https://orcid.org/0000-0002-8072-4134
Figures
Abstract
While ensuring employment opportunities is critical for global progress and stability, workers are now subject to several disruptive trends, including automation, rapid changes in technology and skill requirements, and transitions to low-carbon energy production. Yet, these trends seem almost insignificant compared to labor impact of the COVID-19 pandemic. While much has been written about the pandemic’s short-term impacts, this study analyzes anticipated long-term impacts on the labor force of 2029 by comparing original 2029 labor projections to special COVID-adjusted projections recently published by the US Bureau of Labor Statistics. Results show that future demand for nearly every type of labor skill and knowledge will increase, while the nature of work shifts from physical to more cognitive activities. Of the nearly three million jobs projected to disappear by 2029 due to COVID, over 91% are among workers without a bachelor’s degree. Among workers with a degree demand shifts primarily from business-related degrees to computer and STEM degrees. Results further show that the socialness of labor, which is important for both innovation and productivity, increases in many more industries than it decreases. Finally, COVID will likely accelerate the adoption of teleworking and slightly decrease the rate of workforce automation. These impacts, combined with a shift to more cognitive worker activities, will likely impact the nature of workforce health and safety with less focus on physical injuries and more on illnesses related to sedentary lifestyles. Overall, results suggest that future workers will need to engage more often in training and skill acquisition, requiring life-long learning and skill maintenance strategies.
Citation: Shutters ST (2021) Modelling long-term COVID-19 impacts on the U.S. workforce of 2029. PLoS ONE 16(12): e0260797. https://doi.org/10.1371/journal.pone.0260797
Editor: Kjell Torén, University of Gothenburg: Goteborgs Universitet, SWEDEN
Received: August 25, 2021; Accepted: November 16, 2021; Published: December 1, 2021
Copyright: © 2021 Shade T. Shutters. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data underlying the results used in the study are available from (1) O*NET online at: https://www.onetcenter.org/dl_files/database/db_25_2_text.zip (O*NET version 25.2); (2) US Census Bureau at: https://www2.census.gov/programs-surveys/acs/data/pums/2019/1-Year/csv_pus.zip (2019 American Community Survey, 1-year Public Use Microdata Sample); and (3) Harvard Dataverse at: https://doi.org/10.7910/DVN/DVHBZT (all remaining required data).
Funding: STS was supported by Grant 2029146 from the US National Science Foundation www.nsf.gov. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the National Science Foundation.
Competing interests: The author has declared that no competing interests exist.
Introduction
More than 3 million people globally migrate to cities each week, primarily in search of employment opportunities [1]. In a 2016 U.S. White House report on the future of cities, listed first among challenges facing urban residents was “finding and acquiring a good job, a quality education, and appropriate training” [2]. Similarly, in its 2016 report to Congress, the U.S. National Science Foundation designated “The Future of Work” as one of its 10 Big Ideas for Future Investment and a critical focus for future research. [3]. Thus, it is broadly acknowledged that improving human well-being is critically dependent on providing quality jobs and ensuring workers have appropriate skills.
Yet several ongoing trends have accelerated in recent years to cast uncertainty on the nature of work in the near future. Countries of the world are beginning clean energy transitions that will likely cause severe economic and labor disruptions in the short term [4,5], leading to possible high unemployment, increased migration, uncertainty in tax revenues, and social unrest [6,7]. Workers simultaneously face the specter of the so-called Fourth Industrial Revolution [8]–the use of robots and artificial intelligence to replace human labor. McKinsey Consulting estimates that 800 million jobs will disappear globally by 2030 [9], with unemployment potentially surpassing that experienced during the 2007–09 Great Recession [10]. At the same time, efficiency gains from automation is expected to add $16 trillion to the global economy by 2030 [11]. This means automation-related increases in global wealth will be shared by an ever-shrinking number of workers, intensifying inequality and accelerating regional fragility [12].
In the midst of these disruptive trends the COVID-19 pandemic (hereafter COVID) erupted in early 2020. The impacts of this unprecedented shock have been multiple and complex, including business closures, loss of child-care options, reduced consumer spending, and other impacts, none of which are isolated, but which interact in subtle ways to exacerbate the pandemic’s aggregate effects [13–17]. The pandemic has also exacerbated issues of labor inequality [18–20] and accelerated ongoing trends such as the adoption of telework [21,22]. Yet, most of what has been learned regarding impacts of the covid pandemic is focused on currently observable effects.
