The Graduation Gap

American higher education, once the envy of the world, is losing its competitive edge. Most of the world's top universities are still located in the United States, but our other great accomplishment, making higher education available to an ever-larger fraction of young people, has succumbed to our hatred of taxes. According to the Organisation for Economic Co-operation and Development, young people in Australia, Britain, France, Ireland, Italy, Japan, Korea, the Netherlands, Spain, and Scandinavia, where students and families do not bear such a large share of college costs, are now all more likely to earn a bachelor's degree than are young people in the United States. That is not because American employers no longer want more college graduates. The pay premium for workers with a bachelor's has doubled since the mid-1970s and is now greater than the gap in almost any other rich nation. This trend has been one factor (among many) in the rise of economic inequality. Richard Rothstein discusses the many other steps that would be needed to reverse that rise.

What has gone wrong? The problem has three parts. First, the college graduation rate has traditionally grown in tandem with the high school graduation rate -- which hasn't risen since the early 1970s. In addition, while the proportion of high school graduates entering college has risen, the proportion of college entrants earning a four-year degree has fallen. Meanwhile, college costs have soared, and financial aid has not kept up.

Official statistics underestimate America's high school dropout problem. The Census Bureau, for example, reports that 88 percent of those aged 25 to 29 in 2008 had completed high school, compared to only 75 percent in 1970. These estimates are inflated for two reasons. First, the Census Bureau does not distinguish between those who earn a traditional high school diploma and those who drop out but subsequently pass their state's general educational development exam. Unfortunately, GEDs are not equivalent to regular high school diplomas. Employers pay GED holders less than workers with traditional diplomas, and young people who earn GEDs hardly ever go on to finish college. Second, the Census Bureau's estimates come from a survey that does not interview people in jails or prisons and misses a lot of young people with no fixed address. Both of these groups include a lot of dropouts. If we estimate the high school graduation rate by comparing the number of diplomas that high schools award each year to the number of 17- or 18-year-olds in the same year, the graduation rate was about 75 percent in both 1974 and 2005. Elsewhere in the rich world, secondary school completion rose steadily during this period and now significantly exceeds ours.

For those who enter college, the best predictors of graduation are high school grades and scores on standardized tests. This is no surprise. High school grades predict college grades, and college grades are a major determinant of whether students remain enrolled. Test scores also predict college grades as well as how easy or difficult students find college-level work. In a competitive culture such as ours, many young people respond to poor relative performance by deciding not to compete. For students who have trouble with academic work, that can mean enlisting in the armed forces, having a baby, starting a rock band, dealing drugs, or just taking a low-skilled job. Even if the best alternative is waiting on tables or feeding paper into a copying machine, such work often feels less demoralizing than collecting C's on boring school assignments.

Absolute performance also matters for staying in college, because the material is only interesting to those with a certain level of competence. Although international testing programs show that American 9-year-olds do better in reading and math than their counterparts in most other countries, 16- and 17-year-olds do not. The drop in America's relative standing as students get older suggests that there is something seriously wrong with American secondary schools, American adolescent culture, or both. The modest academic skills of American high school graduates help explain why they are less likely to earn a bachelor's degree.


The federal government and some state governments responded to the poor academic preparation of many would-be college students by launching a campaign to raise reading and math scores that culminated in the No Child Left Behind legislation of 2001. Strangely, however, NCLB focuses on raising scores in the first eight years of school, where the United States performs comparatively well, rather than improving high schools, where American students lag behind those in other rich countries.

NCLB requires American public schools to make "adequate yearly progress" toward ensuring that all fourth- and eighth-graders meet their state's "proficiency" standards by 2014. Many states decided to narrow the gap by setting their proficiency standards so low that most schools would be able to meet them without waiting for some pedagogic equivalent to antibiotics. Manyee Wong, Thomas Cook, and Peter Steiner at Northwestern University report, for example, that the federal government's National Assessment of Educational Progress (NAEP) rated only 35 percent of Maine's fourth- and eighth-graders "proficient" in reading and math before NCLB went into effect, and that the figure was similar in North Carolina. But when Maine and North Carolina devised their own tests to assess students, they defined "proficiency" in radically different ways. In Maine, the percentage of students rated proficient on the state test was very close to the NAEP figure. In North Carolina the bar was set so low that 87 percent of all students were rated proficient. As a result, NCLB led to a lot of pressure for higher scores in Maine but very little pressure in North Carolina.

Wong and her collaborators show that after NCLB was implemented, scores on NAEP's math tests improved more in states like Maine that set high proficiency standards than in states like North Carolina that set low proficiency standards. Overall, they conclude that NCLB raised math scores by about half a grade level but did not have a significant effect on reading. These findings, which are the best currently available, suggest that while pressuring elementary and middle schools to raise math scores probably "works," the gains are small. Gains of this size, especially if limited to math, are unlikely to have much impact on college graduation rates.

Although test scores are clearly important and students' relative performance is fairly easy to improve in the short run, such gains seldom persist over time. One alternative might be to put more emphasis on improving low-scoring students' non-academic skills and character traits. American schools used to emphasize instilling traits like persistence, self-control, and self-knowledge as well as developing students' ability to work with one another. American employers often say that for many jobs such attributes are still more important than reading or math skills. Some of these traits may also be easier to change than academic skills.

