The curious task of economics is to demonstrate to men how little they really know about what they imagine they can design.

F.A. Hayek, The Fatal Conceit (1988)

My previous critique examined the College Payoff report from the Georgetown University Center on Education and the Workforce (CEW) and found four methodological omissions that, taken together, overstated the financial return on a college degree by more than a million dollars. In 2019, the CEW published a response of sorts, “A First Try at ROI: Ranking 4,500 Colleges,” which was updated with new data in 2022.¹

The title is honest. It is a first try. The CEW fixed two of the four problems, partly fixed a third, and introduced new ones.

What the CEW Fixed

The 2022 ROI report is a genuine improvement over the College Payoff. The CEW now factors in the cost of a degree and attempts to account for the time value of money. Both were missing from the original headline figures.

The new report calculates a Net Present Value (NPV) of earnings after costs for each institution, which is the right method. In the 2022 table (2020 dollars), the median NPV across all institutions is $835,000 at a 40-year horizon and $140,000 at a 10-year horizon.

The CEW draws on the College Scorecard, as my book does, though on a different earnings series: earnings 10 years after enrollment, not four years after graduation. The approach to projecting synthetic lifetime earnings, assuming flat earnings beyond 10 years after enrollment, understates lifetime NPV, a choice the CEW says puts its estimates at the lower end of possible estimates, but the assumption is applied consistently across institutions. That is a reasonable simplification.

Real progress, then. The question is what remains.

What the CEW Still Gets Wrong

The first problem is inconsistent time periods. The College Payoff tracked earnings from age 25 to 64. The ROI report instead counts 40 years from enrollment, roughly ages 18 to 58 for a student who starts at 18.

Starting at enrollment rather than at age 25 is the right move. It counts the years spent in college, though the CEW does not subtract the earnings a student gives up while enrolled. But stopping at 58 instead of 65 truncates the comparison at the wrong end. In my experience, college graduates’ earnings often peak late in their careers, particularly in high-earning fields. Extending to 65 would increase the NPV for those fields. Because the CEW holds earnings flat after year 10, cutting those last years lowers the NPV of every institution’s 18-year-old entrants. The truncation understates lifetime value, most of all at the institutions whose graduates earn the most.

The second problem is flat costs. The CEW assumes tuition does not increase over the enrollment period.

Their own notes acknowledge this: “We also assume no increases in the cost of postsecondary investment, including tuition. This assumption matters less at long horizons but would overestimate the economic value in the short term if costs are actually increasing.”¹

Published tuition at four-year colleges has outpaced inflation by 90 to 150 percent since 1990, though the College Board’s 2024 report found that, after grant aid and inflation, the net tuition and fees paid by first-time, full-time students have fallen since the mid-2000s at private nonprofit four-year colleges and, for in-state students, since 2012-13 at public four-year colleges. Where costs do rise during enrollment, the assumption overstates short-term ROI for the students for whom the calculation matters most.

The third problem is taxes. The CEW calculates NPV using pre-tax earnings.

The federal tax system is progressive. In my book’s model, a high school graduate pays about $257,000 in federal income taxes over a lifetime. A college graduate pays about $483,600. The model’s earnings inputs differ from the CEW’s, but at the CEW’s own low discount rates, that federal income tax gap shrinks college’s NPV advantage by roughly 20 to 25 percent. Measuring returns before taxes is measuring something other than what a graduate actually keeps.

The fourth problem is the discount rate. The 2019 and 2022 rankings use 2 percent. That is below the 2.5 percent the 2011 College Payoff used in its appendix illustration, though that report’s headline figures were not discounted at all.² The 2021 edition used no discount rate either.³

Measured against that appendix rate, the discount rate moved in the wrong direction. The CEW explains the choice: they assume the investor is risk-averse and compare returns to a safe asset like Treasury securities, which, the CEW noted, then had a long-term return of about 2 percent. But a college education is not a Treasury security.

The benchmarks below are all well above the CEW’s 2 percent.

Parent PLUS loans carried 9.08 percent for the 2024-25 school year. New York Fed economists Jaison Abel and Richard Deitz used 5 percent in their college ROI work.⁴ San Francisco Fed researchers Mary Daly and Leila Bengali used 6.67 percent, the average AAA bond rate from 1990 to 2011.⁵ In its May 2023 baseline, the Congressional Budget Office projected undergraduate loan rates at 5.91 percent and PLUS loan rates at 8.46 percent.⁶

That range runs from 5 percent to about 9 percent. The CEW chose 2 percent. That choice makes college look like a safer investment than any rigorous comparison of capital costs would support.

A higher discount rate reduces the present value of future earnings more steeply, which favors lower-cost institutions and produces a more complete picture of risk-adjusted return.

The Benchmark Problem

The fifth problem is the most consequential.

When comparing colleges to each other, the obvious question is which college returns more than another college. The relevant one is which colleges return more than the alternative of not attending college at all. The CEW acknowledges this directly in its notes: they exclude from their rankings the benchmark of going directly to work after high school, because, they say, it varies from one student to the next.