In this study I take advantage of a unique dataset recently published by the US Bureau of Labor Statistics (BLS) to isolate long term impacts on the skills and education composition of the US labor force. Because of the importance of person-to-person interaction in innovation and productivity, I also examine how the pandemic is likely to shape the socialness of future work, both in aggregate and within individual industries. Further, I determine whether the pandemic is likely to alter ongoing trends in automation and teleworking. Finally, I explore how COVID-induced changes in the nature of future work might affect the occupational health and safety of workers.
Covid and the future labor force
Each year the BLS projects US national employment by industry and occupation 10 years into the future. This dataset, known as the National Employment Matrix, is typically published with a lag of 1–2 years. For instance, the 2029 projections were published in late 2020. Because of the unprecedented impact of the COVID pandemic on US labor, the BLS took the unusual step of preparing alternate scenarios of future employment projections, each assuming different long-term impacts of COVID on the structure of US labor. Two scenarios, termed medium and strong impact scenarios, were published in February 2021 modelling the impact of long-term effects of COVID, such as the transition to telework and movement from brick-and-mortar retail to eCommerce. These trends not only change labor demand in the affected industries but alter consumer behavior and supply chains in related industries.
A BLS analysis of these scenarios presents an excellent discussion of the industry and occupational shifts under the two COVID scenarios [23]. Among industry sectors, the authors show that accommodation and food services will suffer the largest decreases in employment due to COVID. Employment in arts, entertainment, recreation, retail trade, and construction also decreases significantly under the strong COVID scenario. On the other hand, the information sector as well as professional, scientific, and technical services are projected to increase employment in response to the pandemic. In terms of occupations, the report finds that food preparation/serving and sales jobs will experience the largest decreases in employment, while science, computer, and mathematical occupations will experience significant employment increases under the strong COVID scenario.
Firms, workers, and educators may be significantly affected by these changes in the future workforce. To better plan for those changes, it is important to consider factors beyond demand for individual occupations. Workers and economic planning agencies must be aware of changes in skills requirements while educational institutions must be ready to shift training options to meet the new needs of firms. Planners must also anticipate how trends in teleworking might impact infrastructure. Building on [23] this paper contributes to addresses these lingering issues by analyzing the long-term impacts of COVID on the future labor force in terms of skills composition, degree distributions, automation potential, and ability to telework.
In is important to clarify that this study does not compare employment patterns of today with employment patterns of 2029, which is covered in [23]. Instead, this study isolates the long-term impacts of COVID on the future workforce by comparing the 2029 strong and medium COVID employment scenarios to the original 2029 employment projections, which I refer to as the baseline scenario.
Data and methods
Employment projections
Employment projection data were published as three individual scenarios. For the baseline scenario I use the original 2019–2029 National Employment Matrix [24], published in late 2020. The medium and strong COVID impact scenarios were special data sets prepared by the BLS in 2021 [25]. These scenarios contain a mix of industries and occupations at multiple aggregation levels. Here, I include only ‘Line item’ occupations to avoid duplicate counting.
Occupational skills and activities
To examine the COVID impact on future worker skills and activities I utilize the ONET dataset [26]. ONET decomposes US occupations into two types of attributes describing the nature of an occupation–elements, which have a continuous value, and intermediate work activities (IWAs), which have a present/absent flag. ONET element data are further grouped into element categories, which include skills, knowledge, abilities, and others. Some element categories, such as education, do not have a single value for an occupation but instead have a distribution. Such categories that do not have a single value per occupation are not included in this analysis. Throughout this analysis ONET elements collectively are referred to as skills.
Elements are further defined by a measurement scale, such as importance or level. Here I use values for importance (IM), except in the case of Work Context elements, which have no values for either importance or level. For those elements I instead use the context (CX) scale values.
