In early January, Governor Andrew Cuomo of New York announced his intention to make a public college education tuition-free for most students in the state. The proposal has breathed life back into the free college movement, which supporters feared would lose momentum under the incoming presidential administration. Instead, momentum has simply relocated (back) to the state level. Tennessee and Oregon already have their own “free college” initiatives, and just this week, Governor Gina Raimondo proposed a version for Rhode Island.
Cuomo’s plan, however, would be the first to make up to four years of college tuition-free, not just the first two years at community colleges. It would do this by covering the gap between tuition and existing financial aid (not including mandatory fees) for full-time students from families earning less than $125,000 per year, who have not exceeded their program’s designated time to completion.
The ambitious proposal has roused proponents and skeptics to their battle stations. Skeptics have noted that Cuomo’s proposal provides no additional support for low-income students whose tuition is already covered by Pell Grants and existing state aid, and have argued that the money would be better spent to augment institutional resources for successful programs and services, like the well-documented ASAP program at CUNY community colleges. Proponents have argued that the simple message of “free college” will grab the attention of prospective students—including low-income students who may not even know about existing aid—in a way that “fill out your FAFSA!” does not, and that reducing prices across the board will have more political durability than programs limited only to poor students.
The contours of this debate suggest a relatively strong consensus around the importance of making new state investments in higher education (or, alternatively, the estimated $163 million price tag is so absurdly low that cost hardly seems an issue—a point I discuss further below). Rather than focusing on whether or how much to invest, the discussion has increasingly focused on whether free tuition is the most effective use of additional funds for higher education.
Specifically, does the marginal dollar spent on higher education have a bigger impact on enrollment and completion if it is used to reduce the sticker prices students face, or instead to increase institutional expenditures that affect the experience they receive once they enroll? Just a few days after Cuomo’s announcement, David Deming of Harvard University and Christopher Walters of the University of California at Berkeley presented a new study at the annual meeting of the American Economic Association, using a national database of state funding levels, tuition policies, institutional expenditures, and student outcomes over time to ask precisely this question.
The tradeoff is real: the marginal dollar invested cannot be spent in two places at once. Tellingly, the authors find large effects when state funds are used to increase institutional expenditures but virtually no effect when they are used for across-the-board reductions in sticker price.
The Deming and Walters study is not the only piece of relevant research. But what makes it particularly valuable is that it directly compares the impact of reducing sticker prices versus increasing institutional spending per student—and does so using a rigorous methodology that allows an estimate of causal effects, rather than just correlations between tuition, institutional spending, and student outcomes.
Identifying truly causal effects out of this tangled web of related measures requires some sort of external shock to the system—an outside factor that affects tuition and/or institutional expenditures, but has no independent effect on student enrollment or degree completion. Separating out the independent effects of sticker price versus institutional spending requires an additional factor that influences one of these policy levers, but not the other.
To do this, the authors make clever use of two factors that vary across institutions and over time: budget shocks and tuition caps. Budget shocks are measured by changes in overall state funding for higher education, interacted with a measure of how much each institution generally depends upon such funding for its own budget. By interacting changes in overall state funding with a measure of how much individual institutions generally depend on such funds, the authors are leveraging the fact that even within a given state and year, some institutions are hit much harder by such shocks than others.
When an institution experiences a budget shock, they have two primary tools to use in response: they can modify tuition prices, or they can modify how much they spend per student. To disentangle the effects of these two levers separately, the authors incorporate a second factor: whether or not the institution faces tuition caps (or other limitations on how much tuition can change from year to year). Institutions that can’t adjust tuition when they receive a budget shock are more likely to adjust how much they spend on students.
Putting all of this together, the authors find that a 10 percent increase in institutional spending per student leads to a 3 percent increase in enrollment and even larger percentage increases in degree completion one to three years later. In contrast, they find that sticker prices have no measurable effect on enrollment or attainment.
