Low-Income Students

High-achieving, low-income students attend selective colleges at far lower rates than upper-income students with similar achievement. Behavioral biases, intensified by complexity and uncertainty in the admissions and aid process, may explain this gap. In a large-scale experiment we test an early commitment of free tuition at a flagship university. The intervention did not increase aid: rather, students were guaranteed before application the same grant aid that they would qualify for in expectation if admitted. The offer substantially increased application (68 percent vs 26 percent) and enrollment rates (27 percent vs 12 percent). The results suggest that uncertainty, present bias, and loss aversion loom large in students’ college decisions. JEL codes: I0,I21,I22,I23,I24,I28 *Corresponding author: dynarski@umich.edu.735 State St. #5132, Ann Arbor, MI 48109. Author contact: clibassi@collegeboard.org (Libassi); kmmichel@maxwell.syr.edu (Michelmore); srowen@umich.edu (Owen). This project would not have been possible without our collaborators at the University of Michigan, particularly Kedra Ishop, Steve Lonn, and Betsy Brown. We are grateful to the Michigan Department of Education (MDE) and Michigan’s Center for Educational Performance and Information (CEPI) for providing data. Seminar participants at Boston University, Clemson, Cornell, Harvard, Northwestern, University of Illinois, University of Virginia, Princeton, Chicago, Stanford, the National Bureau for Economic Research, and Syracuse provided helpful comments, while Michael Lovenheim and Sarah Turner generously read initial drafts. The Institute of Education Sciences of the U.S. Department of Education (through Grants R305E100008 and R305B110001), Arnold Ventures, the Smith Richardson Foundation, and the University of Michigan Provost’s Office funded this research. This study is registered at the randomized trial registry of the American Economics Association under RCT ID AEARCTR-0001831, with DOI 10.1257/rct.1831. A pre-analysis plan was filed in April 2017 (Dynarski et al., 2017). The code for replicability purposes has been deposited in the AEA Data and Code Repository, openicpsr-130286. This research was approved by the Institutional Review Board at the University of Michigan (ID: HUM00096289 ) and Syracuse University (ID: 16-264). Elizabeth Burland, Meghan Oster, and Shwetha Raghuraman provided outstanding research assistance. This research uses data structured and maintained by the Michigan Education Data Center (MEDC) (Michigan Department of Education 2020a; Michigan Department of Education 2020b). MEDC data is modified for analysis purposes using rules governed by MEDC and are not identical to those data collected and maintained by MDE and/or CEPI. Results, information and opinions solely represent the analysis, information and opinions of the author(s) and are not endorsed by, or reflect the views or positions of, grantors, MDE and CEPI or any employee thereof. Gaps in educational attainment between lowand high-income students are large and have grown in recent decades. Among children born in the 1980s, those from the bottom quartile of family incomes are 50 percentage points less likely to attend college than those from the top quartile. And while 54 percent of children born into the top income quartile earn a bachelor’s degree, only nine percent of those in the lowest quartile do so (Bailey and Dynarski, 2011). These differences stem in part from disparities in academic preparation. But even among well-prepared students, there are substantial gaps in college enrollment and the quality of college attended (Hoxby and Avery, 2012). The under-representation of low-income students at selective colleges likely exacerbates both educational and income inequality.1 While there is no experimental evidence on the effect of college quality, several studies suggest that attending a college of higher quality (e.g., a flagship instead of a less-selective four-year school or a community college) increases both educational attainment and earnings (Hoekstra 2009; Zimmerman 2014; Dillon and Smith 2018). Among high-achieving students, it is application behavior that drives income differences in college quality. Hoxby and Avery (2012) find that the majority of low-income, high-achieving students apply to zero selective schools, even though doing so would likely lower their costs (Cohodes and Goodman, 2014), increase their chances of completing a college degree, and increase their future wages (Hoekstra 2009, Zimmerman 2014; Andrews, Imberman and Lovenheim 2016). Standard models of human capital investment fall short in explaining these behaviors. Though a lack of information about the (net) cost of college or suitability for an elite school could in theory lead low-income students to underinvest in education, previous interventions targeting these information frictions have shown only modest success (Bettinger et al. 2012; Hoxby and Turner 2013; Bergman, Denning and Manoli 2019; Gurantz et al. 2019; Hyman 2020, although see Jensen (2010) for an exception). Insights from behavioral economics suggest that students’ choices deviate from the classical model in predictable ways. Many observed behavioral patterns, such as present bias, overreliance on routine or defaults, and debt aversion, are particularly pronounced for those facing economic scarcity (as are low-income students) and complex decisions (as presented by the higher education and financial aid systems) (Mullainathan and Shafir, 2013). Within such an environment, small changes in choice architecture can lead to large changes in behavior. We use a randomized, controlled trial to test whether targeted, personalized communications, which reframe but do not increase financial aid, can alter the college decisions of low-income students. The intervention, the “HAIL (High Achieving Involved Leader) Scholarship,”2 was designed in the spirit of previous interventions that make small changes to the framework of decision-making. We collaborated with the University of Michigan in Ann Arbor, the state’s most selective college, 1We interchangeably use the terms “high quality”, “selective”, and “elite" throughout to refer to selective institutions. Such schools tend to spend more per student, as well as enroll high-achieving students who are inputs into the education production function (Black and Smith 2006). 2The acronym “HAIL” is a reference to the University of Michigan’s fight song. “HAIL Michigan” is plastered on t-shirts, bumper stickers, water bottles, tube tops, underwear, beer coolers, dog coats, and billboards across the state and beyond. Go Blue!