Because ONET uses an expanded occupation code, those codes must be mapped to BLS occupation codes with a crosswalk file. Many ONET codes map to one BLS code. Thus, ONET values are mapped to BLS codes by simple averaging. The ONET to BLS occupation mapping must be created manually and is specific not only to the version of ONET but to the year and dataset of the BLS. Here the crosswalk is specific to ONET version 25.2 and the BLS’s 2019 National Employment Matrix.
As with elements, IWAs must be mapped to BLS occupations codes. Since there is no value to collapse by averaging, we adopt the method used in [27] to assign a present/absent indicator for each BLS occupation. If an activity is present in any one of the ONET occupations that map to a single BLS occupation, it is designated as present in the BLS occupation.
College degrees and degree groups
To determine how the future distribution of college degrees changes under each COVID projection scenario I first create a probability distribution function for degrees by occupation derived from empirical data. This is obtained from the US Census Bureau’s 2019 American Community Survey (ACS) 1-year Public Use Microdata Sample (PUMS) [28]. These data comprise a survey of approximately 1% of the US population. Each record in this microdata represents a single individual but carries a weight factor which estimates of the total number of similar individuals in the individual’s geographic area.
While the microdata contains covers several hundred attributes of each survey respondent, I extract only occupation code (SOCP), 1st degree field of study (FOD1P), and person weight (PWGTP). If the field of degree is blank the respondent is designated as having no degree. Individuals without an occupation code are excluded from this study.
Like many US agencies, the Census Bureau maintains an idiosyncratic list of occupation codes that do not exactly match those used by the BLS. Thus, I construct a crosswalk to match many census occupation codes to one BLS occupation code. With this crosswalk, a degree probability distribution by census occupation is restated as a degree probability distribution by BLS occupation. This distribution of approximately 175 degrees is further aggregated into 37 degree groups and results for both degrees and degree groups are applied to occupational projections of each COVID scenario. COVID impact on degree demand is calculated both as the change in projected number of degrees in the national workforce and as a percentage of all workers.
Industry “socialness”
I also calculate the “socialness” of industries under the two COVID projection scenarios. Socialness can significantly impact an industry’s projected productivity as well as an economy’s rate of innovation. I this study I calculate industry socialness as the number of social IWAs performed by workers of an industry divided by the industry’s total labor force. This measure of industry socialness as the number of social IWAs per worker. I adopt the determination of which IWAs are social from Painter et al [27] and apply these to each occupation in the industry’s employment projections. This allows comparison of industry socialness between the baseline projection and the two COVID impact scenarios. Results are calculated for both detailed industries (4-digit NAICS) and industry sectors (2-digit NAICS).
One exception is noted: Industry 453100 (Florists) was included in the baseline BLS projections but not in either COVID impact scenario and is thus excluded from the analysis.
Job automation
To estimate job automation risk at the occupation level I use Frey & Osborne’s metric of automation potential by individual occupation code [29]. Those probabilities area applied to occupation-level employment under each COVID scenario to estimate projected changes in total US workforce susceptible to automation. These measures are also applied to employment in individual industries and industry sectors to understand how COVID may impact automation within industries.
Teleworking estimates
To examine the impact of different scenarios on the future propensity to telework, I use a 2020 study from the U.S. National Bureau of Economic Research [30]. This study analyzes each occupation’s ONET characteristics to determine which occupations can be performed remotely [31]. I then apply those occupational designations to baseline and COVID impact projection scenarios and aggregate to both industry and national levels.
Results
In each of the following analyses the baseline 2029 employment projections are compared to both the medium and strong COVID impact scenarios for 2029 and changes under each COVID scenario are presented.
Impact on occupational elements and worker activities
COVID impacts on projected per capita importance of ONET elements is presented in Table 1. Elements decreasing the most in importance are largely associated with physical tasks, such as standing, walking, running, or moving objects, and with skills likely to be automated, such as interacting with the public. On the other hand, elements associated with data and data manipulation have the largest increases in per capita importance.
Results shown in Table 1 include all elements defined by ONET that have a single value for an occupation. However, this combines five categories of elements which may or may not be comparable despite sharing the same scale and range of values. Therefore, I present also elements having the largest positive and negative changes in each category in Table 2.
Not one element in the skills category exhibits a decrease in importance. Similarly, all but four of the elements in the Knowledge category increase in importance under the COVID scenarios.
