37% 96% 45% 43%

Note: The number of students are in thousands, and the access to internet includes both mobile and broadband internet. Table 5, continued

Education Total % Has In- % Has Province Region % Male % Female % Has TV level students ternet Computer Balochistan Rural Primary 6728 73% 27% 19% 28% 0% Balochistan Rural Secondary 2849 76% 24% 30% 42% 2% Balochistan Rural Tertiary 65 85% 15% 61% 36% 18% 23 Balochistan Urban Primary 2291 69% 31% 31% 57% 11% Balochistan Urban Secondary 1015 71% 29% 44% 70% 19% Balochistan Urban Tertiary 54 82% 18% 74% 83% 47% Balochistan MegaCities Primary 1297 75% 25% 28% 56% 8% Balochistan MegaCities Secondary 1290 76% 24% 43% 66% 10% Balochistan MegaCities Tertiary 55 75% 25% 73% 71% 27%

Note: The number of students are in thousands, and the access to internet includes both mobile and broadband internet.