Department of Labor Logo United States Department of Labor
Dot gov

The .gov means it's official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Article
July 2021

Teleworking and lost work during the pandemic: new evidence from the CPS

To measure the effects of the coronavirus disease 2019 pandemic, the U.S. Bureau of Labor Statistics added questions to the Current Population Survey, the  main U.S. labor force survey, starting in May 2020. This article analyzes the results from questions asking people (1) whether they teleworked because of the pandemic and (2) whether they were unable to work because their employers closed or lost business because of the pandemic. We use the data on telework to refine work completed earlier in the pandemic that classified occupations on their suitability for telework. We then apply the revised classification to examine trends in telework and the extent to which working in an occupation suitable for telework shields workers from unemployment. Our results show that the pandemic resulted in a large increase in teleworking, with 33 percent of U.S. workers reporting teleworking because of the coronavirus in the period May-June 2020, before declining to a still substantial 22 percent in the fourth quarter. Rates of lost work varied widely both by an occupation’s suitability for telework and by demographic category.

The coronavirus disease 2019 (COVID-19) pandemic has had a momentous impact on the U.S. economy and on the labor market, in particular. In addition to eliminating millions of jobs, especially in the early months, the pandemic has dramatically changed the way work is performed. To measure the effects of the pandemic, starting in May 2020, the U.S. Bureau of Labor Statistics added questions to the Current Population Survey (CPS), the main U.S. labor force survey.1 All of these supplemental questions refer to activities at any time during the “last 4 weeks” prior to the survey and follow the monthly labor force questions. These questions ask whether

  1. people teleworked or worked from home because of the pandemic,
  2. people were unable to work because their employers closed or lost business because of the pandemic,
  3. they were paid for that missed work,
  4. the pandemic prevented job-seeking activities, and
  5. anyone in the household was prevented from seeking non-coronavirus-related medical care because of the pandemic.

In this article, we analyze the results from the first two of these added questions. Building on previous work by Jonathan I. Dingel and Brent Neiman early in the pandemic in which they classified occupations on their suitability for telework,2 we look both at the extent of telework and whether the people’s ability to telework mitigates the effect of the pandemic on employment. We use the new CPS data on people teleworking because of the pandemic to revisit the Dingel and Neiman classification scheme and suggest refinements to it. We then apply the revised classification to examine trends in telework and the extent to which working in an occupation suitable for telework shields workers from unemployment.

Telework rates

Table 1 shows how many workers teleworked because of the coronavirus pandemic in the May–December 2020 period. In May-June, fully a third of workers reported teleworking because of the pandemic. This proportion declined to 22 percent by the fourth quarter. Although the surveys are not strictly comparable, note that the 2017-18 Leave and Job Flexibilities Module of the American Time Use Survey (ATUS) showed only 13 percent of wage and salary workers had paid telework arrangements.3

Table 1. Employed people (in thousands) who teleworked because of COVID-19, by suitability for telework classification, May–December 2020
Suitability for telework categoryMay–June 2020July–September 2020October–December 2020
TotalTelework because of COVID-19PercentTotalTelework because of COVID-19PercentTotalTelework because of COVID-19Percent

Employed, 16 years and over

140,13646,67433.3146,50435,83224.5150,08333,39722.3

Original Dingel and Neiman occupation classification [1]

Suitable for telework

61,23535,05457.261,43427,21044.362,05625,56341.2

Not suitable for telework

77,09610,79314.083,1877,9789.686,1967,2888.5

Revised occupation classification

Suitable for telework

65,37335,88654.966,11529,14344.166,54327,02640.6

Not suitable for telework

72,9589,96113.778,5056,0457.781,7095,8247.1

[1] As discussed in the main text, our revised classification modifies that originally developed by Jonathan I. Dingel and Brent Neiman (“How many jobs can be done at home?” Journal of Public Economics, vol. 189, no. 2, September 2020).

Note: COVID-19 = coronavirus disease 2019; O*Net = Occupational Information Network.