In addition to ONET elements, ONET also decomposes occupations into a set of over 200 Intermediate Work Activities (IWAs) which are tabulated as either present or absent. Table 3 presents the impact of COVID on the fraction of total workers performing each IWA.
Under both the strong and medium scenarios of COVID impact, demand for skills shifts to areas concentrated in computers and health care. These include skills in programming, chemistry, biology, and several related scientific and medical fields.
The IWA with the highest increase in demand is maintenance of current knowledge and expertise, suggesting that future workers will be required to engage more in life-long learning and continuous skill acquisition [32].
Impact on demand for college degrees
While the previous results demonstrate the increased importance of skills and knowledge, here we examine how those changes are manifest across college degrees. The largest COVID impacts on 173 different degrees are presented both as total degrees (Table 4) and as percent of total workers (Table 5). Results reveal the largest decreases in demand are concentrated in business degrees, while the largest increases are in STEM degrees, particularly those related to computer science and engineering. Viewing results instead as a percent of workforce (Table 5) changes the story significantly. Only three of 173 degrees are projected to decrease as a percent of workforce–hospitality management, cosmetology and culinary arts, and drama and theater arts.
Detailed degree data are further aggregated into 37 degree groups and COVID impacts area show both in terms of total degrees (Table 6) and percent of workforce (Table 7). At this level of aggregation several degree groups decrease in total quantity, with business degrees having the largest drop. However, only the degree group personal and culinary services decreases as a percent of workforce.
Impact on industry and sector workforce socialness
COVID impacts on industry socialness are calculated for both 2-digit industry sectors and 4-digit detailed industries. Among 2-digit sectors (Table 8), only 3 industry sectors decrease in socialness under the strong COVID scenario, while all others increase. Among 4-digit industries (Table 9) several industries show decreases in socialness. Interestingly, one of these is scientific research and development (5417), which displays a marked decrease in socialness even though the sector of which it is part, professional, scientific and technical services (54), exhibits one of the largest increases in sociality among sectors. This is indicative of a large degree of heterogeneity in attributes of the individual industries that comprise an industry sector.
Impact on workforce automation
Using Frey & Osborne’s [29] occupation level metric of automation potential, I calculate the fraction of jobs in each 2-digit industry sector (Table 10) and 4–digit industries (Table 11) that are susceptible to automation. Results reveal that among 2-digit sectors, only construction (23) and transportation/warehouse (48–49) have an increased probability of automation under the strong COVID scenario.
Impact on teleworking
Analyzing COVID’s impact on the probability that jobs can be performed remotely we find that results are highly skewed toward higher probability of teleworking (Table 12). Among 2-digit industry sectors only construction (23) displayed a decrease in fraction of jobs that can be performed remotely. This suggests that long term impacts of COVID include an acceleration of telework adoption by workers.
While several industries at the 4-digit industry do exhibit decreases in potential for telework, the magnitude of those changes is relatively small compared to the industries expected to increase adoption of teleworking (Table 13).
Discussion
The following discussion covers several topics related to this study. For clarity, the study’s key findings are summarized in Table 14.
The nature of future work: Skill requirements and job activities
Results generally suggest that the pandemic will induce future work to be less physical and more cognitive in nature. For instance, worker abilities with the largest increases in importance (Table 2) include reasoning, comprehension, and fluency of ideas, which are all cognitive abilities. On the other hand, worker abilities decreasing most in importance are physical attributes of workers, such as stamina, strength, and flexibility.
Worker skills in particular increase in importance. In fact, not one ONET skill decreases in importance under either COVID scenario. This suggests that skill acquisition by workers will be central to future employment and further highlights the key role that education, training, and life-long learning will play in preparing and maintaining the future workforce.
A similar pattern is exhibited by COVID impacts on intermediate work activities (Table 3). Future workers will perform fewer routine and repetitive tasks, such as cleaning tools or stocking supplies, and perform more cognitive tasks, such as developing procedures, maintaining records, and designing computer applications. While many IWAs diminishing in prevalence are physical activities that might be automated by robots, others are more cognitive in nature, such as executing financial transactions, that are susceptible to automation through computerization.
