Mitigating the work-security trade-off
Editor’s Note: The paper behind this column appears in the second issue of CEPR’s new initiative, Covid Economics: Vetted and Real-Time Papers.
In this war, timing is essential. Italy has been the first Western country hit by the pandemic, and will likely be the first to be able to lift its most restrictive confinement measures. Four weeks down the road of the lockdown, with a health system almost at the limits of its capacity and some evidence that we are beyond the peak in the contagion, it is necessary to think at ways to mitigate the work-security trade-off and mobilise labour for the war against COVID-19.
Four issues are particularly relevant in this context:
1. How many jobs can be carried out while guaranteeing safety at work?
2. How many of these jobs can be activated as soon as the most severe restrictions to mobility will be lifted?
3. Will the changes required to make jobs ‘safer’ cause large productivity losses?
4. How many of these ‘safe’ jobs are essential in fighting the war against COVID-19?
In this column, we offer below some preliminary answers and hints at to how this analysis can be extended to other countries. In the absence of data in real-time on how firms and workers are adjusting to the crisis, our estimates are necessarily based on data collected in ‘normal times’. Evidence on behaviours adopted by firms and trades, like the provision of protective gear, may affect our conclusions, which must therefore be interpreted as a lower bound on the number of jobs that can potentially be carried out under the current circumstances.
How many safe jobs under the epidemic?
Working from home is obviously the safest way to perform your job without incurring the risk of contracting an infection. Although this risk is never zero, precautionary measures that do not interfere with the work itself can be envisaged (e.g. living separately from family members that work outside and can thus be a vehicle of contagion).
But how many jobs can be carried out remotely? In most European countries, the share of workers covered by teleworking or smart-working arrangements (including fully working from home) in normal times is below 10% (Eurofound 2017). The confinement has induced a spread of these arrangements among persons that so far were only mildly involved in this organisation of work. For instance, in Italy, 7 out of 10 managers interviewed in a survey carried out at the beginning of March by a managerial association (Manageritalia) declared having adopted smart-working practices for their employees – the first experience of this arrangement for about 40% of the workers involved. Taking the survey data at face value, we may expect that the number of workers involved has increased to reach about 15% of employment in the average EU country.
The crucial issue is, however, how many jobs can potentially be carried out remotely.
The results of this exercise are displayed in Table 1, which reports the share of jobs that can be carried out from home (online or through other means). We label these jobs as type 1 jobs. These jobs are mainly concentrated in services, with examples being professors, engineers, lawyers and architects.1 Within manufacturing, type 1 jobs concern mainly administrative and marketing activities. In Italy, manufacturing jobs account for around 20% of all jobs, whereas they account only for 11% of jobs of this category. Indeed, only 14% of all manufacturing jobs in Italy belong to category 1. A similar pattern is detected for the construction industry, with only 7% of all construction jobs included this category.
Table 1 Share of jobs that can be carried out form home.
Source: Authors’ calculations.2
How many more jobs with limited mobility and safe personal face-to-face contact?
As soon as the most restrictive confinement measures are lifted, it will be possible to carry out a wider range of occupations. In particular, there are jobs involving:
- limited mobility from home and no personal contacts (like veterinarians, animal caretakers, foresters and conservation workers, archivists, jewellers, chemists); and
- limited mobility and infrequent and safe face-to-face contacts (like mechanics, plumbers, electricians, drivers).
Table 2 offers our estimates (obtained using the same methodology, but applied to the measurement of the potential for smart working) for two categories of jobs in addition the type 1 jobs in Table 1. Type 2 jobs can be carried out when relaxing the mobility restriction (but not the personal face-to-face contact restraint). Type 3 jobs are possible when both the mobility and the face-to-face constraint are relaxed (assuming that these contacts are infrequent and interactions can happen at safe distance).
The result of these back-of-the-envelope calculations are sobering… even after allowing for limited face-to-face contact and mobility, the share of ‘safe’ jobs remains below 50% in all countries for which we carried out this exercise.
Table 2 Jobs that can be done with further relaxation of containment polices
The result of these back-of-the-envelope calculations are sobering. Overall, even after allowing for limited and face-to-face contact, the share of ‘safe’ jobs remains below 50% in all countries for which we carried out this exercise.
Will the changes required to make jobs ‘safer’ cause large productivity losses?
Even restricting our attention to ‘safe’ jobs, putting people back to work in a world in which COVID-19 has not yet been eradicated will likely require profound changes in the organisation of work. Remote working is one extreme example, which has already been investigated in normal times (e.g. Blinder and Krueger 2013, Bloom et al. 2015, Angelici and Profeta 2020), but also jobs that we categorised as Type 2 or Type 3 will probably need to be carried out differently than before. The impact on productivity, at least in the short-medium term, are likely to be negative, as workers will have to adapt to different ways of working.
