These comments were prepared for the inaugural conference on “Automation and the Middle Class” for the Brookings Institution, Future of the Middle Class Initiative. The original paper can be found here.
Paul Osterman gives us a clear, compact, and shrewd overview of the U.S. training system for mature adults. He does not focus on the implications or effects of automation per se. Instead he offers a timely assessment of the structure, strengths, and weaknesses of the current system.
His focus is appropriate, because the training issues arising from automation are roughly the same as those arising from any other source of job loss or occupational change. This includes disruptions caused by shifts in consumer tastes, company setbacks in competition with their rivals, and international trade as well as technological change. If the training system is equipped to deal with one kind of labor market disruption, it should be capable of dealing with the other ones, too.
Automation or the threat of it can give rise to demand for new training on the part of two kinds of worker. First are the workers who are displaced from their jobs as a result of employers’ adoption of new production technologies, including artificial intelligence and robots. These displaced workers are Paul’s “mature adults.” The second group of workers consists of new workers, the young adults who are just entering the workforce. They need to be trained in the skills required by tomorrow’s employers as well as today’s.
One can imagine creating two systems to address the skill needs of the two groups. As Paul points out, however, the U.S. system is so open and flexible—and, some might say, so mind-numbingly opaque—that exactly the same institutions and programs serve both populations.
In the present system, community colleges do much of the heavy lifting with respect to formal training for middle-skill occupations. Two-year colleges are equally prepared (or unprepared) to serve both new labor force entrants and skilled or semi-skilled workers who suffer displacement.
Aside from community colleges, the markets for both kinds of workers are also served by private, profit-making institutions that serve narrow niches of the training market, for example, beauty schools, health occupation prep schools, and truck driving schools. Some of the profit-making schools try to serve broader markets and to compete with public and private colleges and professional schools. Many of the private training institutions can serve both new labor force entrants and displaced workers. As Paul points out, many of them are not very successful in fulfilling this mission. They may be profitable to their owners, but many are demonstrably ineffective in lifting the earnings of trainees.
In an ideal world—though perhaps only in the fevered imagination of someone teaching undergraduate economics—both new labor force entrants and displaced workers are well-informed about the skill needs of employers that offer well-paid jobs. These workers are knowledgeable about their own work capacities and job preferences. They’re keenly aware of which institutions have the best records for placing graduates in occupations the workers wish to enter. Finally, workers also know which institutions have the greatest success in pushing their enrollees to completion of a degree or training certificate.
In the real world, of course, only a handful of people know much about the actual distribution of occupations, the skill requirements of individual occupations, or the likely future demand for workers in those occupations. Few people know how their capacities and preferences will line up with the skill requirements and day-to-day realities of life in relevant occupations. In the real world, few of us know much about the job-placement success of most training institutions (aside from Stanford, Cal Tech, and Harvard).
For the past 40 years, I’ve lived within 3 miles of the Northern Virginia Community College, also known as NOVA. Despite NOVA’s proximity, I know little about its success in getting entrants to graduate or in placing graduates in a good job or a decent 4-year college. I’m pretty sure local high school students and displaced workers know even less than I do. What local residents do know is that if a worker succeeds in getting an associate’s degree from NOVA, most employers will recognize that the worker has more education than is signified by a high school diploma but less schooling than indicated by bachelor’s degree. There are undoubtedly some local employers who are knowledgeable about the quality of NOVA’s training in specialized fields. I doubt whether that knowledge extends more than 30 or 40 miles beyond the NOVA campus, however.