These results suggest the pandemic has accelerated the key role that education will play in preparing the workers in the future. It is perhaps telling that the work activity with the highest projected increase is Maintaining current knowledge in area of expertise (Table 3). Thus, a key task of future workers will be the activity of continuously upkeeping skills and knowledge.
Education and learning activities
The shift in future labor attributes from physical to cognitive work is further reflected in the forecasted importance of college degrees. Of the nearly 2.9 million jobs expected to disappear under the strong COVID scenario, 91% are jobs that do not currently require a college degree (Table 4). Thus, job-seekers with a high school education or below will bear the brunt of lost opportunities under future COVID scenarios.
In terms of the quantity of college degrees in the workforce, COVID the largest decreases will be among business and related degrees while the largest increases are in computer-related degrees (Table 4). When aggregated to degree groups, the number of business-related degrees in the workforce will decrease by nearly 86,000, while computer-related degrees will increase by over 32,000 (Table 6). Other STEM degree groups, such as engineering, biological and physical sciences, and statistics also increase in number, despite there being almost 3 million less workers in the workforce.
Analyzing aggregate numbers of degrees gives only a partial picture of COVID’s impact on the post-secondary education landscape. It is also important to analyze the relative distribution of those degrees and examining the percentage of the workforce holding each type of degree offers another perspective of COVID’s impact on worker education. For instance, despite having the largest drops in aggregate numbers of degrees, business degrees nevertheless increase as a percentage of the workforce (Tables 5 and 7).
This occurs because, despite the decrease in quantities of certain degrees, the decrease in workers with no degree overwhelms all other categories. Thus, even though nearly 2.9 million jobs disappear under the strong COVID scenario, the percent of the workforce with a degree increases by over 4%, which positively impacts nearly every degree category. In fact, only three of 173 detailed degrees are projected to decrease as a percent of workforce–hospitality management, cosmetology and culinary arts, and drama and theater arts. And while computer science continues to show the largest increase in relative terms, degrees such as nursing and biology also increase significantly (Table 5). Aggregated to degree groups, only Personal and Culinary Service degrees decrease as a percentage of workforce.
It is important to note that the underlying data sources do not address other types of post-secondary education, such as 2-year degrees, high value certificates, microcredentials, and other worker training options. Thus, it is difficult to discern the importance of these credentials with the given data. However, it is likely that these non-traditional pathways for education, which were identified as critical to the future workforce [32], will only increase in importance as the need to continually maintain and update skills accelerates under COVID.
Industry socialness: Impacts on innovation and productivity
When ONET work activities are classified as either social or non-social and applied to employment projections, a measure of labor socialness can be calculated by industry. Social IWAs are defined as those likely to require face-to-face interaction between employees. Painter et al [27] recently demonstrated that the spatial density of social ONET elements is positively correlated with patenting output of cities and that the number of social IWAs per worker in an industry is superlinearly related to the industry’s per worker GDP. Thus, socialness is positively correlated with both innovation and productivity. Here I calculate industry socialness as the number of social IWAs per industry worker.
Results show that, among 2-digit industry sectors (Table 8), socialness increases most in Arts and Recreation (71). Socialness also increase substantially under the strong COVID scenario in the Finance (52), Information (51), and Professional/Scientific Services (54) sectors. Thus, it is plausible that the long-term impacts of COVID may include increased productivity and rates of innovation in these sectors.
Only three sectors exhibit a decrease in socialness under the strong COVID scenario. These are Agriculture (11), Construction (23), and Accommodation/Food Services (72), possibly reflecting a growing trend in mechanization and automation in these sectors.
This result indicates that the nature of work is likely to shift in ways that better facilitate workplace innovation. Thus, industry leaders and policy makers might seek to exploit this increased workforce socialness by implementing policies and strategies that foster innovation and reward creativity.
Trends in workforce automation and teleworking
Workforce automation, which has been ongoing for years, accelerated during the first months of the COVID pandemic [32]. This study suggests that this acceleration in automation may only be a short-term response and that COVID will slow the rate of automation on longer time scales. Among 2-digit sectors (Table 10), only construction (23) and transportation/warehouse (48–49) showed an increased probability of automation under the strong COVID scenario. Though the effect is negligible, it is unclear why this would be, and further research is warranted into the long-term impacts of COVID on trends in workforce automation.