While we do not have sufficient information to venture into predicting the evolution of productivity, we can try to get an idea of the distance between the way jobs are organised now and how they should ideally be organised in a world plagued by COVID-19. To do so, we resort to information collected by the OECD Survey of Adult Skills (PIAAC), which contains detailed information on what people actually do in their jobs. We focus on two dimensions that will be at the core of future changes in the organisation of work:
- the frequency of use of ICT devices, and
- the frequency of interactions with other people.
Arguably, workers who are less used to work with ICT equipment and who need to frequently interact with other people will be more at risk of experiencing a decline in productivity. Unfortunately, these data were collected in 2011/12, so they probably underestimate the current frequency of use of ICT devices. This implies that not only our classification of ‘safe’ jobs is a lower bound, but also that our considerations about the impact of productivity are probably pessimistic.
Putting people back to work in a world in which COVID-19 has not yet been eradicated will likely require profound changes in the organisation of work. … The impact on productivity, at least in the short-medium term, are likely to be negative, as workers will have to adapt to different ways of working.
Across all jobs belonging to types 1, 2 or 3 (‘safe’ occupations) and across the European countries for which PIAAC data are available,3 some 18% of workers are used to have very frequent social interactions, and 12% are not used to work frequently with ICT devices.
Luckily, the combination of these two features (i.e. jobs in which workers have to frequently interact with people and have a low use of ICT devices) are much rarer: only about 1% of workers in “safe” jobs fall in this category. This makes intuitive sense: as one of the main purpose of ICT is to ease communications among people, it is not surprising that occupations where a large share of workers make an intense use of ICT are also occupations where a large share of workers frequently interact with other people.
Only in a few “safe” occupations the share of workers who frequently interact with people and who are not used to work with ICT ranges between 3% to 5%: teachers, salespersons, legal and social professionals, and sports and fitness workers.
These encouraging results are consistent with experimental studies that find smartworking does not reduce individual productivity (Bloom, Liang, Roberts and Ying, 2014). However, these results reflect essentially a partial-equilibrium analysis. The fact that a large share of jobs will either be reorganised (in the case of safe jobs) or will not be able to resume (unsafe jobs) can generate negative productivity spillovers.4 An obvious example who comes to mind is the closure of schools and childcare facilities, who are likely to increase burdens on families and reduce the productivity of people working from home (Alon et al. 2020).
To what extent do these jobs contribute to fighting the war against COVID-19?
Unfortunately, the fraction of safe jobs is particularly low in most of the strategic industries involved the war against coronavirus, notably those that could help relaxing the capacity constraint in the health sector, such as the manufacturing of basic pharmaceutical products, of pharmaceutical preparations, of electromedical and measuring equipment and of medical instruments and supplies. These industries amount to around 1% of all jobs in Italy (0.86%) and are slightly underrepresented among jobs of type 1, but not among jobs of type 2 and 3.
Unfortunately, the fraction of safe jobs is particularly low in most of the strategic industries involved the war against coronavirus, notably those that could help relaxing the capacity constraint in the health sector, such as the manufacturing of basic pharmaceutical products, of pharmaceutical preparations, of electromedical and measuring equipment and of medical instruments and supplies.
The higher prevalence of jobs of type 2 and 3 is probably due to the fact that some activities in these industries need to be carried out in laboratories or other controlled environments where contacts among workers are very limited, but can’t be performed from home. The downside is that, while production is supposed to increase in order to supply sufficient equipment to face this emergency, the increase is likely to be limited given the context in which workers in these industries operate. Land availability is a major factor preventing a more dispersed assembly line and even more so in this specific case. Expanding production will require some reconversion of assembly lines that currently produce other goods and further investment in automation. Ironically, the robots and machines creating a lot of anxieties among workers for the future of their jobs are becoming, under COVID-19, a way to preserve labour, by allowing workers govern assembly lines from a distance and by creating more jobs in maintenance and supervisory roles.
As mentioned, evaluating the speed at which investments in automation or other forms of reorganisation of assembly and production lines is impossible with the data currently available. Scaling up real time data collection is therefore essential, not only for better evaluating how the epidemic is spreading (as rightly pointed out by Baldwin 2020, Bloom and Canning 2020, and Dewatripont et al. 2020), but also for evaluating the possibility of lifting the current restrictions to economic activity.
But how advanced is automation in manufacturing?
Clearly the answer varies from country to country. In Italy a survey carried out by the statistical office together with the employers association suggest that 19.3% of the firms in manufacturing with 10 or more employees use robotics in their activities. The broad sector including the industry producing equipment for the health sector (manufacturing of computer, clocks, optical products, electromedical and measuring equipment) displays an incidence of robotics in line with the average (19.0%).
Another option is to involve in production young workers who have a much lower mortality risk from Covid19 than the other age groups. There are, however, two problems with this strategy.
