January 31, 2018
Recently, labor economists, workforce development policy analysts, and workforce development practitioners gathered at the Federal Reserve Bank of Atlanta to examine the effects of automation and shifting labor demands on the future of work. Presenters included Dan Restuccia of Burning Glass Technologies, Sara Lamback of Jobs for the Future, Daniel Kreisman of Georgia State University, Chad Shearer of the Brookings Institution, Susan Lund of McKinsey Global Institute, and Nancey Green Leigh and Ben Kraft of Georgia Institute of Technology. The researchers collected data from job postings, résumés, and various public sources to evaluate the shifting nature of career pathways, particularly for middle-skill workers, and explore how to assess changing labor markets with new data sources. Several themes stood out from the research and subsequent discussion.
Effects of automation on the labor market
The degree to which job automation affects the labor market will vary across occupational sector, skill level, and geography. According to a study by McKinsey Global Institute, although less than 5 percent of jobs are completely automatable, around two-thirds contain at least 30 percent technically automatable tasks. Predictable physical tasks, data collection, and data processing have higher potential for automation than activities such as solving problems creatively, making decisions, or interacting with and managing people. Workers who carry out technically automatable activities are concentrated in the United States, India, China, and Japan, but the timing and degree to which jobs are replaced with technologies differ across countries. Economic factors contributing to the pace of automation include technical feasibility, costs and benefits, labor market dynamics, regulation, and social acceptance, says McKinsey's Lund. See the McKinsey report Harnessing Automation for a Workforce That Works for further information.
Occupations requiring less education tend to have greater potential for automation. Specifically, two-thirds of jobs requiring less than a high school diploma and half of jobs requiring a high school diploma are automatable, compared to only one-fifth of jobs requiring a bachelor's degree or higher, says Lund. This has clear consequences for workforce development investment. Consider that 41 percent of the U.S. workforce has a high school diploma or less, according to American Community Survey 5-Year Estimates for 2012 to 2016, and only 46 percent of U.S. students who start college earn a degree, according to Georgia State's Kreisman, referring to research by Hansson and Charbonnier. At the same time, some of the low- and middle-skill jobs with good automation potential are currently in high demand. For instance, job trend data suggest demand for warehouse workers has increased since the recession, particularly in core metro areas, according to initial research by Nancey Green Leigh, Ben Kraft, and Heon Yeong Lee. O*NET data indicate that warehouse workers and similar logistics occupations will grow by 5 to 9 percent between 2016 and 2026, on par with the national growth average.
The dual nature of high demand and highly automatable occupations demonstrates the complexity of issues surrounding the future of work and workforce policy. In addition, while automation may widen the gap between high- and low-skill workers, it will likely create new types of jobs and boost global gross domestic product growth as demonstrated by the advent of the personal computer, says Lund. These intricacies leave room for rethinking how the public, private, and nonprofit sectors invest in and train workers for less automatable activities based on up-to-date labor market data.
Careers often move laterally, not upward
Today's career pathways look different from the traditional upward mobility ladder. Careers often take the form of lattices where workers move not only upward, but also laterally into new occupations. More than 60 percent of workers switch jobs within five years of beginning a job, and more than 40 percent of workers change occupation families completely, according to Burning Glass Technologies' Restuccia. Temporary economic recessions as well as long-term shifts in technology exacerbate these transitions. Often workers and workforce development organizations have a very difficult time navigating these complex career pathways.
Researchers at Burning Glass used job posting data to map occupations with overlapping skills across different career sectors. Jobs with high enough similarity scores could offer feasible next-step occupations for individuals looking to make lateral rather than upward transitions. For example, researchers found skill overlap between retail sales associates and personal bankers. Comparing occupations in this way also pinpoints skill gaps across sectors, so that individuals understand what training they need before shifting career paths. Retail sales associates with diminishing job options in the retail field could use this comparison data to identify education and training programs for personal banking (or other jobs with similar skills) and successfully make a lateral career move.
Workers who change occupations tend to have higher earnings mobility (that is, wage increases), while earnings mobility has fallen over time for workers who remain in the same occupation, says the Brookings Institution's Shearer. Restuccia says only 8 percent of workers are in higher-paying occupations within the same career area five years after starting a job. Interestingly, however, occupational transitions have declined since 2000, particularly for workers with less education. To promote upward mobility of low- to middle-skill workers, Brookings Institution researchers built a model using Current Population Survey data and occupational job openings by county to project future job opening rates. Information about locational shifts in demand reveals potential career pathways, allowing researchers to map how workers can transition to different occupations and achieve higher earnings.