The CEW illustrates that alternative only in its appendix. At $15,000 per year, close to the federal minimum wage, the 40-year NPV at their 2 percent discount rate is $397,000. At $10 an hour, it is $547,000.

But the CEW’s own College Payoff report calculated the lifetime earnings of a high school graduate as $1.6 million in undiscounted dollars in the 2021 report.³ The 2011 report also showed discounted values, at a 2.5 percent rate.² Its appendix arithmetic starts from an average-based $2.79 million bachelor’s figure: a discounted bachelor’s NPV of $1,712,000 minus a discounted high school-to-bachelor’s gap of $786,000 leaves a high school NPV of about $926,000, in 2009 dollars. Apply the report’s own 39 percent discount to its median high school lifetime earnings of $1,304,000 instead, and the high school NPV is about $795,000.

These are not compatible figures. The CEW’s own 2011 figures imply a high school baseline of about $795,000 to $926,000. In the ROI ranking report, the high school figures are footnote illustrations of $397,000 and $547,000, well under that range. Against those lower figures, more colleges would look like good investments. The baseline implied by the CEW’s own prior work tells a different story. The relevant benchmark is not the minimum wage floor. It is what a student could realistically earn by going directly into the workforce instead of attending college. The CEW calculated a version of that number in their own prior research, for a steady full-time, full-year career. Then they left it out of the rankings.

The high school baseline implied by the CEW’s own figures, applied to its 2019 ROI rankings, tells a specific story. The online table covers 4,529 institutions, with 40-year NPVs ranging from $240,000 to $2,722,000.⁷ At the $926,000 appendix baseline, only 874 (19%) produce an NPV at or above it, and 81 percent fall below. At the $795,000 median-based baseline, 1,700 (38%) are at or above it, and 62 percent fall below. The table’s NPVs rest on median earnings, so the median-based $795,000 is the closer match of the two. Either way, most institutions in the table fall below the high school baseline implied by the CEW’s own figures. Among the 1,761 that mainly award bachelor’s degrees, 62 percent fall below the $926,000 baseline and 35 percent fall below the $795,000 one.

Figure B-A · Colleges with a 40-year NPV rank of 875 of 4,529, the first below the $926k HS NPV threshold (2019 table) (the left-hand row counter is the table’s display order)

Source · Georgetown CEW ROI rankings · Author’s NPV threshold analysis.

Those figures come from the CEW’s own data, set against the CEW’s own prior research on high school graduate earnings. The two studies measure different people: full-time workers from 25 to 64 in one, federally aided students counted from enrollment in the other. Converting the baseline from 2009 dollars and a 2.5 percent rate to the table’s 2017 dollars and 2 percent rate would raise it, while its full-time, full-year earnings basis cuts the other way, as does the $926,000 figure’s use of an average where the table uses medians, and so does its start at age 25, which counts the years from 58 to 64 in place of the lower-paid years just after high school. The table’s own shortcut also cuts the other way: it holds each college’s earnings flat after year 10, while the 2011 baseline adds up the earnings observed at each age from 25 to 64, and earnings tend to rise with age before they level off. Treat it as a rough test, not a precise count. Even so, the result sits uneasily with the CEW’s view that its findings support the idea that college is a worthwhile investment.

Where the CEW Went Next

The 2022 report was at least attempting to engage with the time value of money, however imperfectly. The CEW’s 2025 ROI update abandoned the effort entirely.⁸

Their explanation: “By doing away with the discount rate, we are attempting to eliminate subjectivity by treating future earnings the same as current earnings.”⁸

That rationale aims to eliminate subjectivity. What it eliminates is basic financial analysis.

Treating a dollar earned in 40 years as equal to a dollar earned today is not a neutral choice. It is the choice that produces the largest possible return figure for college. College graduates’ earnings are backloaded. They accumulate over decades. A zero discount rate maximizes the apparent value of those distant earnings.

Researchers at the New York and San Francisco Federal Reserve banks used discount rates above zero, as standard investment analysis does, and the Congressional Budget Office projected federal student loan rates well above it. The CEW, whose 2 percent rate sits below every benchmark above, has moved to 0 percent.

The methodology has not improved.

It has regressed.

Each choice can be defended on its own terms. Four of the five in the table below tilt the result in the same direction. The methodology produced the conclusion the industry has long believed.

The CEW fixed two methodological problems in 2019, partly fixed a third, introduced new ones, and carried the fixes and the new problems into its 2022 update. In 2025 it undid one of the fixes by dropping the discount rate. Most of the choices that biased the result pointed in the same direction.

