The Rise of the ROI and College Premium Mentality
An underlying driver of higher education policy and public sentiment that needs to avoid overly-simplistic conclusions
Phil Hill
June 03, 2025
Was this forwarded to you by a friend? Sign up, and get your own copy of the news that matters sent to your inbox every week. Sign up for the On EdTech newsletter. Interested in additional analysis? Try with our 30-day free trial and Upgrade to the On EdTech+ newsletter.
In yesterday’s Online Education Across the Atlantic podcast episode, we discussed the rise of the Return on Investment (ROI) and College Premium mindset that is increasingly dominating higher education policy in the US and the UK (also available on Apple Podcast and Spotify). We discussed the rise over the past 10 - 15 years of this economic and utilitarian view, some key issues driving it (with some overlap between the US and UK), whether the movement is a good development overall, and some unintended consequences we are seeing.
One of the points I made was that there is a difference between using ROI metrics for accountability for specific institutions and programs (e.g., whether a school should be responsible for student debt repayment rates) and broad evaluations of higher education sector performance. Morgan and Neil pointed out the need for both where I had been more critical of the former.
But on the latter issue of broad performance, there were two recent reports that highlight that common question of ‘is college worth it’ and for whom.
Is College Still Worth It?
The NY Fed has looked at this question for more than a decade, and in its broad modeling it calculates the college premium in investment terms. In last month’s update:
To weigh the upfront costs of college against the lifetime benefits, we calculate the internal rate of return—a measure investors commonly use to gauge the profitability of different investments. We follow the methodology used in our previous studies, with one important exception. Rather than using standard regression methods to estimate lifetime earnings profiles for the average graduate, we utilize a quantile regression method to estimate earnings profiles and the rate of return for the median graduate.
The overall conclusion is that yes, college is a sound investment.
After rising significantly in the 1980s and early 1990s, the return to college has held between 12 and 13 percent for the past three decades and was 12.5 percent in 2024—easily exceeding the threshold for a sound investment. Indeed, by comparison, the stock market has provided a long-term return of about 8 percent and bonds have returned around 4 percent.
There are problems with this model, however, which the NY Fed acknowledges. This analysis only looks at graduates and not those who fail to complete a degree, and that this aggregate conclusion is not true for all students. In a companion article, the NY Fed researchers again calculate that for close to 25% of US graduates, college is not a good investment when viewed solely in economic terms.
While our baseline estimates focus on the median college graduate, by definition, half of graduates are earning a lower return. Indeed, in the chart below we plot composition-adjusted wages for the 25th percentile of college wage earners compared to the median high school graduate over the past several decades. There is very little difference between the two groups, with an annual college wage premium of well under $10,000. Under our baseline cost scenario, we estimate a 2.6 percent rate of return for the 25th percentile of college graduates in 2024, making college a questionable investment for this group. As we’ve shown before, for at least a quarter of college graduates, college does not appear to pay off.
The article also points out that college major is a significant determinant of whether the investment is worth it.
As I mention in the podcast, the discussion of whether college is worth it becomes problematic when people just focus on the aggregate question. Despite the political rhetoric, there are very few people arguing against the concept of higher education. In a healthier discussion, instead of criticizing or defending higher education, we would be using these data as a feedback mechanism to support continuous improvement. 25% of graduates seeing little to no economic value is a real problem - we should be working to lower this number (but not to naively seek to eliminate it).
This month two researchers from Princeton and Vanderbilt looked at this question of the college premium for different students, with a focus on low-income to high-income demographics upon entering college. The broad conclusion is that over time higher education has become regressive.
The college-going premium was similarly large for lower- and higher-income students born in the 1900s, but by the end of the 20th century it had narrowed for lower-income students and widened for their higher-income peers. We call this new feature of American higher education the rise of “collegiate regressivity”: higher-income students now derive greater average observational value from going to college than the lowerincome students who make the same choice. This study documents when and why the labor market returns to college became positively correlated with childhood parental income.
The authors attempt to describe why this is happening at a policy level, in a fairly complicated mechanism. I question whether you can really use regression analysis to remove major factors to enable comparison over a 100+ period when higher education has completely changed (e.g., GI Bill and college-going rates, globalization, etc). But this regressivity also provides a useful feedback mechanism to support continuous improvement.
Avoiding the Aggregate
The move to measuring ROI and the college premium is not going away. There will be benefits to this emphasis and there will be unintended consequences. But I think we need to avoid broad discussions such as ‘is college worth it’ and instead focus on students - for whom is college worth it, and are we improving this service to students over time?
The main On EdTech newsletter is free to share in part or in whole. All we ask is attribution.
Thanks for being a subscriber.