The 2025 One Big Beautiful Bill Act raises the federal tax on private university endowment income sharply, with the top rate of 8% falling on schools holding more than $2 million in assets per student. Harvard, MIT, Princeton, Stanford, and Yale face this maximum rate. Coming on top of cuts to federal research funding and a wave of civil-rights litigation, the tax will reduce operating income at some of America’s most prestigious universities. Responding to financial pressure, Harvard has moved to admit 50% fewer PhD students in the sciences (Mao and Paulus 2025), MIT graduate admissions for the autumn of 2026 were down by 20% (Ryan 2026), and the University of Chicago has paused or trimmed admissions to many of its programmes (Glick and Eisenstadt, 2025). Will enrollment cuts at schools that train the most successful economics researchers handicap scientific progress?

Graduates of eight elite US schools make up over half of full-time economics faculty employed by top departments (Amir and Knauff 2008). These elite eight schools include the five most highly taxed plus Berkeley, Chicago, and Northwestern. ‘Elite eight’ graduates publish more influential research than do graduates of other US and foreign programmes. This elite advantage may result from superior training at elite doctoral programmes. But not necessarily.

The research impact of elite PhDs may be an artifact of prestigious schools’ ability to attract the most promising young scholars. After all, elite economics programmes are highly selective. The challenge of how to measure causal value-added in graduate education parallels that addressed in Dale and Krueger’s (2002, 2014) influential studies of selective undergraduate institutions. As in these studies, our empirical strategy identifies causal school effects by exploiting uniquely informative controls for PhD applicants’ ability and motivation for advanced doctoral work.

Our analysis exploits a novel dataset covering the roughly 200 highest-rated applicants to the MIT Economics PhD programme in selected years between 1996 and 2019. For each of these applicants, we observe the ratings assigned by the MIT Economics admissions committee. These numerical scores, averaged across faculty readers and converted to a within-cohort rank, strongly predict long-run research success. For some applicants, we also see the full list of programmes to which applicants applied. We link publication data to applicant data using a matching workflow similar to that used in our earlier study of Economics PhD advisors and students (Angrist and Diederichs 2024, 2025). Our primary measure of research output is an index that weights each publication by the influence of the journal it appears in, giving heavy weight to the field’s most-cited outlets and fractional credit to coauthored work. We also track narrower and broader outcomes, namely, the count of articles in the ‘top five’ journals, which carry outsized weight in hiring and tenure decisions (Heckman and Moktan 2018); an unweighted count of publications in 137 leading journals; a count of papers in any journal indexed by EconLit; and an indicator for having published at all.

Our analysis starts by comparing the research success of graduates with degrees from the five high-tax schools, graduates from the remaining three elites, graduates of other US schools, and foreign degree holders. Even within our select pool of MIT Economics PhD finalists, research output varies widely across groups. Figure 1 shows the distribution of impact-weighted publications and top five articles for elite graduates, alongside statistics for graduates of other US schools and foreign PhDs. Roughly 80% publish of the high-tax group at least one paper in a high-impact journal, a statistic that falls to around 70% for other elite graduates and 60% for foreign graduates. Gaps in top five success are even wider. Over 40% of high-tax graduates publish at least once in a top five journal, and around 25% land three or more top fives. Other elite graduates see somewhat less success, while the contrast in top five success is largest between elite graduates and the rest.

Figure 1 Publication distribution by school type: Impact-weighted and top five output

Notes: The bars show the distribution of publications for graduates of high-tax schools (MIT, Harvard, Stanford, Princeton, and Yale), other elite schools (Chicago, Northwestern, and Berkeley), other US schools, foreign schools, and applicants who did not graduate. The left panel (WDI) shows co-authorship- and journal-influence-weighted “deep impact” publications, grouped into 0, (0, 1], (1, 2], and 2+; the right panel shows top-five journal counts, grouped into 0, 1, 2, and 3+. The sample is limited to 1996–2010 MIT application cohorts. 
Source: Angrist et al. (2026).

MIT admissions ratings are a powerful predictor of research success. As can be seen in Figure 2, almost everyone in the highest-rank group publishes at least one high-impact paper, and nearly two-thirds place a paper in a top five journal. The publication gradient flattens for lower-ranked applicants. At the same time, many lower-ranked applicants see considerable success. This reflects the fact that research outcomes have a large idiosyncratic component that even MIT’s admissions committee cannot foresee. Importantly for our purposes, Figure 3 shows substantial overlap in the distribution of school types conditional on rank.

Figure 2 Publication distribution by MIT admissions rank: Impact-weighted and top five output

Notes: The bars show the distribution of publications across six MIT admissions-rank groups (1–5, 6–25, 26–50, 51–100, 101–150, 151–200). The left panel (WDI) shows coauthorship- and journal-influence-weighted “deep impact” publications, grouped into 0, (0, 1], (1, 2], and 2+; the right panel shows top-five journal counts, grouped into 0, 1, 2, and 3+. The sample is limited to 1996–2010 MIT application cohorts. 
Source: Angrist et al. (2026).

Figure 3 Graduate school attended by admissions rank

Notes: This figure charts the distribution of graduate school attendance across six admissions rank groups (1–5, 6–25, 26–50, 51–100, 101–150, 151–200). MIT is separated from other high-tax schools (Harvard, Stanford, Princeton, and Yale). Other elite eight schools are Chicago, Northwestern, and Berkeley. Remaining school groups are: other US, foreign, and did not graduate. The sample includes application cohorts through 2019.

