Who Merits Financial Aid?: Massachusetts’ Adams Scholarship

Most states now fund merit-based financial aid programs, the effects of which depend on how strongly students react to changes in college costs. I estimate such reactions using quasiexperimental aspects of a recent Massachusetts merit scholarship program intended to attract talented students to the state’s public colleges. Despite its small monetary value, the Adams Scholarship induced 6% of winners to choose four-year public colleges instead of four-year private colleges, suggesting an elasticity of demand for public college enrollment above unity. Nonetheless, most funds flowed to students who would have enrolled in public colleges absent the scholarship and the aid had no effect on winners’ overall college enrollment rate, which already exceeded 90%. Regression discontinuity estimates are larger than those from difference-in-difference specifications because winners with relatively low academic skill, and thus nearest the treatment threshold, reacted much more strongly to the price change than did highly skilled winners. Conditional on academic skill, low-income winners reacted similarly to their higher income peers, suggesting that previous research may have mistaken income heterogeneity for skill heterogeneity. ∗For their helpful comments, I thank Janet Currie, Susan Dynarski, Jonathan Gruber, Thomas Kane, David Lee, Johannes Schmieder, Miguel Urquiola and two anonymous referees. I am also grateful to Robert Lee at the Massachusetts Department of Education, who generously provided me with data and explanations of Massachusetts’ school system. College costs in most industrialized countries have increased rapidly in recent years, presenting a serious public policy dilemma given most governments’ stated goals of improving access to higher education. Many of these countries have experimented with various forms of student support in an attempt to improve access while targeting aid toward students for whom such aid might bring a high return on public investment. Germany has, for example, loosened the income limits for aid eligibility and tried transforming student loans into grants, though Baumgartner and Steiner (2006) find these reforms had no impact on the eligible low-income students. Australia, along with Great Britain and New Zealand, has expanded income contingent loans that base a student’s repayment amount on her current ability to pay. Chapman and Ryan (2004) argue that such reforms did increase enrollment rates, but only of students from the middle of the wealth distribution. The United States and Canada have increasingly favored merit-based financial aid, in which grants to students are contingent on demonstrated academic skill. This last form of student support is the focus of this paper. In the U.S., the increased popularity of merit-based aid can be partly explained by rapidly rising college costs. From 2000 to 2005, the price (tuition and fees) of public and private colleges rose annually by an average of 9.1% and 5.6% respectively, far exceeding the inflation rate of 2.5%. State governments have reacted particularly strongly to political pressures to reduce these costs. Since 1980 the amount of financial aid offered by the states has more than doubled relative to that offered by the federal government. By 2005, state and local governments were spending $59 billion (7.4% of revenue) on students enrolled in public postsecondary institutions, in the form of both financial aid for students and direct support for those institutions.1 As the states have increased their funding for financial aid programs, they have also shifted the mix of funding away from need-based programs and toward programs based partially or solely on academic merit. By 2005, 31 states were providing undergraduate financial aid at least partly on the basis of merit. This merit-based aid totalled $2.1 billion (29% of all state-based financial aid), of which $1.2 billion was based solely on merit and $0.9 billion was based on a combination of merit and need.2 Advocates argue that these programs raise college attendance rates, incentivize student achievement, prevent state-level brain drain, and reduce the financial burden of college. Opponents argue that targeting aid based on merit rather than need diverts funds toward highincome students and away from those low-income students whom the funds would most benefit.3 Evaluating merit-based programs depends in large part upon understanding how students, particularly at the upper end of the skill distribution, respond to changes in the costs of various college options. I explore this question in the context of Massachusetts’ Adams Scholarship, which is assigned on the basis of a standardized test score and reduces the price of in-state public colleges by about 17% in order to attract more talented students to those colleges. Using data provided by the state’s Board of Education, I exploit two quasi-experimental aspects of the scholarship in order to identify its effects. First, the scholarship’s unexpected introduction allows construction of a difference-in-difference estimator that compares the college intentions of winners and losers See Tables 280 and 282 of the U.S. Census Bureau’s Statistical Abstract of the United States: 2007 (126th Edition). U.S. merit aid data come from Table 8 of National Association of State Student Grant and Aid Programs (2006). Gucciardi (2004) finds that Canada now spends $200 million annually on merit scholarships, of which $80 million or so comes from the national or provincial governments. Heller and Rasmussen (2002) show that the association between income and test scores causes Florida and Michigan, two of the largest funders of merit-based aid, to have disproportionately low representation of low-income students in their pool of merit scholarship winners. The authors argue that this diminishes the ability of the scholarships to raise college attendance rates, which are lowest among low-income students.