Achieving greater socioeconomic diversity at highly endowed colleges and universities

Elite private colleges and universities educate a small slice of American college students, but their large endowments create opportunities to provide an affordable, high-quality education to lower-income students. Historically, though, they have provided this opportunity to relatively few such students.

This report tracks recent trends in the enrollment of Pell Grant recipients at these institutions. Pell Grant eligibility is based mostly on the financial situation of a student’s parents. In 2019-2020, those with family incomes below roughly $75,000 (in 2023 dollars) would qualify. I analyze trends in the Pell share of enrollment between 2007-08 and 2022-23 for two groups of elite private institutions: those with the highest endowments—above $500,000 per full-time equivalent (FTE) student—and those with slightly smaller endowments but still above $250,000 per FTE.¹ Going forward, the two categories are labeled as having “very large” and “large” endowments, respectively. Institutions in the first category are subject to the current endowment tax, and the latter may be at risk should the scope of that tax expand. In previous research, I show that these institutions have the financial resources to provide an excellent educational experience for their students and increase college access at the same time.

In this analysis, I find that representation of Pell Grant recipients increased by around 20% at institutions with the highest endowment levels over the last decade, though such students remain substantially underrepresented. Changing economic conditions and changes in the formula used to determine Pell Grant eligibility were not meaningful contributors to this trend in the last decade, though they likely explained earlier increases.² Test score data suggest that elite colleges and universities did not lower their admissions standards to increase enrollment of Pell Grant recipients. Instead, lower-income students appear to be more apt to apply and enroll if admitted to these institutions now. Greater outreach to high achieving, lower-income students by institutions likely contributed to that trend. This is a substantial increase in the representation of Pell-eligible students at these institutions, but there is still considerable room for further improvement. 

Pell Grant receipt roughly tracks socioeconomic diversity and may change over time for several reasons

To accurately track enrollment patterns of lower-income students, one would need data on family income for enrolled students for the institutions of interest. Pell grant receipt is an imperfect proxy for lower-income students. Eligibility for a Pell Grant is, indeed, partially based on lower-income status, but other factors play a role, including family composition and asset holdings. More importantly, the formula used to calculate eligibility has changed over time. This could change the measured representation of Pell-eligible students on campus without any change in socioeconomic diversity.

Tracking Pell Grant receipt does have significant advantages, though. The Pell share is calculated by institutions using their administrative databases. It is readily available from the federal government since 2007-08, and they are reasonably up-to-date, now extending through 2022-23. I rely on those data to measure trends in socioeconomic diversity at highly endowed institutions.

There are multiple factors that could lead to changes in an institution’s Pell share, including:

• Changes in eligibility: The number of students who qualify for Pell Grants can change over time if there are trends in family economic circumstances. Economic conditions can generate such trends. Eligibility will also change if the formula determining Pell Grant receipt changes. When the formula becomes more generous, more students will be eligible.
• Changing admissions standards for lower-income students: Colleges and universities may adopt a form of affirmative action based on socio-economic status. The presence and extent of such policies may change over time.
• Changing student application behavior: Students cannot attend an institution where they did not apply. If lower-income students become more aware of the opportunities available at elite private colleges and universities and apply at greater rates, this may increase enrollment. Institutions can facilitate awareness by increasing their recruitment efforts, encouraging applications from highly qualified lower-income students. There might also be changes in the likelihood that admitted, lower-income students enroll when they are admitted, which is likely affected by the financial aid offered.

The remainder of this analysis will focus on distinguishing between these competing explanations for the recent increase in representation in Pell Grant recipients at elite institutions. I can assess the validity of changes in eligibility and admissions standards directly given the available data. It is difficult to directly assess changes in application and enrollment behavior, but its likely importance increases to the extent the alternatives can be ruled out.

The Pell share at elite institutions jumped after the Great Recession because eligibility increased

Figure 1 shows trends in the percentage of students who are Pell-eligible at institutions with very large and large endowments. The data used to track these trends is available from the Integrated Postsecondary Education Data System (IPEDS). At the beginning of this period, Pell shares were almost identical in the two categories of institutions. Both similarly jumped between 2008-2009 and 2010-2011. Increases in this period are likely attributable to changes in eligibility. Economic conditions certainly contributed to this jump. The Great Recession reduced family incomes, enabling more students to qualify for a Pell Grant.

