The expectation is widespread that colleges should provide pathways to prosperity and success for individuals–and also make the distribution of wealth and power in the whole society more fair and equitable. Higher education seems to have that potential because it does promote upward mobility for individuals. Even if you are born in the bottom fifth of the income distribution, if you attain a college degree, your chances of staying in the bottom fifth are only 1 in 10.

But the results at the national scale are disappointing. Only a few (14.5%) of children born in the bottom quintile complete a four-year degree. The pressure for higher education to promote economic mobility better comes from various quarters–conservative governors angry that college degrees aren’t aligned with employers’ needs, technocratic liberals in the Obama administration trying to push down college costs, and campus activists demanding that student populations be representative of the national population in terms of race and social class.

The pressure is understandable, and much of it is valid, but we must squarely face the reasons that higher education does not promote mobility if we want to improve results.

College graduates represent the upper strata of society. About 40% of US adults hold college degrees, and they increasingly fill the top 40% of the slots in the whole socioeconomic hierarchy. That is why it is good advice to an individual student to complete college. Four-year colleges intentionally prepare their students to be doctors, lawyers, engineers, business leaders, and teachers, not bus drivers, receptionists, or construction workers. That’s what they offer; that is what they’re for. It means that the alumni body, almost by definition, will not represent the socioeconomic profile of the country.

The alumni could, however, represent the racial and gender diversity of the country if race and class were decoupled. And the alumni could come from economically representative backgrounds. Half of college students could be raised in homes below the median income, even though, as alumni, they would occupy the top 40% of the socioeconomic scale. That would imply a great deal of upward mobility, and it’s one criterion of justice (justice as an equal chance to succeed).

The barriers, however, are serious. One obstacle is that we don’t have an expanding middle class. Adjusted for inflation, the median US family earned $54,000 in 2014, about $4,000 less than in 1999. If families in the middle are becoming worse off, then upward mobility for some implies downward mobility for others.

That zero-sum model is an oversimplification, but there really are only so many jobs for lawyers, doctors, engineers, and corporate executives. Families who already stand in the top 40% will fight tooth and nail to keep their own kids in that tier instead of trading their spot with someone from further down the hierarchy. They will spend whatever they must on housing, k-12 education, and extracurriculars; apply whatever pressure they need on their alma maters; and take full advantage of their social networks to keep their kids ahead of the median. They may value both racial and socieconomic diversity as educational assets, but not to the extent that they are willing to give up a spot at a top college to someone else’s kid. And it is highly implausible that raising their consciousness about injustice will change their behavior. I can think of no comparable outcome in human history. People at the bottom of hierarchies do achieve change by obtaining power, but not by making the people at the top feel bad.

We might think that by investing in human capital, we could expand the middle class and make space for more people above a real income of $54,000. To some extent, that is both possible and important. But my sense is that left-of-center economists are now criticizing the human capital thesis. Yes, individuals who have more valuable skills, more elite social networks, and more cultural capital will beat out the competition for desirable jobs, which is why a college degree predicts high earnings. But raising the proportion of people with college degrees–as we have done–does not make the society more equitable, because other factors are increasing inequality. Prime suspects are deregulation, capital mobility, and monopoly power.

Finally, at the individual level, a person must move a long way from the levels of financial, human, social, and cultural capital available at the bottom of the economic hierarchy to reach the levels that are expected for candidates for great jobs. It is implausible that you can move nearly far enough during four years that start at age 18. To the extent that higher education is a path to upward economic mobility, it is near the last mile of that path. Colleges and universities compete for applicants from relatively disadvantaged backgrounds who have already received unusual support from families, community groups, and k12 schools. Colleges can play a useful role by a) being aggressive about finding these applicants, the ones who are ready to move the last mile, and by b) serving such students well. But they are not going to get disadvantaged kids 80% of the way. They offer too little, too late.

These are reasons that higher education is unlikely to make more than a modest dent in the inequality problem. To address that problem in a serious way, we need different economic policies. In a different political-economic climate, colleges and universities could do their jobs (educating students and producing knowledge) with less injustice. Given the reality, I believe our top priority is to understand the inequality problem better and to make the knowledge that we produce more politically significant. I don’t mean only economic research and policy advocacy but also normative inquiry, cultural critique, social experimentation, public dialogue, and a range of other research-related activities. Meanwhile, we must deal with an unjust context as ethically as we can, which means handling matters like admissions, hiring, and retention thoughtfully  while also acknowledging the inadequacy of these responses to the underlying problem.

[PS: I read later today that “colleges themselves are responsible for about 5 percent of the variation in students’ earnings later in life,” which reinforces my argument here.]