Source: U.S. Bureau of Labor Statistics, Current Population Survey supplement.

Researchers have noted that many jobs cannot be performed remotely and require that workers be physically present at their worksites. The Occupational Information Network (O*NET) contains occupation-level measures that include not only the knowledge and skills required by an occupation but also how the work associated with the occupation is conducted and in what environment. Dingel and Neiman use these data to construct a division between occupations suitable for telework and those not suitable.

Table 1 also shows how the percentage of workers who teleworked because of the pandemic differs between occupations classified as suitable and those classified as not suitable for telework by Dingel and Neiman. (The revised classification shown in the table is discussed later.) Occupations classified as suitable for telework reassuringly had a much higher percentage of workers teleworking because of the COVID-19 pandemic. Over half of workers in these occupations teleworked because of COVID-19 in May–June, with the percentage declining to approximately 40 percent in the fourth quarter.

Occupations classified as not suitable for telework had a much lower percentage of workers responding that they were able to telework because of the pandemic. However, the proportion was still appreciable, with 14 percent reporting telework in May–June, declining to 8 percent in the fourth quarter. This finding raises the question of whether the classification can be improved.

Revising the classification of suitability for telework by occupation

We attempt to improve the Dingel and Neiman classification scheme by revising several occupations. We add several criteria classifying occupations as not suitable for telework if O*NET measures indicate working conditions are unlikely to be replicated in a home office environment. For example, we classify an occupation as unsuitable for telework if it involves workers spending a substantial amount of time kneeling, crouching, stooping, or crawling. We eliminate the criterion that a job is not suitable for telework when “Performing for or working directly with the public” is rated as very important. Finally, we replace the Dingel and Neiman condition that in order to be suitable for telework, an occupation must include using electronic mail at least once a month, with a three-prong condition that an occupation must involve regularly interacting with computers, spending time sitting, and working in an environmentally controlled indoor setting.4 (See appendix table 1.)

The appendix table 2 shows the proportion of workers in each four-digit North American Industry Classification System industry who are in occupations considered suitable for telework according to our classification (labeled “DFLP,” which stands for Dey, Frazis, Loewenstein, and Piccone) and the Dingel and Neiman classification. The appendix table 3 provides breakdowns by Metropolitan Statistical Area. These estimates are constructed with the use of microdata from the May 2019 Occupational Employment Statistics survey. With wage data from the same survey, both tables also show the average wage earned by workers in occupations that are deemed suitable for telework. Wages in these occupations are markedly higher. The average wage is $35.22 in occupations that we deem suitable for telework, compared with $20.31 in occupations that we classify as unsuitable for telework (data not shown in appendix table 2).5

The last two rows of table 1 show the proportions reporting teleworking because of COVID-19 in the revised classification. In the latter half of the year, the proportion of workers who report telework despite being in occupations classified as not suitable for telework is reduced by about 20 percent (1.5 to 2.0 percentage points) relative to the Dingel and Neiman equivalent. (The change in the proportion teleworking in occupations suitable for telework is small.) Overall, of workers who report being in occupations not suitable for telework, an appreciable fraction still reports teleworking. Rather than taking the labels “suitable for telework” and “not suitable for  telework” literally, we find that the two categories may be better interpreted as containing occupations in which conditions are favorable or not favorable for telework. Given the serious concerns about working onsite during a pandemic, some workers may still telework, although their productivity is substantially lower as a result. Teachers are a well-known example of workers who fall into our unable-for-telework category. Research indicates that remote learning by elementary and high school students is less effective than onsite instruction.6 Yet, averaged over the May-June period, the percentage of preschool and kindergarten, elementary and middle school, secondary school, and special education teachers teleworking was 78.0 percent. This rate declined to 38.1 percent in the fourth quarter, a percentage decline similar to that for other occupations in the not-suitable-for-telework category. If one were to remove teachers from the not-suitable-for-telework category, the percentage of workers in this category who are teleworking would be 8.9 instead of 13.7 in the May-June period and 4.9 instead of 7.1 in the fourth quarter.