Why? Sticker price reductions may spread benefits too thinly across many students who would have gone to college and graduated regardless. Across-the-board tuition reductions can also distort college choices in ways that actually harm some students: a separate study by Cohodes and Goodman (2014) finds that when a Massachusetts program lowered the cost of public college, students were less likely to enroll in better-funded private and out-of-state colleges, and ultimately less likely to graduate as a result. While Deming and Walters cannot disentangle precisely what it is that institutions do with their money that makes the difference, prior research suggests that targeted financial assistance, improved advising, performance incentives, or some combination of interventions may be effective.
Of course, no study is without caveats. For example, the study focuses on incremental shifts in sticker prices—in most cases, comparing the effect of larger versus smaller increases—rather than the effect of reducing tuition to zero. The “free” message may have a distinctive impact beyond the dollar amounts involved. Taking this new research into account, here are some suggestions for moving the debate forward:
First, establish reasonable cost estimates. When cost is underestimated, it creates the illusion the tradeoffs examined by Deming and Walters are beside the point. The Cuomo administration says their proposal would cost just $163 million per year—at that price, why not do it and argue about institutional funding later? But this strains plausibility: it represents just 3 percent of the state’s current $5.5 billion dollar higher education budget, and averages out to less than $300 per full-time equivalent student. The Independent Budget Office of New York City has previously estimated that a similar plan for CUNY community colleges alone would cost between $138 and $232 million per year; Cuomo’s statewide plan including four-year colleges could easily cost two to four times that.
Carefully consider the right level for the income cap. Broad-based eligibility need not mean universal eligibility, and a lower income cap would free up resources to invest directly in institutional programs and services that the new study implies better support student success. Existing programs in Tennessee and Oregon limit eligibility to community college enrollees, and the Cuomo administration has already opened the door to a less-than-fully-universal program by suggesting an income cap of $125,000. Beyond conserving costs, lowering the cap further may preserve many of the hoped-for benefits of the program while minimizing potentially harmful side effects: for low income students, the “free” message is more likely to tip the scales towards attending college at all, while for upper income students the same message might simply change where they choose to go.
Pair increased support for students with increased support for institutions. People vote, institutions don’t. It is undeniable that “free college” is more appealing as a message than “increase support for public colleges by 3 percent.” But if the goal is improving student outcomes over the long term—not just generating short-term excitement—then policymakers may want to tie “free college” proposals to broader investments in higher education that make the college experience better, not just cheaper. This is precisely what the successful ASAP program achieved, nearly doubling graduation rates by not only covering students’ direct costs, but also simplifying course scheduling, enhancing academic advising, and providing free public transportation to get students to class.
Moving the “free college” debate back to the state level is a productive shift in more ways than one. State governments have greater capacity to actually enact such plans than does the federal government, and their longstanding direct support for institutions—not just the students who attend them—gives them a closer view of the tradeoffs involved. States can also function as laboratories for experimentation: as long as new programs include mechanisms for evaluation and improvement, then even an imperfect policy can represent a step in the right direction.
— Judith Scott-Clayton
Judith Scott-Clayton is Associate Professor of Economics and Education at Teachers College, Columbia University, where she teaches courses on labor economics and causal inference.
This post originally appeared as part of Evidence Speaks, a weekly series of reports and notes by a standing panel of researchers under the editorship of Russ Whitehurst.
The author(s) were not paid by any entity outside of Brookings to write this particular article and did not receive financial support from or serve in a leadership position with any entity whose political or financial interests could be affected by this article.
 For additional details of how the plan would work, see the official website (https://www.ny.gov/programs/tuition-free-degree-program-excelsior-scholarship) and resulting news coverage (e.g., http://www.nbcnewyork.com/news/local/New-York-Free-Tuition-Plan-Governor-Andrew-Cuomo-College-SUNY-CUNY-409845535.html).
