How many jobs can be done at home?

Updated 28 Apr 2020


While her daughters play in the room next door, Yuki Sato, a Japanese employee of a start-up company, works at home. — AFP/File
While her daughters play in the room next door, Yuki Sato, a Japanese employee of a start-up company, works at home. — AFP/File

THE coronavirus is changing the world faster than ever. In just a few months, it has forced the advocates of austerity to spend aggressively and even the biggest proponents of globalisation to argue for a worldwide shutdown, at least temporarily.

Within the world of employment too, it has brought remote working from the niche to the mainstream. Companies that two months ago had nothing to do with tech are jumping on the bandwagon, allowing telework wherever possible and quickly adopting tools like Zoom and Slack, which seems to be the only way to survive in the short term.

And that has brought a number of tech evangelists to start preaching the virtues of technology: how it could save the world and how it is inherently well-suited for social-distancing measures. News reports about Amazon hiring en masse and online sales surging are being used to support their theses on the growing obsolescence of the physical world. But how well does it match with the ground realities? The answer to that could be found in the labour market trends.

A recent white paper written by two Chicago Booth professors, Jonathan I. Dingel and Brent Neiman, is a good place to begin with as it explores the question: how many jobs can be done at home?

Technology’s potential is not limited, but its ability to save us from this pandemic is

For Pakistan, they find that the share of such jobs comes in at a miniscule 13.47 per cent — ahead of only 11 countries in the sample, including Afghanistan, Bangladesh and Mozambique (with the lowest value of 5.2pc) among others. On the other hand, developed nations have a proportion as high as 53.4pc in Luxembourg, 41.6pc in the United States and 36.7pc in Germany.

The percentage of teleworkable jobs in the country falls well short of the sample average of 26.4pc and is also considerably below the low and middle income group’s mean at 18.6pc.

Using the responses to Occupational Information Network (O*NET) surveys from the United States, they come up with a classification on what professions and roles are teleworkable — in other words, tasks that can be performed from home.

In addition to identifying which occupational groups and the US states fare well on teleworking using a quantitative score, they also compiled the rankings of 85 other countries based on the International Labour Organisation statistics. As one would expect, a positive correlation is found between the per-capita income (plotted on the horizontal axis) and the share of jobs that can be performed at home.

Among jobs that can be done at home, education services top the chart with the un-weighted share of teleworkability at a significant 83pc; followed by the professional, scientific and technical category 80pc; management of companies and enterprises 70pc; finance and insurance 76pc; and information 72pc.

Meanwhile, those with the lowest proportion were accommodation and food at just 4pc; agriculture, forestry, fishing and hunting 8pc; retail trade 14pc; construction 19pc; and transport and warehousing 19pc.

Keeping these occupational category scores and using sector-wise figures from the Pakistan Bureau of Statistics (PBS), we can do some back-of-the-envelope calculations to have our own rough estimates.

According to the available data, the five least teleworkable occupation groups employ a cumulative 68.7pc of the labour force (63.4m people), which comes in at 43.53m. If we factor in the share of jobs among these that can be worked from home, that gives us a rough figure of 4.18m, representing 9.6pc of the group’s total.

On the flip side, the four occupational categories with the highest proportion of jobs that can be performed at home (excluding management of companies and enterprises since there is no corresponding head for this in the PBS dataset) account for only 5.63pc of the labour force, that too driven by education’s 4.05pc.

In absolute terms, that means just 3.43m people. Adjusting for the respective teleworkability weights of these categories yields us a value of 1.44m, with education again accounting for a disproportionate share.

However, it must be pointed out that this methodology might run into one problem: the occupational telework scores given for the United States are possibly much higher than for Pakistan in the corresponding sectors. So the values must be taken as a weak proxy due to the absence of country-specific data.

One imperfect way to overcome that is to apply a discount factor to the US scores so they become more relevant to the local labour market conditions. After a number of iterations, 0.5 appears to be a reasonable value for that, which closely corresponds with the results provided by the two academics. Other works such as Gottlieb, Grobovšek and Poschke (2020) could also be possibly used to get an idea of the developing nations’ teleworkability though they did not have Pakistan in the dataset.

These estimates reveal the extent of potential of work from home, which takes us back to whether technology is really the panacea some believe it to be. Even in the oft-cited case of booming e-commerce amid the novel coronavirus, what is generally ignored is how that creates more demand for logistics, a very labour-dependent sector, thus ignoring the very human costs of people involved in providing those services.

Maybe a more nuanced understanding of labour markets and their underlying dynamics, especially within the context of a developing country like Pakistan, would help. This is not to say that technology’s potential is limited but rather its ability to save us from this (or the next) pandemic is.

Published in Dawn, The Business and Finance Weekly, April 27th, 2020