Table 2 shows telework rates of employed people in occupations classified as suitable and unsuitable for telework for various demographic groups averaged over the May-December 2020 period. For most demographic breaks, telework rates are similar within the suitable-for-telework categories. Different concentrations in the suitable- and not-suitable-for-telework occupations appear to account for much of the variation in total telework rates across demographic groups. Educational attainment is an exception to the similarity within occupation categories. The teleworking rate for bachelor’s degree holders who worked in occupations classified as not suitable for telework was 18 percent, and for advanced degree holders in those occupations, the rate was 35 percent. In all other categories of educational attainment, the teleworking rate for occupations classified as not suitable for telework was 6 percent or less. Similarly, the teleworking rate for bachelor’s degree holders in occupations classified as suitable for telework was 53 percent, and for advanced degree holders, the rate was 62 percent. In contrast, the comparable rate was 33 percent or less for other categories of educational attainment.

Table 2. Employed people (in thousands) in occupations classified as suitable and unsuitable for telework, who teleworked because of COVID-19, by demographic category, May–December 2020
Demographic categoryAllSuitable for teleworkNot suitable for telework
TotalTelework because of COVID-19PercentTotalTelework because of COVID-19PercentTotalTelework because of COVID-19Percent

All workers

146,25437,62925.766,09030,03545.478,3206,9418.9

Gender

Male

77,92217,75622.832,16814,71245.745,0482,8586.3

Female

68,33319,87329.133,92115,32345.233,2724,08312.3

Race

White only

114,32728,78525.251,92322,80943.961,0045,4809.0

Black only

17,5443,78021.56,7262,88042.810,5438097.7

Asian only

9,3073,88041.75,4413,43363.13,76141010.9

All other

5,0761,18423.32,00091345.73,0122438.1

Age, years

16 to 24

17,1762,05111.94,3641,51534.712,6114793.8

25 to 54

94,22127,17828.844,40821,66048.848,6125,06110.4

55 to 64

25,1846,24524.812,4365,11241.112,4551,0408.4

65+

9,6732,15522.34,8821,74835.84,6423617.8

Hispanic ethnicity

Hispanic

25,6114,21916.58,0363,11938.817,2821,0065.8

Non-Hispanic

120,64333,41027.758,05326,91646.461,0385,9359.7

Marital status

Married

80,14222,93028.640,09118,41945.939,0984,17610.7

Never married

46,77010,31022.017,5468,17046.628,6111,9096.7

Other marital status

19,3434,39022.78,4533,44640.810,6118568.1

Educational attainment

Less than a high school diploma

10,4143453.31,06114413.59,2821982.1

High school graduate, no college

36,7223,2248.89,4732,23723.626,8909173.4

Some college, associate’s degree

39,1126,61216.915,7445,12532.622,8321,3495.9

Bachelor’s degree only

37,53915,22640.623,79312,64153.113,1912,32717.6

Advanced degree

22,46812,22354.416,0189,88861.76,1252,15135.1

Note: COVID-19 = coronavirus disease 2019; O*Net = Occupational Information Network.

Source: U.S. Bureau of Labor Statistics, Current Population Survey supplement.

In previous work, Matthew Dey, Harley Frazis, Mark A. Loewenstein, and Hugette Sun showed that in occupations in which telework is feasible, the proportion of workers who actually teleworked (hereafter, the “takeup rate”) was particularly high before the pandemic for workers in management, professional, and sales occupations—over 20 percent in all three groups—in the ATUS and the National Longitudinal Survey of Youth.7  Takeup rates in all other occupation groups were 10 percent or less. Grouping occupations into high and low takeup-rate categories accordingly, chart 1 shows trends in telework because of the pandemic in the May-December 2020 period for our three occupational categories: suitable for telework and high prepandemic takeup rate, suitable for telework and low prepandemic takeup rate, and unsuitable for telework.

