Does going to college make economic sense? How much does the answer to that question depend on the college major you choose?
A new study by Temple University economist Douglas Webber calculates the lifetime earnings premium accrued by college graduates in various subject areas, relative to the earnings of high school graduates with no college attendance. Webber corrects for “selection bias”—that is, for the fact that people who are intrinsically (due to any number of factors) more likely to earn more are also more likely to go to college. The conclusion: Even when selection bias is corrected for, college still makes economic sense, and the benefits are larger in some areas of study than in others. (STEM majors did especially well.) Webber provided Science Careers with an exclusive glimpse at data that fill out the big picture, including lifetime earnings by specific major. It’s a fascinating analysis with many interesting and surprising results, so read on.
Over their lifetimes, graduates with majors in science, technology, engineering, and mathematics (STEM) can expect to earn far more than high school graduates with no college attendance, with an earnings premium of $1.5 million over and above the $1.73 million that high school graduates with no college attendance can expect to earn.
The big picture. Over their lifetimes, graduates with majors in science, technology, engineering, and mathematics (STEM) can expect to earn far more than high school graduates with no college attendance, with an earnings premium of $1.5 million over and above the $1.73 million that high school graduates with no college attendance can expect to earn. Business majors do slightly worse than STEM majors, with a lifetime earnings premium of $1.4 million. Social scientists stake out the middle ground, earning $1.05 million more than noncollege high school graduates over a lifetime. Arts and humanities majors can expect to earn about $700,000 more, on average, than high school graduates with no college attendance.
The worst STEM majors earn more than the best high school graduates. Those in the bottom quintile of ability who go on to major in STEM have lifetime earnings of about $2.3 million, compared to $2 million for high school graduates in the top quintile of ability; business majors do slightly worse than STEM majors. The worst social science majors earn about the same as the best high school graduates, and the worst arts and humanities majors earn less.
Selection bias is real, but the earnings advantage persists. Webber ran the numbers both ways: with and without correcting for selection. Without correction, the simulation showed that STEM majors could expect an even larger lifetime earnings premium: $2.2 million more than high school graduates with no college attendance, instead of $1.5 million.
STEM majors' lifetime earnings haven’t improved much over time. Webber studied three separate cohorts, born in the decades following 1955, 1965, and 1975, respectively. He found that earnings have improved considerably for most college majors—but the gains in STEM fields are surprisingly small. For example, the most recent cohort of arts and humanities majors can expect to earn in excess of 50% more than the earliest (1955) cohort. For social science majors, the increase in lifetime earnings is about 40%. The latest cohort of STEM majors, though, will earn just 14% more than the 1955 cohort of STEM majors.
There’s an ability premium. Webber’s selection correction says, in effect, that those with the propensity and ability to major in STEM can expect to earn a premium over other high school graduates, even if they don’t go to college. It follows that those who have a propensity to go to college and to major in a STEM field—and then actually do so—could expect to earn the uncorrected lifetime earnings premium. (I wasn’t sure about this interpretation, so I sent Webber an e-mail. “I think you are spot on,” he replied.)
The ability premium (also known as the selection-bias correction) has declined over time. College isn’t as exclusive as it once was; a larger cross-section of the population is attending. This shows up in Webber’s analysis in the size of the selection-bias correction and how it changes between cohorts. In the earliest cohort of STEM majors, for example, the ability premium was about $568,000. In the most recent cohort, the ability premium had fallen to about $317,000.
The decline in the ability premium over time means that the college premium—the actual value of a college education—has grown more over time than is apparent at first glance. Using the corrected values for STEM in Table 8, we see that the college premium for STEM majors has increased by about 27% from the first cohort to the last—a significantly larger increase than the 14% reported above in the growth of total lifetime earnings.
Lifetime earnings vary by major field. For statistical reasons, Webber’s study reported lifetime earnings only by major category—STEM, business, social sciences, and arts/humanities—and reported results only for men. (See below for an explanation of why Webber’s analysis excludes women.*) Science Careers, however, obtained data from Webber on average lifetime earnings by specific major; these data include both men and women:
|Total earnings||Total Earnings, corrected|
|All STEM graduates||$3.05||$2.75|
|All Non-STEM graduates||$2.66||$2.36|
|High School only||$1.27||$1.27|
Caveats first: Sample sizes are now smaller, so these estimates are less precise than those for the broader fields: STEM, social sciences, and so on. Also, the correction factor used for specific majors was calculated for all STEM fields together (not field by field). Finally, these numbers—like all the numbers in this article—do not factor in the costs of college: tuition, servicing student loans, and so on. Webber presents those numbers (by major category) in his Table 4.
Webber calculated these earnings over the course of a working lifetime, from 18 to 64 years; that’s 46 years or 552 months. So, for example, a person with a bachelor’s degree in physics could expect to earn (including the correction) $3170 a month, on average, more than someone with just a high school degree, over the course of a career.
Women earn less. The numbers in the table above, for all STEM graduates and for high school graduates, are much smaller than the numbers that include men (e.g., about $3.5 million for men in STEM fields, compared to $3.05 million in all fields). We have no information about how much of this difference is attributable to discrimination and how much to fertility-related working-pattern differences.
Biology earnings are weak; engineering earnings are strong. Engineering is the most lucrative field, followed by physics—then computer science and chemistry. The least lucrative STEM field by far is biology, which has lifetime earnings significantly below those of the average non-STEM college graduate. These differences are not small; over the course of a career, a graduate with a physics degree can expect to earn an average of about $1532 a month more than a graduate with a biology degree.
These results do not account for subsequent graduate training. To get into graduate school, you need to earn an appropriate undergraduate degree first. For those who take advantage of it, that’s an added benefit of an undergraduate degree that isn't measured here; indeed, Webber excluded from his sample people with postgraduate training.
There’s a lot that’s interesting here, but the big-picture result is that even in times when it can seem difficult to get a job in most fields, a college STEM degree remains valuable. The closest thing to an exception is biology, where a degree still results in a substantial lifetime-earnings premium, but one that’s less than in some non-STEM fields.
* Why were women excluded? Using only men in the study is, Webber writes in the article, "consistent with many labor market studies." However, "[t]his is a particularly important restriction for this study given the relatively weaker labor force attachment of women in the 1979 cohort of the [National Longitudinal Survey of Youth], and the drastic differences in major choice among women (e.g. STEM fields) relative to today." Webber elaborated in an e-mail: "Sample size limitations prevent me from doing things separately by gender. Putting them together but not dealing with issues such as discrimination and fertility will just give me the wrong answer for everyone. Thus, I look only at men so that I can get the 'right answer' for at least some subgroup of the population. This is much more important though for some of the in-depth questions in the review process, and much less important for the top-line numbers that I am giving to you."