A further long-term impact of COVID is an increase in nearly every industry of the amount of work performed remotely (Table 12). Among 2-digit industry sectors only construction (23) displayed a decrease in the fraction of jobs that can be performed remotely.
At the 4-digit industry level results are mixed and several industries do exhibit decreases in potential for telework. Yet, there is no immediately obvious trend among those that increase and decrease at the 4-digit level. For instance, charter buses (4855), other transportation activities (4889), and sightseeing transportation (4871, 4879) are all part of the same industry sector, yet charter buses and other transportation activities both decrease in teleworking potential while sightseeing transportation increases (Table 13).
Therefore, we can only conclude from this study that COVID will induce a broad increase in teleworking across industry sectors and across the workforce in aggregate. Still this may be informative for planning infrastructure investments for the future workforce, as this broad trend will likely increase the need for services such as broad band internet [33,34], while decreasing the demand for other services [35,36].
It should also be noted that teleworking changes the nature of worker interactions. Because those interactions are central to the previously discussed worker socialness that is linked to innovation and productivity, accelerated adoption of teleworking may mitigate projected impacts on workplace innovation. Further research is thus warranted to understand how interpersonal interactions and communication change in telework environments.
Occupational health and safety considerations
The COVID-induced shift away from physical activities to more cognitive activities in the future workforce may have significant implications for workplace health and safety. Because workplace injuries are often associated with physical tasks [37], it is reasonable to expect that workplace injuries will decrease under the strong COVID scenario. At the same time, reduced physical activity can induce new health issues related to a sedentary lifestyle, including mental health and stress-related illnesses [38]. Indeed, the ONET element “Spend Time Standing” decreases more than any other Work Context element, while “Spend Time Sitting” exhibits the second largest increase among Work Contexts. Thus, occupational health and safety professionals and policy makers should be aware of these potential shifts in workplace health risks.
At the same time, the anticipated increase in telecommuting means that more workers will perform tasks remotely in often isolated and unsupervised environments without typical workplace safeguards [39,40]. This creates additional health and safety considerations [41] as well as numerous legal and liability questions and thus should be a target for further research.
Caveats and opportunities
It is critical to understand that the conclusions of this study are based on long-term employment forecasts and, even though these forecasts are the product of sophisticated and well-vetted methodology, such forecasts are inherently uncertain. It is also important to realize that the changes in projected skills and education due to COVID are aggregate economic impacts and do not consider changes that may occur in the job requirements of individual occupations over the next 10 years. In other words, while this analysis uses the skills and knowledge requirements of today’s jobs those job requirements may change by 2029 –both because of long-term impacts of COVID and simply because of ongoing trends in technology, training, and innovation. Such changes in occupational requirements happen frequently and are especially rapid in STEM fields [42].
This issue applies equally to the potential for automation to radically change the activities and requirements of individual occupations. Factors used in this study for susceptibility to automation and potential for telecommuting apply to today’s jobs but will almost surely change in the future. For instance, advances in connectivity and changes in corporate culture will influence the prevalence of telecommuting while unforeseen advances in technology will likely accelerate automation. Both trends can also be significantly affected by policies, such as tax regimes or labor laws.
Given that such changes in occupational structure will occur in parallel to changes in demand for various occupations, this analysis may nevertheless help regional policy makers reprioritize planning for industry growth and educational provision. Overall, this analysis indicates that COVID will increase the need for post-secondary educational options. While this study points to increased demand in traditional degrees it is more likely that a suite of innovative options will be needed to prepare the workforce of the future for economies shaped by the long-term impacts of COVID.
Finally, readers should be aware of shortcomings of the underlying data and issues that may arise from linking disparate datasets. Consider the ONET data used in this study to assign skills, knowledge, and activities to individual occupations. The ONET data provides occupation specific values for attributes of the occupation “Accountant”. However, an accountant in New York City may have significantly different skills requirements than an accountant in a small rural village, while an accountant in a large financial consulting firm may have very different activities and knowledge requirements than an accountant in a small family-owned restaurant. Such differences are ignored as they are simply beyond the scope of such a dataset. Thus, geographical and economic context must be considered when using the results of this study to inform policy.