Although the CPS and the prepandemic surveys are not entirely comparable, the pandemic has clearly caused an increase in telework for both low and high takeup-rate occupations. However, although one might speculate that low takeup-rate occupations would have had a particularly large increase in telework and that this increase would have grown with time as employers adapted to pandemic conditions, this did not occur. In both high and low prepandemic take-up rate occupations, telework rates were high relative to their levels in prepandemic surveys.8 Telework rates decreased for both occupation groups later in the year but remained high compared with prepandemic levels. Throughout the second half of 2020, the telework rate in high prepandemic takeup-rate occupations remained substantially above the rate in low takeup-rate occupations.

 

Chart 2 shows how teleworking rates varied in the May-December 2020 period by educational attainment. Teleworking rates declined rather rapidly from May to July for bachelor’s degree and advanced degree holders in occupations classified as not suitable for telework. However, the rates later in the year were still much higher than those in lower educational attainment groups who were also in occupations classified as not suitable for telework. This finding suggests that telework was a temporary expedient for at least some workers with high levels of education.

Rates of lost work because of COVID-19

What were the direct effects of the pandemic on employment, and how did this vary by the suitability for telework? Among the questions added to the CPS to track the effects of COVID-19 was “At any time in the last 4 weeks, were you unable to work because your employer closed or lost business due to the coronavirus pandemic?”9 Chart 3 shows trends in the number of persons who lost work because of the coronavirus pandemic. The underlying population for chart 3 is the experienced labor force—the employed, plus the unemployed with previous work experience. The number of workers who lost work in the last 4 weeks declined from 42 million in May 2020 to 12 million in October 2020 before leveling off. In each month, most of these workers were in occupations classified as not suitable for telework, although both types of occupations showed declines in lost work as the year progressed.

Chart 4 shows the percentage of workers reporting lost work because of the pandemic. In May 2020, 27 percent of workers lost work, declining to approximately 8 percent in each month of the fourth quarter. As might be expected, workers in occupations classified as suitable for telework were less likely to report that they had lost work because of the pandemic. In May, 34 percent of workers lost work in occupations classified as not suitable for telework, in contrast to 19 percent of workers in occupations classified as suitable for telework. These rates declined to 9 percent and 6 percent, respectively, by the fourth quarter.

Some persons may not have searched for work and instead may have withdrawn from the labor force in response to losing a job because of the pandemic. Charts 5 and 6 repeat charts 3 and 4 for the population of persons who withdrew from the labor force after working within the last 12 months. Over 4 million persons who were out of the labor force in May 2020 and who had worked within the previous 12 months reported losing work in the last 4 weeks because of the pandemic. This number declined to less than 2 million by the end of the year. These were predominantly persons whose most recent job was in an occupation classified as not suitable for telework—over 3 million such persons in May, for example.

As chart 6 shows, the numbers in the previous paragraph are a large percentage of withdrawals from the labor market by the recently employed. In May, approximately 43 percent of those not in the labor force who had worked in the last 12 months reported losing work in the last 4 weeks because of the pandemic. The percentage declined to a still substantial 20 percent by the end of the year.

Table 3 shows averages for May to December 2020 by demographic category for workers reporting lost work because of the pandemic. Both the demographic group and whether the occupation is classified as suitable for telework are important determinants of lost work. For example, 12.9 percent of Hispanics in occupations classified as suitable for telework reported losing work, compared with 19.9 percent of Hispanics in occupations not suitable for telework. For comparison, the equivalent numbers for non-Hispanics were 9.9 percent and 15.7 percent, respectively.

