Can we retrain older workers?
You can, in fact, teach an old dog new tricks
Hello, and welcome back to Forked Lightning! If you’re just tuning in, last week I wrote exclusively about my recently released paper with Raj Chetty and John Friedman about selective college admissions. Starting today, this newsletter will come out weekly and will cover a range of issues related to education, skills, jobs, and technology.
In case you haven’t already, subscribe for free to keep up with my posts!
Most of you came here for a conversation about elite college admissions (the bait) but today I want to talk about job training for older workers (the switch). I promise to link them together eventually, just give me a few hundred words to explain.
If you care about higher education, you should care about job training too, because that’s what most Americans want from colleges when they hand over their hard-earned tuition money. The transformative power of a liberal arts and sciences education is a truly wonderful thing, but it’s also a luxury good. Most people have more earthly concerns, and rightly so. As the leaders of the Harvard Union of Clerical and Technical Workers (HUCTW) would say, “We Can’t Eat Prestige.”
Nearly all U.S. four-year colleges were founded before 1940, when less than 6 percent of people age 25-29 got a bachelor’s degree. As a result, they were not set up to be engines of upward mobility, but rather as finishing schools for the well-to-do. Now, two-thirds of high school graduates enroll in college, although not all of them finish. College today is less like a luxury brand and more like a mass market commodity. And yet the total number of public colleges and universities has increased by only 90 (from 1,497 to 1,587) since 1980.1 We are asking much more of American higher education than it was originally designed to deliver.2
The swiss cheese model of workforce development
This history helps explain why the U.S. does so little to help people transition from schooling to (and from) the labor market. We spend only 0.1 percent of GDP on Active Labor Market Policies (ALMPs), which includes programs like job search assistance, training, relocation for workers displaced by trade or technology, apprenticeships, and employment subsidies. The average among other OECD member countries is 0.5 percent of GDP, so we are spending in per capita terms only 20% of the average of other advanced economies. Despite have a strong education system and a strong economy, our pipeline from one to the other is incredibly underdeveloped.
This works well enough if you have the resources to navigate between education and work. Young people who attend selective colleges or who join a family business have no problem finding their way to opportunity. But many talented people slip through the cracks.3
There’s good recent evidence that we can do better – but it mostly applies to people early in their careers. Sectoral employment programs take youth without college degrees and train them for employment in growing industries like health care and information technology. They typically combine “soft” skills development with occupation-specific training, and most provide job search assistance as well as wraparound supports (bus passes, resume creation assistance, etc.).
A team of authors led by Larry Katz reviewed randomized controlled trial (RCT) evidence from four sectoral employment programs (SEIS, WorkAdvance, Year Up, and Project Quest) and found earnings gains between 14 and 38 percent that persisted for several years after the program was complete. That’s very exciting.4
Retraining after work accidents
The problem for older workers is different. We know much less about how to help them. That’s why I want to talk about an exciting paper I saw at the NBER Summer Institute (aka Nerd Summer Camp). The paper is about work accidents in Denmark.
What do work accidents in Denmark have to do with job training in the US? The paper, written by Anders Humlum, Jakob Munch, and Pernille Plato, studies what happens to workers who decide to go back to school after being seriously injured on the job. Danish administrative data are extraordinarily detailed, allowing the authors to track people as they move through the employment, education, health, and social services systems. Workplace injuries are surprisingly common, affecting about 0.6% of workers each year in Denmark and 2.3% in the U.S. They are most common in physically demanding jobs like carpentry, truck driving, and plumbing.
In Denmark, compensation for workplace injury comes as a one-time transfer (with an amount set by a board of industry specialists) and does not depend on subsequent employment or on the subsequent receipt of disability insurance. Injured workers can also receive disability insurance payments if they are unable to work. These benefits are set at the equivalent of about $2,700 per month and can be received through retirement age.
Here's where it gets interesting. As an alternative to disability payments, injured workers may elect to receive rehabilitation benefits to participate in formal education or vocational training with a firm. These benefits are also set at $2,700 per month. However, education in Denmark is completely free. In other words, injured workers who go back to school are paid a stipend equal to the amount they would receive if they took disability benefits instead.
Still, only about 13 percent of injured workers take up these generous benefits, even though most of them only have a high school degree.5 They almost always enroll in four-year degree programs rather than vocational degrees (our equivalent of a 2-year program) or shorter training courses, a fact that the authors use to their advantage. They notice that some vocational programs are “stackable” – meaning they lead to bachelor’s degree programs – while others are not. For example, a vocational degree in carpentry leads to a BA in construction architecture, while a vocational degree in landscape gardening has no upper-level equivalent. Other examples include electrician to service engineer and social-health assistant to social worker.