 Focusing on public, non-selective institutions between 1990 and 2013, the authors use institution-level data on enrollments, degree completions, tuition prices, total revenues, and expenditures from the Integrated Postsecondary Education Data System (IPEDS), and augment this with data on how much state legislatures appropriate for higher education each year. The IPEDS is a national database covering all institutions that enroll federal student aid recipients (postsecondary institutions not in this database are primarily trade schools offering credentials below the associate’s degree level). Appropriations data come from an annual survey jointly administered by the State Higher Education Executive Officers (SHEEO) and the Center for the Study of Education Policy at Illinois State University.
 This study is not the first to examine the causal impact of tuition costs or of institutional resources on college enrollment and completion. For reviews of the research on strategies to improve college access and completion, see Deming and Dynarski (2009) and Page and Scott-Clayton (2016). Most of that prior research suggests that both factors matter. But because of variations in method, sample, and context from study to study, the previous body of literature has not provided a decisive answer regarding which strategy, at scale, has the biggest impact per dollar. This ambiguity leaves the door open for politicians to imply that we don’t really have to choose, and to prioritize whatever they prefer.
 Enrollments and students’ family backgrounds may affect tuition and institutional resources just as tuition and institutional resources affect enrollments and student outcomes. Economic conditions may independently affect all of the above, making these variables move together even in the absence of any causal relationship. And tuition revenue is a direct input into institutional resources, making it hard to separate out the effect of one versus the other.
 For simplicity, I use “state funding” here to refer to state and local appropriations; that is, the money that flows directly from state and local governments to institutions. Changes in state funding alone would be a problematic measure, because such changes correlate with broader economic conditions that may independently affect enrollment and completion. Because states must maintain balanced budgets, state appropriations for higher education tend to fall when the economy (and therefore tax revenues) are weak. But enrollments also tend to rise when the economy is weak.
 Specifically, they interact a per-capita measure of overall state appropriations (which varies by year, but is fixed across institutions within the same state) with the proportion of total institutional revenues that comes from appropriations (which is measured in 1990 and fixed over time, but varies across institutions).
 They use a two-stage-least-squares (2SLS) regression, with the budget shock and tuition cap measures as the instrumenting variables, to compute these effects.
 Cohodes, Sarah R. and Joshua S. Goodman. 2014. “Merit Aid, College Quality, and College Completion: Massachusetts’ Adams Scholarship as an In-Kind Subsidy.” American Economic Journal: Applied Economics, 6(4): 251-85.
 For reviews of the research on strategies to improve college access and completion, see Deming and Dynarski (2009) and Page and Scott-Clayton (2016). For evidence on the ASAP program, which combines several elements, see http://www.mdrc.org/publication/doubling-graduation-rates.
 Indeed, a study by Carruthers and Fox (2015) of the Tennessee Promise program, which makes community college tuition-free for eligible students, found large effects on enrollment and credit completion despite the fact that many students didn’t receive much additional support beyond the financial aid they were already receiving. However, the Tennessee program also provided college coaching, a feature which is not incorporated into the New York proposal.
 Total state appropriations taken from State Higher Education Executive Officers (SHEEO) database: http://www.sheeo.org/sites/default/files/Unadjusted_Nominal_Data_FY15.xlsx. Given over 567,000 students enroll in SUNY and CUNY institutions each year, this represents less than $300 per student, or only about $350 per full-time student. It is unclear how an average of less than $300 per student would be sufficient to cover the gaps between tuition and existing financial aid, even given the program’s limitations. Currently, families with income near the NY state median ($60,000) qualify for approximately $2,500 in state and federal grants, well below typical tuition charges around $6,400 for four-year and $4,400 for community colleges in New York.
 See http://www.ibo.nyc.ny.us/iboreports/free-cuny-community-college-tuition.pdf. My own rough calculations, assuming 460,000 full time students split equally between two and four-year colleges, with two-thirds of these within the required “on-time” degree timeframe, with an income distribution roughly approximating the overall state income distribution (such that approximately 85% of students would be income-eligible), and accounting for estimated Pell and TAP eligibility, suggests that Cuomo’s proposal could easily cost closer to $482 million.
 See http://www.mdrc.org/news/press-release/new-study-shows-cuny-s-asap-program-nearly-doubles-three-year-graduation-rate.