For a few lucky, talented and highly educated workers, the U.S. job market is booming. Artificial-intelligence researchers sometimes make more than $1 million a year. In the more general field of data science, salaries continue to rise despite a flood of new supply:

Meanwhile, salaries for electrical, mechanical and software engineers are rising as well, if a bit less spectacularly. For those with the skills to “tell computers what to do” (as venture capitalist and inventor Marc Andreessen once put it), capitalism still looks like a good deal. And there’s also the financial industry, which has long offered attractive salary premiums for highly talented people willing to endure the competitive culture and potential moral ambiguity.

But there’s a hidden downside to this high-end labor market. Many of these good jobs require Ph.D.s. A survey by Paysa found in 2017 that about 35 percent of AI jobs required a doctorate. In finance, Ph.D.s are heavily recruited for top quant trading jobs — as a professor at Stony Brook University, I helped advise applied-math doctoral students who were aiming for that industry. Plenty of workers at top tech companies such as Intel have Ph.D.s too. And more Ph.D. economists are going to work for industry. A quick Google search reveals a vast array of tech industry positions that now require this most advanced of degrees.

Why are so many companies asking for Ph.D.s? One reason might be that there are simply more Ph.D.s on the market. The number of doctorates awarded in the U.S. has increased in recent decades (though it fell a bit in 2017):

Meanwhile, tenure-track academic jobs — the kind of positions that doctoral programs groom people for — are on the wane:

So with more Ph.D.s looking for alternative careers in industry, why shouldn’t companies demand these degrees? Doctorate holders need money, tech and finance companies need expertise, so the market is matching the two.


Another reason might simply be the increasing need for both specialization and independent research in top technical jobs. Continuous innovation is the norm in knowledge industries, which compete by constantly offering new products. Ph.D.s teach students independent research skills. Also, the increasing technical complexity of the jobs might simply require Ph.D. levels of talent — corporate AI research, for example, is arguably as cutting-edge as anything in a university lab.

But the practice of requiring Ph.D.s for technical jobs could have some drawbacks. First, if the practice becomes an entrenched norm — that is, if companies start to assume that top jobs should go to Ph.D.s — it could create a segmented labor market, where qualified job candidates with only master’s or bachelor’s degrees would be overlooked and ignored.

If Ph.D.s become de rigueur for top tech jobs, it could also entice many more Americans to get Ph.D.s. But the program of study required to earn the degree isn’t optimized for sending people to industry. Because doctoral students work under professors, they get trained for the academic life. Academia is more independent than corporate research, especially for students trying to prove their research skills through single-authored papers. It’s also driven by different imperatives — an academic may choose to research an esoteric topic of interest, while a corporate research team tends to be driven by the demands of the market. The culture mismatch is so great that there is a whole industry devoted to helping academics transition to the private sector.

Doctoral programs can also take a heavy psychological toll on students. A recent paper in the journal Nature Biotechnology found that more than a third of doctoral students report symptoms of depression and anxiety — a rate about six times higher than the general public. About 40 percent experience severe symptoms. Other studies have found similar results. (Though it’s worth noting that the tech and finance industries have their own problems with stress and depression.)

Finally, Ph.D. programs come with a high opportunity cost. Grad students tend to spend six or seven of their peak years in school. During that time, they’re learning advanced topics and research skills, but they’re also spending time signaling their professorial abilities to potential academic employers. That signaling process could be wasted if they end up going into industry.

So there are reasons to think that the Ph.D. system is not ideal for producing the employees that U.S. industry needs. One possible solution is to offer Ph.D. tracks that guide students toward industry. This could involve having some advisers in the private sector, and doing more research at university-affiliated labs on or off campus. This research could be compensated by the companies, making the poverty of grad school less acute. Dissertations written by Ph.D.s aiming for industry could be team efforts instead of individual demonstrations of prowess. Industry-focused Ph.D.s might take less time and convey more certainty about students’ future careers.


But companies should also think twice about requiring Ph.D.s for research jobs. Master’s or even bachelor’s degree holders are often highly talented, and many can learn Ph.D.-level research skills on the job as they go. Employers should be careful not to overlook the vast pools of talent among those who lack the most polished credentials.


Noah Smith is a Bloomberg Opinion columnist. He was an assistant professor of finance at Stony Brook University, and he blogs at Noahpinion.