Upward mobility depends also on geographic mobility—the ability to move elsewhere for a new job—since the number of living wage jobs available to low- and middle-skill workers varies greatly by geographic region. In Atlanta, for instance, 30 percent of jobs in 2015 will move low- to middle-skill workers from jobs that pay less than a living wage (defined here as $15 per hour) to those that pay a living wage by 2027, says Shearer of Brookings. Other regions around the country will offer more or fewer opportunities for earnings mobility (within the same or a new occupation), and so workers' capacity to move physically to a better-paying job greatly affects their potential career pathways. Workforce policy should account for the geographical mobility of individuals, which varies due to local, state, and federal policies, current economic conditions, and factors such as age, income, housing status, family size, and social networks.
Job stability and advancement
Middle-skill occupations and associated skill credentials offer varying opportunities for job stability and career advancement. In a recent study funded by the Lumina Foundation, analysts from Jobs for the Future and Burning Glass Technologies looked at résumés of 15 million middle-skill workers, defined as having jobs requiring less than a bachelor's degree and with wages of at least $15 per hour. Of the four targeted career areas—business, health care, information technology (IT), and production—workers in business and IT sectors had higher rates of career stability and advancement over a five-year period. While health care workers tended to remain within the same career, fewer workers advanced (defined as earning a median salary at least 10 percent higher than with the starting occupation).
Education and training for these middle-skill jobs also offer varying opportunities. Credentials like the American Welding Society certification prepare workers for entry-level positions, but they typically are not associated with career advancement. Others, such as the Professional in Human Resources certification, provide the same potential for advancement as would two years of college, says Jobs for the Future's Lamback.
Career advancement is also associated with soft skills development not always gained through traditional credentialing and certification programs. Management, problem solving, and idea generation skills correspond to career advancement and increased wages, according to Lamback and Shearer. Workforce strategies should reflect the importance of developing soft skills just as much as earning official credentials.
Data do not tell the whole story
Longitudinal studies of job posting and résumé data provide new insights for career pathways, but they should be used with caution. Workforce development policymakers and practitioners can use long-term résumé data to track students' careers and wages before and after school/training programs and show which jobs lead to career advancement. Job posting data could also improve alignment between such programs and the needs of local employers. As researchers point out, these data sets can be used as complements to existing public census data, and like anything, they come with limitations. Both job postings and résumés are nonstandard and relatively more subjective data sources. While job postings indicate how job qualifications shift over time, these changes are not necessarily linear. Qualification levels deemed necessary at one time may fluctuate with business cycles and decrease during times of low unemployment. In addition, job postings in certain industries are less likely to appear online (for example, many unionized occupations, some lower-skill occupations, and positions in the skilled trades are not traditionally recruited online and thus, may not have as reliable information). Up-to-date numbers of job openings for some occupations, therefore, may be lacking.1
Information ascertained from résumé data also depends completely on what skills, experiences, and education individuals choose to highlight. Some may leave off seemingly irrelevant or negative information, and therefore create a bias toward higher job stability. For instance, Georgia State's Kreisman says individuals who dropped out of college tend to omit partial schooling from their résumés. On the other hand, people who remain in one job for many years are less likely to have publicly available résumés, creating a data bias toward workers with less job stability. Further, since résumés do not include wage information, researchers can only measure "career advancement" based on job title change, excluding some number of workers who experience wage increases but maintain the same job position.
The seminar stimulated insightful conversation around current job trends and shifts in the labor market. Data analyses on the future of workforce demand from organizations like McKinsey Global Institute, Burning Glass Technologies, Jobs for the Future, the Brookings Institution, and colleges or universities can inform workforce education and training decisions and shape career pathways toward the best quality job opportunities—both now and in the future.
Note: This event was planned and hosted by Stuart Andreason, director of the new Center for Workforce and Economic Opportunity at the Atlanta Fed, in collaboration with Keith Wardrip, community development research manager at the Philadelphia Fed. For more information, contact Stuart Andreason or Ashley Bozarth.
Ashley Bozarth is a research analyst in the Center for Workforce and Economic Opportunity.
1 For a longer discussion of online job posting data, see Wardrip et al. (2015), Identifying Opportunity Occupations in the Nation's Largest Metropolitan Economies and Wardrip et al. (2017), Uneven Opportunity: Exploring Employers' Educational Preferences for Middle-Skills Jobs.