To distinguish causal value-added from selection effects, we adapt a framework long used to evaluate K–12 schools and college quality (as, for instance, in Angrist et al. 2024 and Dale and Krueger 2002). This framework asks how much more an applicant who attends Northwestern, say, would publish were he or she to attend MIT. Our study identifies such difference in potential outcomes by assuming that, conditional on MIT admissions rank and a few other key covariates, school attended is independent of potential outcomes. This identification strategy is implement using an exponential model that controls for functions of admissions rank, NSF and MIT award status, and (in a subsample) the list of each applicant’s target schools. Our identification strategy is validated by showing that, conditional on the reciprocal of MIT admissions rank, other controls matter little for estimates of elite value-added.

Inclusion of rank controls shrinks raw elite–non-elite publication gaps considerably, suggesting that selection accounts for part of the elite advantage. Figure 4 summarises school-by-school estimates. But the elite-school premium remains large when estimated with admissions and other key controls. Graduates of high-tax and other ‘elite eight’ schools produce 60–75% more impact-weighted publications than do otherwise comparable graduates of non-elite US schools, with the advantage especially pronounced for top five articles. Differences within the elite eight, by contrast, are modestly modest and rarely statistically significant. Finally, in the spirit of Dale and Krueger (2002), the portfolio of programmes each applicant applied to offers a revealed measure of applicant ambition. In support of a causal reading of our value-added estimates, the addition of target-school dummies leaves value-add estimates like those in Figure 4 unchanged.

Figure 4 School value-added by selectivity

Notes: This figure plots estimates of individual-school value-added for impact-weighted (WDI) publications, with other US schools as the reference group. Blue dots show estimates without controls for applicant ability; red diamonds show estimates that control for MIT admissions rank and related measures of applicant promise. Schools are ordered from most to least selective (by the average MIT admissions rank of their graduates), shown by the gray bars (right axis, reversed). Vertical lines are 95% confidence intervals. The sample includes application cohorts through 2019. 
Source: Angrist et al. (2026).

We weigh two explanations of elite PhD research success that, if true, weigh against research productivity concerns raised by the endowment tax. If the elite advantage arises merely by attracting clusters of talented students, tax effects can be mitigated by redirecting elite students to untaxed schools. In this scenario, elite value-added is a peer effect. We find little support for this view: research outcomes are uncorrelated with MIT applicants’ classmate quality.

We also ask whether the elite premium can be dismissed as home-journal favouritism, that is, an ‘inside job’. Harvard and MIT graduates often place articles in the Quarterly Journal of Economics, which is edited at Harvard. Editorial connections like those between Harvard and the QJE or between the University of Chicago and the Journal of Political Economy may well matter for the publication success of these schools’ PhDs (Colussi 2018, Diederichs, 2026). But the estimated elite advantage remains large and statistically significant even after stripping out the journals most tied to particular departments.

What does this mean for the endowment tax? Our estimates suggest that the relative research success of elite-school PhDs is, to a substantial degree, a causal effect rather than something these schools merely inherit. This effect cannot be attributed to the marshalling of high-quality peers or editorial favouritism at department-affiliated journals. Tax-induced reductions in graduate enrollment are therefore likely to handicap the economics research enterprise. This finding might be mirrored in other disciplines as well. We note, however, that a vast endowment is not the only route to high value-added. Graduates of Berkeley and Northwestern, which have smaller per-student endowments, publish almost as well as PhD graduates of the highly taxed elites.

References

Amir, R and M Knauff (2008), “Ranking economics departments worldwide on the basis of PhD placement”, Review of Economics and Statistics 90(1): 185–190.

Angrist, J and M Diederichs (2024), “Dissertation paths: Advisors and students in the economics research production function”, NBER Working Paper 33281. 

Angrist, J and M Diederichs (2025), “Prolific PhD advisors are no guarantee of graduate student research success”, VoxEU.org, 16 February. 

Angrist, J, M Diederichs and G Ellison (2026), “Will the university endowment tax slow scientific progress? Evidence from elite economics PhD programs”, NBER Working Paper 35244.

Angrist, J, P Hull, P A Pathak and C Walters (2024), “Credible school value-added with undersubscribed school lotteries”, Review of Economics and Statistics 106(1): 1–19.

Colussi, T (2018), “Social ties in academia: A friend is a treasure”, Review of Economics and Statistics 100(1): 45–50.

Dale, S B and A B Krueger (2002), “Estimating the payoff to attending a more selective college: An application of selection on observables and unobservables”, Quarterly Journal of Economics 117(4): 1491–1527.

Dale, S B and A B Krueger (2014), “Estimating the effects of college characteristics over the career using administrative earnings data”, Journal of Human Resources 49(2): 323–358.

Diederichs, M (2026), “Effects of editorial connections on publications and citations in economics”, Working paper, University of Passau.

Glick, I and E Eisenstadt (2025), “UChicago to cut some Ph.D., master’s admissions for 2026–27”, Chicago Maroon, 2025. 

Heckman, J J and S Moktan (2018), “Publishing and promotion in economics: The tyranny of the Top Five”, VoxEU.org, 1 November. 

Mao, W C and V H Paulus (2025), “Harvard to admit 50% fewer Ph.D. students in science, walking back deeper cuts”, The Harvard Crimson, 19 November. 

Ryan, A (2026), “‘A loss for the nation’: MIT president says research has fallen 10 percent, and grad student enrollment is down”, The Boston Globe, 14 May.