Acknowledgments
Lindsey Ice assisted with data acquisition and Teresa Farinha and Maricel Lawrence provided helpful feedback on this study.
References
- 1.
International Organization for Migration. World Migration Report 2015: Migrants and Cities: New Partnerships to Manage Policy. Geneva, Switzerland: International Organization for Migration; 2015.
- 2.
PCAST. Technology and the Future of Cities: Report to the President. Washington, DC: Executive Office of the President of the United States, President’s Council of Advisors on Science and Technology, 2016.
- 3. National Science Foundation U.S. 10 Big Ideas for Future NSF Investments: Report to Congress2016. Available from: https://www.nsf.gov/about/congress/reports/nsf_big_ideas.pdf.
- 4.
Buchan D. Energiewende–Germany’s gamble. Oxford, UK: Oxford Institute for Energy Studies; 2012.
- 5. Gawel E, Lehmann P, Korte K, Strunz S, Bovet J, Köck W, et al. The future of the energy transition in Germany. Energy, Sustainability and Society. 2014;4(1):15.
- 6. Abraham J. Just Transitions for the Miners: Labor Environmentalism in the Ruhr and Appalachian Coalfields. New Political Science. 2017;39(2):218–40.
- 7. Morton T, Müller K. Lusatia and the coal conundrum: The lived experience of the German Energiewende. Energy Policy. 2016;99:277–87.
- 8. Schwab K. The Fourth Industrial Revolution: What It Means and How to Respond. Foreign Affairs. 2015;Dec-2015:1–7.
- 9.
Manyika J, Lund S, Chui M, Bughin J, Woetzel J, Batra P, et al. Jobs Lost, Jobs Gained: Workforce Transitions in a Time of Automation. Brussells: McKinsey Global Institute; 2017 Dec-2017.
- 10.
Gownder JP. The Future of Jobs, 2027: Working Side By Side With Robots. Cambridge, Massachusetts: Forrester Research, Inc., 2017.
- 11.
PriceWaterhouseCoopers LLC. Sizing the prize: What’s the real value of AI for your business and how can you capitalise? London: Price Waterhouse Coopers LLC, 2017 170905-115740-GK-OS.
- 12.
U.S. Department of Defense. 2014 Quadrennial Defense Review Report, available at http://archive.defense.gov/pubs/2014_Quadrennial_Defense_Review.pdf. Washington, DC.: 2014.
- 13. Khamis M, Prinz D, Newhouse D, Palacios-Lopez A, Pape U. The Early Labor Market Impacts of COVID-19 in Developing Countries: Evidence from High-Frequency Phone Surveys. World Bank Group Open Knowledge Repository, 2021 Jobs Working Paper;No. 58.
- 14. Cortes GM, Forsythe E. The Heterogeneous Labor Market Impacts of the Covid-19 Pandemic. SSRN Electronic Journal. 2020;
- 15. Bayham J, Fenichel EP. Impact of school closures for COVID-19 on the US health-care workforce and net mortality: a modelling study. The Lancet Public Health. 2020;5(5):e271–e8. pmid:32251626
- 16. Groshen EL. COVID-19’s impact on the U.S. labor market as of September 2020. Business Economics. 2020;55(4):213–28. pmid:33204039
- 17. Larue B. Labor issues and COVID-19. Canadian Journal of Agricultural Economics/Revue canadienne d’agroeconomie. 2020;68(2):231–7.
- 18. Reichelt M, Makovi K, Sargsyan A. The impact of COVID-19 on gender inequality in the labor market and gender-role attitudes. European Societies. 2021;23(sup1):S228–S45.
- 19. Blundell R, Costa Dias M, Joyce R, Xu X. COVID-19 and Inequalities. Fiscal Studies. 2020;41(2):291–319. pmid:32836542
- 20. Abedi V, Olulana O, Avula V, Chaudhary D, Khan A, Shahjouei S, et al. Racial, Economic, and Health Inequality and COVID-19 Infection in the United States. Journal of Racial and Ethnic Health Disparities. 2021;8(3):732–42. pmid:32875535
- 21. Karako K, Song P, Chen Y, Tang W. Shifting workstyle to teleworking as a new normal in face of COVID-19: analysis with the model introducing intercity movement and behavioral pattern. Ann Transl Med. 2020;8(17):1056–. pmid:33145275.