A strength of the nation’s community colleges is that they are widely distributed around the country and are inexpensive from the perspective of enrollees, especially low-income students who qualify for federal aid. They can easily accommodate the schedules of students who cannot attend college full-time. Many 2-year colleges are also flexible in setting up specialized training programs aimed at the local employer community. A weakness is that, relative to 4-year public institutions, these institutions are underfunded. The combination of state and local institutional aid plus student fees and other sources of funding yields much lower spending per student in community colleges compared with public 4-year colleges. The U.S. Department of Education estimates that the instructional costs per full-time equivalent student in public 2-year institutions were about $6,300 in the 2015-16 school year. In public 4-year institutions, these costs averaged about $12,530 per full-time equivalent student, or twice as much. Spending on academic support services is also sharply lower in 2-year compared with 4-year public colleges. One effect of smaller instructional and support service budgets is the very low graduation and certification rates that Paul mentions in his survey.
The skill needs of new labor force entrants and displaced workers are somewhat distinct. Perhaps we shouldn’t rely as heavily as we do on the ability of displaced workers to negotiate a system that was originally geared toward the needs of new workforce entrants. At a minimum, we might want to create clearer markers in the system so displaced workers can negotiate the system more easily.
The prototypical displaced worker is someone who has held a decent, possibly well-paid, job for a number of years. He or she might lose the job because a plant has closed, a production line has been shut down, or part of an employer’s workforce is laid off as a result of slumping demand. Potential employers know something about this worker, and the displaced worker knows a lot about at least one niche in the labor market – namely, the last job that he or she held.
What do prospective employers know about the displaced worker? They know or can easily learn that the worker can hold a job for a reasonable span of years without getting fired. This implies the worker can perform a job at least adequately. This information is helpful compared with what employers know about new labor force entrants, who do not yet have a job record. In this respect, displaced workers have an advantage compared with new labor force entrants. However, prospective employers also know a displaced worker’s age. Labor economists recognize that the older the job applicant, the greater the odds the worker will face employer discrimination as a result of advancing age. The existence of age discrimination in hiring can make it risky for displaced workers to make big investments in their own retraining.
Let’s contrast the displaced worker with a new workforce entrant. New entrants have not yet demonstrated whether they can hold a job or acquire the specific skills needed to keep one. They probably know little about working, except what their older relatives and friends have passed along. Who knows whether a new entrant can acquire the skills and work habits needed to keep to a job? The nation’s education and training providers are supposed to help teach young people some of these things. Of course, we learn a great deal about work and occupational skill requirements on the job rather than in a classroom. But the American on-the-job training system, while highly flexible, is also decentralized and messy. In many occupations, the training system does not provide clear markers to inform interested job seekers about how they can acquire the credentials needed to enter an occupation or make their resumes more attractive to employers.
The German, Swiss, and Austrian systems offer much clearer guidance to inexperienced labor force entrants. For young people with the inclinations and aptitude to work in the professions, there is a university system, similar to our own. For youngsters without those preferences or aptitudes, there is the so-called “dual system,” which combines formal occupational instruction in classrooms and on-the-job apprenticeship in a workplace. The short-term workplace positions have low hourly pay, which makes the system attractive for many employers who are seeking a source of low-paid labor. The system seems well suited to the needs of non-college-bound youngsters who want to enter one of the hundreds of occupations in which apprenticeship training is available. Germany, Switzerland, and Austria traditionally have below-average youth unemployment rates, even when their economies are struggling. But the German-style system offers less attractive prospects to 45-year-old displaced workers who must change their occupations as a result of technological change.
One thing we know about the German system is that it devotes less real resources (per enrollee) to labor force entrants who pass through the country’s university system than does the equivalent system in the United States. Figure 1 shows educational spending per full-time-equivalent student in 19 OECD countries.
