College Major, Employment Status and Earnings

 

The relationship between college major and employment status has been getting a lot of attention. The Mellon Foundation claims that “around 95 percent of terminal humanities bachelor’s degree holders between the ages of 23 and 32 were fully employed” based on this study by the American Academy of Arts and Sciences. The New York Times suggests that computer science majors have to take jobs at Chipotle based on this study by the Federal Reserve Bank of New York. Numerous people on Bluesky have shared these articles, sometimes combining the claims about employment and unemployment with claims about higher earning potential for humanities majors, like historians than computer science and engineering majors.

 

The actual reports that these claims based on are worth looking at. Here are a few things to note.

1.       These are estimates of unemployment based upon  the American Community Survey (ACS) done by the Census Bureau. ACS is a one percent sample taken every year. Because it is a very large survey (they ask a lot of people a lot of questions) there is some delay in getting the data. The estimates from the NY Fed are are from the data collected in 2023, and the American Academy of Arts and Sciences study uses 2021 data.

2.       The estimates based on the ACS define unemployment in the same way as the Bureau of Labor Statistics, which produces the widely cited monthly estimates of unemployment based upon the Current Population Survey of 60,000 households. Someone is unemployed if they do not have a job and are actively seeking one. The labor force is composed of all the people who are employed and all the people who are unemployed. The unemployment rate is the percentage of the labor force that is unemployed.

3.       This definition of the unemployment rate means that you can’t just subtract the unemployment rate from 100 and declare that to be the unemployment rate as was done by the Mellon Foundation. Imagine you are looking at a labor force composed of a particular group of people, say people aged 22- 27 with a degree in history. For simplicity imagine there are 100 people, 95 are employed and 5 are unemployed. The unemployment rate is obviously 5 percent. But I can’t say anything about the employment rate for the entire population of history majors unless I also know how many history majors are not in the labor force, i.e., do not have a job and are not looking for one. If there are five of them then the percentage of the population that is employed would be 95 divided by 105 or 90 percent; if there are 10 of them the employment rate would be 95 divided by 110 or 86 percent. The Mellon Foundation's claim assumes a one hundred percent labor force participation rate.

4.       ACS and CPS use the same definition but collect data differently. BLS does a survey every month, ACS surveys are collected throughout the year. People who report unemployment to the CPS were all unemployed around the same time, people who reported unemployment to ACS were not necessarily unemployed at the same time. ACS estimates are higher than CPS estimates. For 2022, the ACS estimate of the U.S. unemployment rate was 4.3 percent while the CPS estimate was 3.6 percent.

5.       The NY Fed’s report also provides underemployment estimates for each major. Underemployment is defined as employment is a job that does not typically require a college degree. The NY Times article compared computer science and computer engineering that had unemployment rates of around 6 and 7 percent to biology and art history, which had rates of around 3 percent. But computer science and computer engineering had underemployment rates less than 20 percent while biology and art history had underemployment rates over 40 percent. The low underemployment in comp sci is consistent with a story in which some of the unemployment is voluntary in the sense that it results from people not taking the first job that comes along. The low underemployment rate is not consistent with a lot of computer science majors accepted jobs at Chipotle.

6.       Putting together points 3 and 5 suggests that the Mellon claim that “around 95 percent of terminal humanities bachelor’s degree holders between the ages of 23 and 32 were fully employed” is doubly misleading: you can’t tell from the unemployment rate how many are employed or how many of the employed are “fully employed.”

7.       The story that the NY Times wants to tell about over supply of computer science grads because of too much hype about it being the lucrative major compounded with AI replacing computer scientists may be true. But the evidence of it is not that clear in the data from 2023. Both the continued high starting salaries, about double those of art history and biology, and the low rates of underemployment are still pretty consistent with strong demand for computer science majors at that time. This would seem to be supported by the BLS estimates of occupational unemployment rates in July 2025 the estimated unemployment rate for computer and mathematical occupations was 2.9 percent.

8.       None of the cited studies support the claim that over the course of their careers history majors out earn computer scientists and engineers. The NY Fed study shows both early and mid-career salaries that are substantially higher for computer science and engineering majors than for history majors: mid-career median was $77,000 for history majors and $122,000 form computer engineering majors.

9.       None of this is to say that you should abandon history and study computer science. You should try to maximize your satisfaction (or utility in economic terms) not your monetary income. Yes, we get satisfaction from the stuff we buy with our income. Many people also get satisfaction from the prestige associated with a higher income. But many of us also get satisfaction from what we do in our job. 

10.   By the way, I didn’t have an undergraduate major. I went to The Evergreen State College, which does not have majors. I have a B.A. in Liberal Arts.