- 22. Lee M, Zhao J, Sun Q, Pan Y, Zhou W, Xiong C, et al. Human mobility trends during the early stage of the COVID-19 pandemic in the United States. PLOS ONE. 2020;15(11):e0241468. pmid:33166301
- 23.
Ice L, Rieley MJ, Rinde S. Employment projections in a pandemic environment. Mon Labor Rev. 2021;U.S. Bureau of Labor Statistics, February 2021: https://doi.org/10.21916/mlr.2021.3
- 24.
US Bureau of Labor Statistics. National Employment Matrix 2019–2029 2020 [Dec 13, 2020]. Available from: https://www.bls.gov/emp/ind-occ-matrix/matrix.xlsx.
- 25.
US Bureau of Labor Statistics. Employment Projections: Pandemic Alternate Scenarios Projections 2021 [March 28, 2021]. Available from: https://www.bls.gov/emp/publications/pandemic-scenarios.htm.
- 26.
National Center for O*NET Development. O*NET database version 25.2. Available at: https://www.onetonline.org/. 2021.
- 27. Painter DT, Shutters ST, Wentz E. Innovations and Economic Output Scale with Social Interactions in the Workforce. Urban Science. 2021;5(1):21.
- 28.
US Census Bureau. American Community Survey (ACS), One-Year Public Use Microdata Sample (PUMS), 2005–2019: https://www.census.gov/programs-surveys/acs/microdata/access.html; 2020.
- 29. Frey CB, Osborne MA. The future of employment: How susceptible are jobs to computerisation? Technol Forecast Soc. 2017;114(2017):254–80.
- 30. Dingel JI, Neiman B. How many jobs can be done at home? National Bureau of Economic Research Working Paper Series. 2020;No. 26948. pmid:32834177
- 31. Dingel JI, Neiman B. Data for "How Many Jobs Can be Done at Home?". github. 2020;https://github.com/jdingel/DingelNeiman-workathome.
- 32.
World Economic Forum. The Future of Jobs Report 2020. Geneva, Switzerland: World Economic Forum; 2020 October 2020.
- 33. Joice W. Implementing Telework: The Technology Issue. Public Manager. 2007;36(2):64–8. pmid:236184284.
- 34. Kord H, Noushiravani Y, Bahadori MD, Jahantigh M. Review and Analysis of Telework Perspective in the Administrative Systems. Dutch Journal of Finance and Management. 2017;1(2):44.
- 35. O’Brien W, Yazdani Aliabadi F. Does telecommuting save energy? A critical review of quantitative studies and their research methods. Energy and Buildings. 2020;225:110298. pmid:32834440
- 36. Advantages Harpaz I. and disadvantages of telecommuting for the individual, organization and society. Work Study. 2002;51(2):74–80.
- 37. Oranye NO, Wallis B, Roer K, Archer-Heese G, Aguilar Z. Do Personal Factors or Types of Physical Tasks Predict Workplace Injury? Workplace Health & Safety. 2016;64(4):141–51. pmid:27056936.
- 38. Dėdelė A, Miškinytė A, Andrušaitytė S, Bartkutė Ž. Perceived Stress among Different Occupational Groups and the Interaction with Sedentary Behaviour. Int J Environ Res Public Health. 2019;16(23):4595. pmid:31756951.
- 39. Robertson MM, Maynard WS, McDevitt JR. Telecommunting: Managing the saftey of workers in home office environments. Prof Saf. 2003;48(4):30–26.
- 40. Healy ML. Telecommuting: Occupational health considerations for employee health and safety. AAOHN J. 2000;48(6):305–13. pmid:11249378
- 41. Wöhrmann AM, Ebner C. Understanding the bright side and the dark side of telework: An empirical analysis of working conditions and psychosomatic health complaints. New Technology, Work and Employment. 2021;Online First:.
- 42. Deming DJ, Noray KL. Stem careers and the changing skill requirements of work. National Bureau of Economic Research Working Paper Series. 2019;(No. 25065).