Average spending in a country is measured as the spending per full-time-equivalent student relative to the country’s GDP per capita. The OECD analysts calculate spending amounts at two levels of schooling, secondary schools (where spending is indicated by diamonds) and post-secondary institutions (where spending is indicated with bars). The United States ranks along with Japan as having the OECD’s most expensive post-secondary institutions. U.S. secondary schools, however, are somewhat less costly than the OECD average. In contrast to the U.S., Germany spends more per full-time-equivalent student on its secondary schools than on its post-secondary institutions. Among the 19 rich countries included in the chart, Germany ranks third from the bottom in its spending per enrollee on post-secondary institutions. The U.S.-Germany gap in spending on post-secondary institutions would appear even greater if we focused on average spending in the most selective post-secondary institutions in each country. (Note that U.S. post-secondary spending includes expenditures on students in community colleges, which as we have seen are far less expensive to educate than students enrolled in 4-year institutions.) If 4-year college and professional-school graduates seem exceptionally well paid in the United States, part of the explanation is the heavy investment made in their schooling. Some, though certainly not all, of the investment is made from public funds and tax preferences.
The flip side is that the German system devotes comparatively more resources to training young labor force entrants who participate in its “dual system” and on older workers who receive public subsidies for retraining. Figure 2 shows OECD estimates of total public spending on adult training in 18 rich countries.
The United States ranks third from the bottom in spending on adult training. Germany spends six times more than the U.S. on this activity. To be sure, part of the Germany-U.S. difference is traceable to the classification of U.S. community college subsidies as an educational rather than training expense. Note that the highest spending countries—Denmark and Finland—devote exceptionally large fractions of their public budgets to adult training and retraining.
How well does the US job training system work for the two classes of workers I’ve considered? Pretty well for new labor force entrants who begin and complete a 4-year college degree program. Economists have uncovered abundant evidence this group experiences below-average unemployment and above-average earnings profiles throughout their post-graduation careers. The lifetime income advantage of students who earn a post-college degree is even bigger, as is the advantage of graduates of the most selective (and richly resourced) public and private colleges. The words “Stanford,” “California Institute of Technology,” and “Harvard” on a resume can get job applicants a 2nd, 3rd, or even 10th look. Employers’ interest is sparked not only by the national reputations of these institutions, but also by the immense resources devoted to selecting, instructing, and providing counsel to the students fortunate enough to gain admission.
The U.S. training system works less well for youngsters in the bottom third of a typical 9th-grade class. Few of these students flourish in classroom settings, whether in a high school or community college. Only a minority receives much clear guidance on what they can do to improve their credentials after leaving school. Nonetheless, the overwhelming majority could learn useful job skills in a workplace setting, possibly supplemented with small doses of classroom instruction. The training opportunities open to this population after it leaves school are diverse, sometimes widely advertised (if offered by a for-profit firm), but almost never generously resourced. We do not lavish the resources on training the less skilled that we do on young adults who have the grades and persistence to gain admission to a selective public or private college.
The U.S. training system also works poorly for the displaced workers who are the focus of Paul’s paper. I agree with him that we can find many examples of excellence and first-rate practice in programs aimed at these workers. But the typical training opportunity tends to produce mediocre outcomes and subpar returns. Many of the training opportunities offered by for-profit companies fail to lead to course completion or an improved earnings trajectory. They frequently are financed with student loans the borrower cannot repay.
Paul suggests that “The American job training, or human capital development, system is complicated, hard to navigate, and under-funded.” Few observers who’ve studied the system will disagree. The islands of excellence Paul points to offer heartening evidence we can do better. But unless the country boosts the resources it invests in good adult training, it’s hard to be optimistic the system will improve.
 U.S. Department of Education, National Center for Education Statistics (2018), Digest of Education Statistics 2017, Table 334.10. “Total expenditures of public degree-granting postsecondary institutions, by purpose and level of institution: Selected years, 2009–10 through 2015–16.”
 Joanna N. Lahey (2008), “Age, Women, and Hiring: An Experimental Study,” Journal of Human Resources 43(1): 30-56; and David Neumark (2018), “Experimental Research on Labor Market Discrimination,” Journal of Economic Literature 56(3): 799-866.
 Adam Looney and Constatine Yannelis (2015),”A crisis in student loans? How changes in the characteristics of borrowers and in the institutions they attended contributed to rising loan defaults,” Brookings Papers on Economic Activity (Fall), 1–68.