Injured workers do a lot better when they go back to school
Since the decision to go into a “stackable” vocation or not is probably unrelated to whether one suffers a workplace accident, they compare people who had a stackable BA option to those who did not, before and after the accident (this being an econ paper, they also do a thousand other robustness checks, but I’ll spare you the gory details).
Figure 5b shows the cumulative share of injured workers who go back to college compared to their non-injured counterparts (each line in the figure is the difference between workers in stackable and non-stackable fields). Being in a stackable field makes you about 10 percentage points more likely to enroll in higher education after a workplace accident.
Assuming that “stackability” is unrelated to getting injured, we can use this difference to study the impact of going back to school on earnings. That’s Figure 6a.
In years 5 through 10, earnings growth was higher for the people who were injured. You read that right. Injured workers who go back to school earn 24% more than they earned before their injuries. I hesitate to say that they are better off because they suffered a very serious injury. But given that the injury happened, they are definitely better off having gone back to school. Injured workers who do not go back to school mostly end up on disability benefits and are also more likely to be prescribed antidepressants.
Adding all these things together, the government’s investment in rehabilitation benefits for injured workers pays for itself four times over. This is because workers who go back to school do not receive as much in disability and other social assistance payments, and their higher earnings lead to higher tax revenue for the government. This calculation doesn’t even include the direct benefits to workers of higher earnings, or the mental health benefits implied by reductions in antidepressant prescription. Education benefits for injured workers reduce the net fiscal burden for taxpayers – they are the proverbial $100 bill lying on the sidewalk.
More people should go back to school, and we should pay them to do it
The proximate lesson is that we should immediately copy this program in the U.S., including increasing the generosity and duration of benefits for injured workers so that they view going back to school at least as favorably as going on disability.
More broadly, it tells me three important things about job training for older workers. First short-term vocational programs are generally not very appealing, and very few injured workers in Denmark took them up despite generous benefits. Jobs obtainable through short-run training are typically worse than the jobs that injured or displaced workers held previously, in which they have often accumulated specific expertise. The certainty of disability benefits may be preferable to having to start over in a lower-paying job. Ben Hyman studied Trade Adjustment Assistance (TAA) benefits in the U.S. for a similar population of older workers and found that the benefits were initially large but faded away quickly, which he attributes to the short-run nature of the programs that workers completed (TAA was almost always used for vocational training, not degrees).
Second, we should train vertically rather than horizontally. The injured workers in Denmark were able to keep some of their expertise. They transitioned from physically demanding jobs to cognitively demanding jobs, but in the same general field. This seems more promising than expecting coal miners to learn to code. There is some recent evidence on the value of stackable credentials in community college settings, but they top out at an associate’s degree. Personally, I would like to see colleges offer more stackable vocational-themed bachelor’s degrees, and this paper shows how it can work.
Third and most importantly: it’s never too late to go back to school. Education and training really can work, even for older people. But you must give them the time and space to truly reskill. In the U.S. we do it on the cheap. But this paper shows that low-dose retraining is insufficient to counteract a high-dose economic shock like injury or mass layoff. The authors find that workers aged 40-50 are especially likely to underestimate the benefits of education; only 6 percent go back to school, even though 37 percent would come out ahead doing so. Middle-aged workers usually have families and other obligations, so we must make retraining especially generous if we want it to work.
I take this paper as an important proof of concept that lifelong learning really is possible. It just gets harder with each passing year. As someone in the dreaded age 40-50 bracket myself, I can certainly relate.
While the number of public institutions has increased by only 90, the total number has increased by 700. About 70 of these are nonprofits, and the other 540 are for-profit colleges, which expanded dramatically in recent years. If you are interested in the topic of for-profit colleges, see this colorfully titled overview paper by me, Goldin and Katz (all glory goes to Claudia for the name!).
And yet, despite its many flaws, higher education is probably America’s most important global export. Depending on the origin country of the rankings, either 17 or 19 of the top 25 global universities are located in the U.S. Talented people still come from all around the world to be educated in America, and many of them want to stay here. The U.S. also has a very dynamic labor market, with the highest GDP per capita in the world and one of the lowest unemployment rates among advanced economies.
That’s why I started the Project on Workforce at Harvard, along with my colleagues Peter Blair, Joe Fuller, and Raffaella Sadun – to help create better pathways between education and the labor market. I’ll probably have more to say in subsequent posts about the work we do, but for now here’s the website if you’re interested.
Based on this evidence and on my broader reading of “what works” in the college-to-job transition, I think we should be asking college career services offices to copy some of the things that sectoral employment programs do well, including engaging more actively with employers. That might be hard to do, but not because we don’t know how. It’s a problem of institutional rigidity and political will.
Primary school goes through grade 9 in Denmark, then students split into a vocational degree track lasting 3-4 years or regular high school, which prepares you for a four-year college degree. Most injured workers completed the vocational track.