Table 3. People (in thousands) in occupations classified as suitable and unsuitable for telework who were unable to work because of COVID-19, by suitability for telework designation from O*NET, for experienced labor force, May–December 2020
Demographic categoryAllSuitable for teleworkNot suitable for telework
TotalUnable to work because of COVID-19PercentTotalUnable to work because of COVID-19PercentTotalUnable to work because of COVID-19Percent

All workers

159,77222,13713.969,9917,18010.387,69114,61416.7

Gender

Male

84,79811,19913.233,8423,2979.750,1347,76715.5

Female

74,97510,93814.636,1493,88310.737,5576,84718.2

Race

White only

123,76216,36013.254,7485,44810.067,45110,66515.8

Black only

19,9843,23716.27,30691612.512,3622,27418.4

Asian only

10,2911,53915.05,7625138.94,40499822.7

All other

5,7351,00117.52,17530514.03,47467819.5

Age, years

16 to 24

19,9353,09115.54,88959312.114,7922,45616.6

25 to 54

102,10213,62313.346,7724,4689.653,9948,96116.6

55 to 64

27,2083,65713.413,1361,38810.613,7442,20516.0

65+

10,5281,76616.85,19473114.15,16199319.2

Hispanic ethnicity

Hispanic

28,7255,08917.78,6981,12112.919,7013,91519.9

Non-Hispanic

131,04817,04913.061,2936,0599.967,99010,69915.7

Marital status

Married

85,35310,62512.441,8423,9389.442,4636,52215.4

Never married

53,1628,20915.419,1122,17011.433,3185,91717.8

Other marital status

21,2573,30315.59,0381,07311.911,9092,17618.3

Educational attainment

Less than a high school diploma

11,9032,21318.61,16317915.410,6552,01518.9

High school graduate, no college

41,0986,35015.510,2251,20711.830,4385,06616.6

Some college, associate’s degree

43,1846,65715.416,9962,12412.525,5814,42917.3

Bachelor’s degree only

40,1514,73111.825,0302,3719.514,5112,28215.7

Advanced degree

23,4372,1869.316,5781,2997.86,50682312.6

Note: COVID-19 = coronavirus disease 2019; O*Net = Occupational Information Network.

Source: U.S. Bureau of Labor Statistics, Current Population Survey supplement.

As with the reported teleworking category, educational attainment shows a particularly strong effect on lost work. For example, advanced degree holders in occupations classified as not suitable for telework report lost work at approximately the same percentage as those with some college in occupations classified as suitable for telework (12.6 vs. 12.5 percent). Recall from table 2 that workers with more education had higher rates of telework even after controlling for occupational suitability for telework. So, at least part of the advantage of more highly educated workers with respect to lost work likely is due to suitability for telework of their jobs not captured by our occupational classification.

Chart 7 shows trends in lost work because of the pandemic in the May-December 2020 period, by educational attainment and occupational suitability for telework. The chart shows that the greatest differences between groups occurred early in the period when the percentage reporting lost work was greatest.

Conclusion

This article has examined the reaction of the U.S. labor market to the COVID-19 pandemic, using questions added to the CPS in May 2020. We analyzed the prevalence of both telework and lost work because of the pandemic in the May-December 2020 period. Our major focus was on how these outcomes varied by an occupation’s suitability for telework. To aid in our analysis, we revised a commonly used classification to reduce the incidence of telework by workers classified as unable to telework.

The pandemic resulted in a large increase in teleworking, with 33 percent of U.S. workers reporting that they had teleworked because of the coronavirus pandemic in May-June before the percentage declined to a still substantial 22 percent in the fourth quarter. The suitability of occupations for telework is, unsurprisingly, an important determinant of this rate.

The pandemic also directly caused substantial rates of lost work. Rates of lost work varied widely both by an occupation’s suitability for teleworking and by demographic category. As having an occupation classified as suitable for telework is itself correlated with demographic characteristics, workers with characteristics associated with high-telework occupations enjoyed a substantial advantage in weathering the pandemic.

Although falling from their peak at the start of the pandemic, teleworking rates are still considerably higher than before the pandemic. It seems likely that some of the increase in teleworking will be permanent as workers and employers gain experience with teleworking arrangements and with the information technology that helps facilitate teleworking.10 Although teleworking entails some costs and limits some of the interactions that occur among coworkers, it still provides benefits even in the absence of a pandemic. For example, employers can potentially economize on office space. Workers who telework only part of the time save on commuting time and costs and have more flexibility in managing their household tasks. Workers who are full-time teleworkers are not constrained to live near their employer, thereby enlarging the set of potential employer-worker matches.