Changes in the formula used to determine eligibility also contributed to the increase in this early period. Pell Grant eligibility is based on a student’s Expected Family Contribution (EFC—recently renamed Student Aid Index, or SAI), which is calculated when they complete the FAFSA form. It is an approximation of the amount a student can afford to pay to attend college. A student is eligible for a Pell Grant equal to the difference between the maximum grant level and EFC. Currently, the maximum Pell Grant is $7,395; a student with an EFC of $3,000 would qualify for a Pell Grant of $4,395. Students with a $0 EFC are eligible for the maximum Pell Grant. Those with an EFC above the maximum award are not eligible for a Pell Grant.

During this period, the value of the maximum Pell Grant increased considerably, as shown in Figure 2. In inflation-adjusted terms (2023 dollars), the maximum grant rose from $6,355 to $7,780 between 2007-2008 and 2010-2011; much of that increase took place in the latter two years of this period. Mechanically, the higher the maximum award, the more people will qualify. This large increase in grant receipt also occurred nationally.

More recently, the Pell share rose modestly because lower-income students were more likely to apply and enroll if admitted

Over the past decade, the Pell share at these institutions continued to rise at a slower rate. It increased from 15.6% in 2014-15 to 18.5% in 2022-23 at institutions with very large endowments (a 2.9 percentage point increase, or 19% from the base of 15.6%). At the next endowment tier, the rate ticked up from 15.6% to 16.8% (a 1.2 percentage point increase, or 8% from the base of 15.6%). Across higher education more broadly, the share of students who received Pell Grants actually fell, although from a higher baseline rate.

Throughout this period, the size of the maximum inflation-adjusted Pell Grant was reasonably stable. Economic conditions fluctuated little prior to the start of the pandemic in 2020. The ensuing recession was brief, albeit severe, and it was combined with a massive infusion of governmental aid to families that offset the financial losses, limiting the impact on Pell Grant receipt. Overall, external conditions likely had little impact on Pell Grant receipt over this period.

If changes in Pell eligibility cannot explain the increasing Pell share since 2014-15, perhaps these institutions relaxed admissions standards to enroll more low-income students. Criteria for admissions are not easy to quantify, but changes in test scores of enrolled students can provide an indication of whether admissions standards were changing.

To examine this issue, I examine trends in SAT scores by Pell Grant receipt among students enrolled at institutions with large and very large endowments.³ I use student-level data from four waves (2007-2008, 2011-2012, 2015-2016, and 2019-2020) of the National Postsecondary Student Aid Study (NPSAS). See the Data Appendix for more details.

The results of this analysis are presented in Figure 3. I aggregate the data to two periods to increase sample sizes, combining 2008-09 with 2011-12 and 2015-16 with 2019-20. At institutions with very large endowments, average SAT scores for Pell Grant recipients were 72 points lower in the earlier period. This finding is consistent with recent evidence documenting that highly endowed institutions often consider the difficulties that these students have in achieving higher scores in admissions decisions. They are considered strong candidates even with somewhat lower test scores. In the latter two surveys, that gap fell to 58 points (although the reduction in the gap is not statistically significant). The gap narrowed even further at institutions with large endowments (that change is statistically significant). The test score data suggest that neither group of institutions lowered their admissions standards for Pell Grant recipients over this period.

Changes in economic conditions and the Pell Grant formula do not explain rising Pell representation at highly endowed private colleges, nor do falling admissions standards. This points to an increase in applications from lower-income students as a leading explanation. The timing of this increase is correlated with the release of influential research identifying the existence of lower-income students with strong academic credentials who were not attending highly selective institutions (“undermatching”). Efforts to help overcome the barriers reducing such access may have increased around that time.

Some of those efforts were unrelated to the college recruitment activities of specific colleges. For instance, nonprofit organizations like College Advising Corps and uAspire that focus on college access significantly expanded around this time. Such interventions may have contributed to the trend.

These elite colleges and universities also pursued their own initiatives to help attract and enroll more lower-income students. For example, they expanded their own financial aid programs, reducing their net prices. They also partnered with nonprofit organizations, like QuestBridge and Posse, designed to help identify and recruit lower-income students. Many elite institutions now include socioeconomic diversity as a stated goal of their enrollment policies. Although pinpointing the exact contribution of such interventions is difficult, it seems likely that they helped, at least somewhat, in explaining the increase in enrollment of Pell Grant recipients.

Discussion

The results of this analysis suggest that colleges and universities with very large endowments have achieved a modest increase in enrollment of Pell Grant recipients. Over the past decade, mechanical changes in eligibility or changes in student selectivity do not appear to explain this trend. It seems that highly qualified, Pell-eligible students apply to these institutions at a higher rate, increasing their enrollment. Greater recruitment efforts on the part of these institutions along with increased financial aid likely contributed to that outcome.