Most young people in Italy live with their parents, who face a much higher mortality risk … we estimate that only one out of four ‘unsafe’ jobs are currently held by workers aged under 35.
First, most young people, notably in Italy live, with their parents, who face a much higher mortality risk from Covid-19 contagion. It is also in this case possible to get some numbers and have an order of magnitudes. In particular, we estimate that only one out of four ‘unsafe’ jobs (the complement to the three categories described above) are currently held by workers aged under 35. Moreover, almost 6 out of 10 of these workers still live with her parents or older relatives. This implies that only a bit more than 10% of the unsafe jobs can be reactivated by young workers without running a substantial risk of increasing even further the death toll of COVID-19. Indeed, as suggested on Vox by Anelli et al. (2020), in principle one should separate the young workers who go back to work from the elderly and the immunocompromised.
The second problem is that, unsurprisingly, young people are overrepresented among the smart workers, the only jobs that have survived confinement. Hence there will be a potential production loss in other sectors if we were to reallocate young people from their current job to the strategic industries.
This is essentially a mismatch problem: we would like the elderly to have jobs that can be carried out from home, rather than the young. One possibility to solve this problem could be to mobilise workers who have been put in short-time work or temporary unemployment benefits. The former could be allowed to take up temporarily jobs in the strategic sector without losing the option to go back to their original occupation when the emergency is over (Giupponi and Landais 2020). This reallocation can be encouraged by integrating even further the income of the short-time workers, which means that we would combine the income support provision with a wage subsidy for the strategic industries.
Alon, T, M Doepke, J Olmstead-Rumsey and M Tertilt (2020), “The impact of COVID-19 on Gender Equality”.
Anelli, M, G Calzolari, A Ichino, A Mattozzi, A Rustichini and G Zanella (2020), “Transition steps to stop COVID-19 without killing the world economy”, VoxEU.org, 25 March.
Angelici, M and P Profeta (2020), “Smart-Working: Work Flexibility Without Constraints”, CESifo Working Paper No. 8165.
Baldwin, R (2020), “COVID-19 testing for testing times: Fostering economic recovery and preparing for the second wave”, VoxEU.org, 26 March.
Blinder, A S and A B Krueger (2013), “Alternative measures of offshorability: a survey approach”, Journal of Labor Economics 31(S1): S97-S128.
Bloom, D and D Canning (2020), “How widespread is coronavirus in New York? We need to know”, VoxEU.org, 29 March.
Bloom, N, J Liang, J Robertsand Z J Ying (2015), “Does working from home work? Evidence from a Chinese experiment”, The Quarterly Journal of Economics 130(1): 165-218.
Dewatripont, M, M Goldman, E Muraille and J Platteau (2020), “Rapidly identifying workers who are immune to COVID-19 and virus-free is a priority for restarting the economy”, VoxEU.org, 23 March.
Eurofound and the International Labour Office (2017), Working anytime, anywhere: The effects on the world of work, Publications Office of the European Union, Luxembourg, and the International Labour Office, Geneva.
Giupponi, G and C Landais (2020), “Building effective short-time work schemes for the COVID-19 crisis”, VoxEU.org, 1 April.
1 Indeed, professional and scientific activities are overrepresented in this category: workers in these occupations account for 17% of type 1 jobs, whereas they represent a share of around 6% of all Italian occupations. 64% of professional and scientific occupations belong to category 1, almost twice the average across all industries reported in the Table below.
2 In order to answer this question, we have used the O-Net (Occupational Information Network) classification, listing 968 occupations, and describing to which extent they require personal contacts. Unfortunately, O-Net does not specify whether these required contacts are face-to-face contacts or can also be organized on line or at least at a distance preventing COVID-19 contagion. Thus, we had to complement the O-Net classification with information from a survey of the Italian Statistical Office and INAPP (http://fabbisogni.isfol.it/) and our personal assessment. In particular, we classified every occupation listed in O-Net according to whether or not it could be carried out remotely. As the EU Labour Force Survey uses the ISCO classification, we mapped the O-Net coding into ISCO, passing through the SOC classification for which we had conversion tables available. Whenever there was not a 1-to-1 correspondence, we weighted averages using the US employment shares for each O-Net occupation between 2012 and 2014. This procedure should be able to deliver estimates that take into account the current level of technology and ICT present in occupations, without being based on US technology levels or on any specific dataset.
3 In order to achieve sufficient sample size at the three-digit occupational level, we pool data from Austria, Flanders (Belgium), Czech Republic, Denmark, Germany, Spain, Estonia, Finland, France, England, Northern Ireland, Greece, Ireland, Italy, Lithuania, the Netherlands, Poland, the Slovak Republic, Slovenia and Sweden.
4 Another factor we do not take into account, as it goes way beyond the scope of this exercise, is that a fall in aggregate demand may also affect productivity.
5 Anecdotal evidence tells us that this is already happening, with some textile companies starting to produce masks rather than clothes, for instance.