Appendix. Breakdown of teleworking suitability of occupations

Appendix table 1. O*NET categories and their variables and cutoffs used to classify occupations as suitable for telework by Jonathan I. Dingel and Brent Neiman and DFLP

Appendix table 2. OES-based employment and mean wage estimates, by four-digit NAICS and suitable for telework

Appendix table 3. Occupational Employment Statistics-based employment and mean wage estimates, by detailed MSA and suitable for telework category

Suggested citation:

Matthew Dey, Harley Frazis, David S. Piccone Jr, and Mark A. Loewenstein, "Teleworking and lost work during the pandemic: new evidence from the CPS," Monthly Labor Review, U.S. Bureau of Labor Statistics, July 2021, https://doi.org/10.21916/mlr.2021.15

Notes


1 For more information regarding the added questions that measure the effects of the coronavirus disease 2019 pandemic, see https://www.bls.gov/covid19/measuring-the-effects-of-the-coronavirus-covid-19-pandemic-using-the-current-population-survey.htm.

2 Jonathan I. Dingel and Brent Neiman, “How many jobs can be done at home?” Journal of Public Economics, vol. 189, no. 2, September 2020. See also Matthew Dey, Harley Frazis, Mark A. Loewenstein, and Hugette Sun, “Ability to work from home: evidence from two surveys and implications for the labor market in the COVID-19 pandemic,” Monthly Labor Review, June 2020, https://doi.org/10.21916/mlr.2020.14, assessing the Dingel and Neiman classification using telework rates in earlier datasets.

3 See Harley Frazis, “Who telecommutes? Where is the time saved spent?” BLS Working Paper 523, April 2020, https://www.bls.gov/osmr/research-papers/2020/ec200050.htm. Frazis considers workers as teleworkers if they work entirely at home on some days. Current Population Survey respondents are not asked how many hours or days they worked at home.

4 See appendix table 1 for a comprehensive list of the Occupational Information Network variables and cutoffs used in both the Dingell and Neiman and our classification schemes (called “DFLP,” which stands for Dey, Frazis, Loewenstein, and Piccone).

5 Dingel and Neiman have also observed that wages are higher in occupations that are suitable for telework.

6 See Susanna Loeb, “How effective is online learning? What the research does and doesn’t tell us,” Education Week, March 20, 2020, https://www.edweek.org/technology/opinion-how-effective-is-online-learning-what-the-research-does-and-doesnt-tell-us/2020/03.

7 Dey et al., “Ability to work from home,” p. 9.

8 Telework rates most likely jumped up in April 2020, but May is the first month for which we have data.

9 This question was asked of all individuals, regardless of their labor force status at the time of the survey.

10 Both the popular press and the economics literature widely speculate that telework will be substantially higher after the pandemic than before. For example, see Jose Maria Barrero, Nicholas Bloom, and Steven J. Davis, “Why working from home will stick,” Working Paper 28731 (Cambridge, MA: National Bureau of Economic Research, April 2021), https://www.nber.org/papers/w28731, who surveyed workers about their expectations to telework after the pandemic.

article image
About the Author

Matthew Dey
dey.matthew@bls.gov

Matthew Dey is a research economist in the Office of Employment and Unemployment Statistics, U.S. Bureau of Labor Statistics.

Harley Frazis
frazis.harley@bls.gov

Harley Frazis is a research economist in the Office of Employment and Unemployment Statistics, U.S. Bureau of Labor Statistics.

David S. Piccone Jr
piccone.david@bls.gov

David S. Piccone Jr is a statistician in the Office of Employment and Unemployment Statistics, U.S. Bureau of Labor Statistics.

Mark A. Loewenstein
loewenstein.mark@bls.gov

Mark A. Loewenstein is a senior research economist in the Office of Employment and Unemployment Statistics, U.S. Bureau of Labor Statistics.

close or Esc Key