The college earnings gap isn't one number

How the wage premium varies across 64 college majors

It’s no secret that workers with a bachelor’s degree earn substantially more over their careers than workers with just a high school diploma. According to the latest available data, the average gap is around $2.14M¹.

But we also know that this gap differs quite a bit by field of study. For example, a worker with a degree in chemical engineering earns an average of $6.68M from age 25 to 64, $4.04M higher than the average high school graduate.

On the other hand, a worker with a degree in early childhood education tends to earn slightly less than a worker with just a high school diploma.

This is the only one of the 64 college majors where this is the case. The top ten college majors in terms of career earnings include economics and finance, as well as several engineering fields. The bottom includes education and social services majors.

The linked interactive page lets you look up any of the 64 college majors in the data to view their respective earnings gaps and other metrics in comparison to the average career of a high school diploma-holding worker.

It is not just earnings where college graduates differ from high school-level workers. College graduates also have lower unemployment rates across their careers compared to high school graduates at similar ages.

And even as this earnings gap closes at older ages, it masks a more persistent gap in labor force participation. Because college graduates are more likely to participate in the labor market, they have a roughly 13-percentage point lead on high school-level workers in overall employment rate across prime working age.

Finally, the data also challenges the common trope of college graduates being stuck in jobs that don’t require a college degree. There certainly are cases where this is true. But overall, two thirds of college graduates are in an occupation that requires college-level preparation.

And what about the on third that aren’t? They still tend to earn considerably more than their high school-level coworkers.

The majors that most commonly result in employment in high-preparation jobs include nursing and several engineering fields. The ones that tend to lead to jobs that don’t require much preparation are criminal justice and fire protection and hospitality management.

Do these gaps reflect a degree’s return on investment?

The topic of education and related employment outcomes gets a lot of attention these days, so it’s important to understand what these gaps in earnings and other metrics represent.

First of all, these earnings numbers do not represent returns on investment. A proper ROI estimate for an educational program requires far more information than I’m providing here. The first question is one of perspective: an ROI can reflect the private return to the individual student, the fiscal return on the government’s investment in education programs, or the social return, which adds in benefits that accrue to the surrounding society.

On the returns side, it would need:

• The wage increase caused by the educational program. The raw earnings gap is polluted by selection bias. Students who complete college have preexisting differences in ability from students who do not. Likewise, students who earn education degrees are likely different from students who major in engineering. Also, if the perspective is the social benefits, the analysis would need to net out any signaling benefits from the returns.

• The proper comparison group is not the average or the median college graduate, but the marginal one. Students who would complete college (or not) regardless of additional investments are not relevant for the ROI calculation.

• Personal earnings do not represent the total potential benefits of an education program. College students likely receive substantial non-wage benefits from schooling. These might include non-work-related skills that improve other life outcomes, such as happiness, health and longevity. It is also likely that the occupations available to college graduates, especially some of the low-earning ones, include other job amenities. This might involve fringe benefits like insurance and retirement accounts. But it can also include perks that are harder to measure, like higher job satisfaction and perceived meaningfulness of work, remote work, flexible scheduling and safer working conditions.

• There are also substantial benefits of college education that accrue to society, rather than to the graduates themselves. These can include monetary benefits like higher tax receipts and reduced welfare assistance due to increased productivity and earnings, which would be relevant for the fiscal ROI calculation. But a social ROI would also count increases in civic participation, higher political stability and commitment to democracy, and lower crime. There is also the fact that many of the lower-paying jobs for college graduates, such as childcare workers and teachers, are themselves socially beneficial and are often underfunded relative to what would be socially optimal. Measuring the value of external benefits is difficult and often subjective, but ignoring them implies that they are all zero, which is certainly incorrect.

• Finally, the benefits would need to be discounted according to how far in the future they occur. This can be difficult since it is largely based on subjective preferences about the timing of benefits.

On the investment side the calculation must have:

• The direct costs of college attendance, such as tuition, fees and books. Lodging and food should be excluded here since they would need to be covered regardless of college attendance. Subsidies of these costs would need to be counted for the fiscal or social ROI calculation.

• The opportunity cost of attending college. This includes the lost earnings during school attendance, and also the longer-term costs of delaying the start of one’s career.

• The risk of not completing school. Students who don’t graduate tend to have much poorer outcomes than those who do, and it is often high cost programs that have lower completion rates.

Ultimately, some of the most important factors in determining the ROI of a program will be external/social benefits, cost of attendance and risk of not completing. Few attempts to measure the ROI of college successfully account for all of these.

One of the most popular and comprehensive attempts to calculate ROI leaves out non-wage benefits and social externalities. While this is understandable given the subjectivity of these factors, leaving them out effectively assumes a value of zero, which is certainly incorrect. These estimates are useful, but they are incomplete in a way that matters a lot.

Less rigorous attempts to quantify the value of college degrees include the U.S. Department of Education‘s recently announced Do No Harm earnings test, which will determine eligibility for federal student loan programs. The metric used in this rule, early career earnings relative to those of high school-level workers, doesn’t reflect the effect of the marginal student, accounts for only wage benefits of education and doesn’t account for the costs.

1. The analyses presented here use American Community Survey data to construct a synthetic work-life earnings estimate similar to Julian and Kominski 2011. This creates hypothetical average people who live every year from age 25 to 64 during the current time period (2020-2024). Earnings numbers are calculated after restricting the population to full-time, year-round workers.

Some might question the inclusion of 2020 and 2021, since they were likely affected by the pandemic recession. But I tested excluding them for these analyses and the results were basically unchanged. The main difference in the statistics is that the unemployment rates, 3.2% for college graduates vs. 5.6% for high school graduates, dropped to 2.6% and 4.6% in the 2022-2024 sample. The main practical outcome of restricting the years would be lower sample sizes and a large reduction in the number of college majors with usable results.