A recent Swedish study illustrates the importance of such traits. Until recently all Swedish men were subject to conscription. To assess potential recruits' suitability for service, the Swedish armed forces gave each a cognitive test similar to the American military's Armed Forces Qualification Test. In addition, a psychologist conducted a 25-minute interview with each and rated his fitness for military service on a scale running from one to nine. These ratings correlated only moderately with recruits' test scores. Economists Erik Lindqvist and Roine Vestman report that the psychological evaluations predict Swedish men's future earnings considerably better than their test scores do. This was particularly true for men in the bottom half of the test-score distribution, where variation in test performance had little predictive power. Although no one has replicated this study in the United States, many American employers clearly think they can predict a job applicant's performance better by conducting a brief interview than by giving applicants a reading or math test. If they are right -- and perhaps even if they are not -- schools should be taking the development of non-academic virtues and skills a lot more seriously than they have in the recent past.

The other major determinant of whether students earn a four-year degree is what they must do to pay for the experience. That depends on how much their college costs, how much their parents can contribute, how much financial aid they can get, and how much they have to borrow. As Michael Hout reports, state legislatures have curtailed support for public higher education over the past generation, so public colleges have tried to keep pace with rising costs by raising tuition. The average inflation-adjusted cost of tuition, room, and board at a public four-year college rose 67 percent between 1987 and 2008. The median income of parents with children under 18 rose only 10 percent during this period. Washington could have made up the difference by expanding its Pell grant program, but Congress decided to let Pell grants lag behind inflation. As a result, prospective college students from families of limited means had to borrow more and more if they wanted to earn a four-year degree.

Many members of Congress opposed raising the value of Pell grants because they believed (with reason) that making Pell grants more generous would encourage further reductions in state spending on higher education and further increases in tuition. Many legislators also felt that since the economic value of a four-year degree was rising, college students should be expected to repay some of their college costs by taking out loans instead of being offered grants.

The idea that college graduates should repay some of what society has spent on them has obvious appeal, but such a policy needs to take account of the fact that college is a risky investment, especially for students from less affluent families. A recent paper by Ron Haskins, Harry Holzer, and Robert Lerman for the Pew Charitable Trusts suggests that college graduates whose parents had incomes above the 80th percentile were unlikely to end up in the bottom 40 percent of the distribution when they grew up -- only one in nine was that poor. But college graduates raised in families with incomes below the 40th percentile were quite likely to remain there. One in three was still below the 40th percentile as an adult.

Forty percent of all American households had incomes below $40,000 in 2008. Four years of tuition, room, and board at a public four-year college cost an average of $57,000. If a college graduate has borrowed $57,000 to attend college and has an annual family income of $40,000 or less, repaying the loan is likely to be difficult, especially once the graduate has children. For such students, the hope that a bachelor's degree would serve as a ticket into the middle class will turn out to have been illusory. Worse yet, while most students who borrow money to attend college plan to get a bachelor's, half the college entrants who come from less affluent families drop out before graduating. For dropouts, income is usually lower and repaying loans, tougher.

Implicitly recognizing these risks, many students whose parents cannot or will not help pay for college reject going heavily into debt for a diploma of uncertain value. Instead, they decide to find a job, hoping to save some money before going to college. But the jobs available to recent high school graduates seldom pay much, so saving is nearly impossible. Students who take such jobs also tend to acquire partners and children. Once a child arrives, neither saving nor borrowing enough to attend college is usually feasible.


Two policies seem to follow. First, if we want students to attend college, we need to get them there as soon as possible, before they start families or start to feel that taking classes and exams is infantilizing. Second, if we are to encourage less affluent students to enroll and persist, we need to make college less of a financial ordeal.

A less satisfactory solution is to rely on two-year community colleges to educate less affluent students who are reluctant to borrow. Most Americans now live within commuting distance of a community college, so most high school graduates can attend such a college while continuing to live with their parents. Community colleges also have lower tuition than public universities, so their students can often cover their out-of-pocket costs with a Pell grant and a part-time job. The articles on pages A14?A20 discuss the promise and pitfalls of a community college solution.

At present, students who enter public two-year colleges seldom earn a two-year degree, and even fewer transfer to a four-year college and earn a degree. But that outcome may not be inevitable. A new study by Jennifer Stephan and James Rosenbaum at Northwestern University and Ann Person at Mathematica Policy Research reports that students who attend public community colleges are very similar to those who attend private two-year colleges. Private colleges charge far more than the public ones do, but they also provide prospective students with far more assistance in lining up Pell grants and low-interest loans. As a result, students in public and private two-year colleges have relatively similar out-of-pocket costs (and family incomes). The two groups are also very similar on a wide range of other measures, like parental education and income. Nonetheless, private two-year colleges have substantially higher graduation rates than public ones. Stephan and her two co-authors attribute the private two-year colleges' success to the fact that they offer more structured programs, monitor student progress more closely, and have more intensive advising. Julie Strawn, page A15, reports a similar conclusion. If these policies are in fact the keys to better graduation rates, state legislatures should start offering community colleges incentives to embrace them.

So which route makes the most sense? Should we try to raise test scores, so that students find college easier and are more inclined to stick around? Or should we try to make college more affordable for students who view big loans as too risky? The easy answer is that we should do both. But making college a lot more affordable is a challenge governments know how to meet, while making students learn a lot more is a challenge we do not currently know how to meet. Under those circumstances, starting with affordability is probably the best bet.