There are potential reasons to question this interpretation, though. For instance, some analysts have demonstrated that some institutions have increased their enrollment of those students just below the Pell Grant eligibility threshold at the expense of students just above it. Such behavior would be consistent with the findings of this analysis, but the social benefit associated with increased enrollment of Pell recipients in this manner would be small.

Such activity is unlikely to explain enrollment patterns, particularly at institutions with endowments this large. Most of them have “need-blind” admissions policies. Applicants are evaluated without regard for financial status. An admissions officer may be able to determine an applicant’s broad socioeconomic status based on their background but not with the precision necessary to engage in this form of strategic behavior.

As I have argued elsewhere, the modest progress in the enrollment of Pell-eligible students that has taken place at highly endowed institutions needs to continue. There are many barriers that limit the likelihood that lower-income students will be academically qualified to enroll in these highly selective colleges. But those who can succeed in that environment deserve the opportunity to ascend the economic ladder. The role of these colleges and universities in promoting social mobility needs to be balanced with their goals of academic excellence. Even institutions with very large endowments have budget constraints. But tilting the scale perhaps a bit more towards social mobility would likely benefit society and the institutions themselves.

Appendix

Data Appendix

The two data sources used in this analysis are Integrated Postsecondary Education Data System (IPEDs) and restricted-use, individual-level data from four waves (2007-2008, 2011-2012, 2015-2016, and 2019-2020) of the National Postsecondary Student Aid Study (NPSAS). From IPEDS, I use the share of first-year, full-time undergraduates receiving Pell Grants, total undergraduate enrollment, and the 25^th and 75^th percentiles of the institution-wide distribution of SAT scores. Pell Grant receipt is available in these data beginning in 2007-08 and currently extend through 2022-2023. Data for the other measures extends further back in time.

The NPSAS data include information on tens of thousands of enrolled students nationwide in each wave. They provide critical components that are not available in other sources: access to students’ financial information, financial aid receipt, and academic background for students enrolled in college. The institutions those students attend are individually identified (with restricted use access).

There are important limitations, though, in using these NPSAS data for this analysis related to representativeness and sample size. For the full data set, sample weights are available to obtain nationally representative statistics, but there is no guarantee that those weights will adequately represent the student bodies at the highly endowed institutions that are the focus of this study.

Those institutions also represent a sliver of total college enrollment. Although each wave of the NPSAS includes tens of thousands of students enrolled in college in that year, few are enrolled at highly endowed colleges and universities. In each wave there are several hundred students enrolled at these institutions. Sample size limitations are compounded when focusing on specific subgroups of students. To increase the sample size, I aggregate consecutive waves and label them by the end of the academic year (2008/2012 and 2016/2020).

Where possible, I compare the aggregate estimates obtained for these institutions with IPEDS data to see whether there are any obvious divergent patterns. Appendix Table A.1 reports total undergraduate enrollment statistics in IPEDS compared to the estimates obtained from NPSAS generated by applying the available sample weights. It appears that NPSAS somewhat under samples (or underweights) students from institutions with high endowments, more so at institutions whose endowments are in the second tier.

The two sets of data also can be used to compare the percentage of students receiving Pell Grants, shown in Appendix Table A.2. For the highest endowment tier, the comparisons are reasonably close. The gap is bigger for the next tier down, with a higher percentage of Pell Grant recipients recorded in NPSAS. Sample sizes are smaller for this group of schools, but the calculated margins of error for the sample proportions are still in the vicinity of 4%. If there is any identifiable weakness in these data, it occurs among this group of schools.

In terms of test scores, descriptive statistics obtained from students included in the survey seem roughly in line with IPEDS data. IPEDS reports SAT scores at the 25^th and 75^th percentile at each institution. Using the NPSAS data, I estimated mean scores. I compare those means to the IPEDS percentiles separately along with their midpoint to the NPSAS average. All values reported focus on combined math and verbal scores with adjustments made for changes in scoring over time.

The results are displayed in Appendix Table A.3. Although direct comparisons are imperfect, it does appear that these data are roughly comparable. IPEDS midpoints are slightly higher than NPSAS means, but SAT score distributions at highly endowed institutions may be skewed left. The patterns over time, though, are comparable.

Overall, these results suggest that the analysis provided in the body of this report provides a reasonable approximation of the characteristics of the institutions studied. Although the exact statistics reported should be interpreted with some caution, they are likely to broadly capture patterns in their operation.