Methodology choice	CEW reports, across versions	My book’s treatment	Direction of bias on result
Time period analyzed	Ages 25–64 in the College Payoff (2011, 2021). Shifted to 40 years from enrollment, roughly 18–58, in the ROI Rankings (2019, 2022) and ROI Update (2025).	Ages 18–65, working life through standard retirement.	For a student who starts at 18, stopping at 58 drops the last seven working years. Understates NPV, most in high-earning fields.
Cost inflation assumption	No costs in the College Payoff’s lifetime figures. Flat costs in the ROI Rankings and ROI Update. The CEW’s own notes acknowledge this would overstate short-term ROI if costs rise.	Published tuition has outpaced inflation by 90–150% since 1990 (net tuition after grant aid has fallen since the mid-2000s at private nonprofit and since 2012-13 at public four-year colleges). Costs treated as rising.	Where costs rise, flat costs overstate ROI for the short-horizon students for whom the calculation matters most.
Tax treatment	Pre-tax earnings in the College Payoff, the ROI Rankings and the ROI Update.	After-tax earnings.	At the CEW’s low discount rates, federal income taxes alone make the NPV advantage roughly 20–25% smaller.
Discount rate	College Payoff (2011): headline undiscounted, with a 2.5% illustration in the appendix. None in 2021. 2% in ROI Rankings (2019, 2022). 0% in ROI Update (2025).	7.8%, inside the 5% to 9% range of credible government, market, and Federal Reserve benchmarks.	Each step lower inflates the present value of future earnings further. The 2025 rate maximizes the apparent return.
High school baseline benchmark	About $795,000 to $926,000 NPV implied by the 2011 College Payoff’s own figures, in 2009 dollars. The ROI Rankings set no high school NPV benchmark. The 2022 update adds only the share of students who earn more than a high school graduate 10 years after enrollment. The rankings’ footnote illustrations are low-wage streams worth $397,000 to $547,000.	The CEW’s own 2011 figures as a consistency test. My book’s model puts the high school NPV at $484,376, after federal taxes at a 7.8% rate, so it is not comparable to the pre-tax 2% figures.	Measured against the footnote illustrations instead of the CEW’s own earlier figures, far more colleges look like good investments. Against the baseline implied by the CEW’s own figures, 62% to 81% of institutions in the 2019 table fall below (35% to 62% of bachelor’s institutions).

Source · CEW ROI Rankings¹ · CEW College Payoff^2 3 · CEW ROI Update⁸ · My book’s NPV model in Chapters 29 and 31

What This Means for the Rankings

The CEW’s ROI rankings rest on real data. College Scorecard earnings and institution-specific costs are sound. The methodology converting that data into a national ranking is where the difficulty enters. The discount rate, the benchmark, and the framing each push in one direction.

The rankings are not useless. The underlying earnings data, drawn from the College Scorecard, is real and institution-specific. A student comparing two schools using the raw earnings and cost data available at collegescorecard.ed.gov is doing something genuinely valuable.

The problem is the aggregate ranking, the headline NPV, and the conclusion that college is a worthwhile investment, reached by choosing benchmarks and discount rates that make it easier to reach.

The underlying data is available. The methodology for putting it to use is in my book. The CEW’s rankings are not the right tool. The data behind them is.

---

About this analysis

A version of this critique appears as Appendix B (Georgetown, ROI Rankings) in We Need To Talk About Higher Education by Leon Shivamber.

Get the book →   Read the argument in full →   Run your own numbers →

---

Notes

1. Carnevale, Anthony P., Cheah, Ban, & Van Der Werf, Martin. (2019). Georgetown University Center on Education and the Workforce, “A First Try at ROI: Ranking 4,500 Colleges.” And Georgetown University Center on Education and the Workforce. “Ranking 4,500 Colleges by ROI (2022).”
2. Carnevale, Anthony P., Rose, Stephen J., & Cheah, Ban. (2011). “The College Payoff: Education, Occupations, Lifetime Earnings.” Georgetown University Center on Education and the Workforce, 2011
3. Carnevale, Anthony P., Cheah, Ban, and Wenzinger, Emma. “The College Payoff: More Education Doesn’t Always Mean More Earnings.” Georgetown University Center on Education and the Workforce, 2021.
4. Abel, Jaison R., and Deitz, Richard. “The Value of a College Degree.” Liberty Street Economics. Federal Reserve Bank of New York, September 2, 2014.
5. Daly, Mary C., and Leila Bengali. 2014. “Is It Still Worth Going to College?” FRBSF Economic Letter 2014-13, May 5.
6. Congressional Budget Office. “Baseline Projections: Federal Student Loan Programs.” May 2023. cbo.gov. Accessed March 2026.
7. Georgetown University Center on Education and the Workforce. “A First Try at ROI: Ranking 4,500 Colleges” (2019), online table of 4,529 institutions ranked by 40-year NPV, in 2017 dollars.
8. Cheah, Ban, Van Der Werf, Martin, Morris, Catherine, and Strohl, Jeff. “Ranking 4,600 Colleges (2025) by ROI.” Georgetown University Center on Education and the Workforce, 2025.

Related critiques

This is one of four close readings of the major college ROI studies. See the others, then read the audit that weighs all four together and the plain-language case beneath them.

• Georgetown’s College Payoff, examined.
• IHEP’s Rising Above the Threshold, examined.
• The Fed’s college ROI math, examined.
• All four studies, audited together: Auditing the Numbers That Say College Pays.
• The plain-language version: College Is a Bet.
• The same problem priced against the alternative: Compared to What?
• The full argument: Is College Worth It?