Does Federal Aid Raise Tuition? The Bennett Hypothesis Tested

Hand buyers more money to spend on something in limited supply, and the price goes up. That is a law of economics that Washington has been trying to repeal since the Johnson administration, and so far the law is winning. Ask any family that has tried to rent an apartment in a city that subsidizes rents. Ask anyone who has watched the price of a ticket after the airline learned you were on an expense account. And ask yourself why a four-year degree costs what it costs, while the people who pay for it keep being told the price is a force of nature, like weather.

In 1987, Ronald Reagan's Secretary of Education, William Bennett, wrote an opinion piece in the New York Times arguing that the federal government had a hand in this. Colleges, he said, were raising tuition because aid was available to cover it. The idea has carried his name ever since: the Bennett hypothesis. It has been celebrated, mocked, tested and retested for almost forty years. Today I want to do something rare in this argument. I want to state the theory fairly, give the other side its best shots, and then tell you what the evidence supports and where the honest answer is "contested." If you want to make this case at a dinner table, you ought to be able to make it without a footnote catching you out.

What the theory actually says

Strip away the politics and the Bennett hypothesis is a claim about prices. It says that when the federal government expands the money students can borrow or receive, some of that money ends up in the college's pocket as higher tuition, instead of in the student's pocket as lower out-of-pocket cost.

Notice what it does not say. It does not say aid is the only reason tuition rose. It does not say every dollar of aid becomes a dollar of tuition. It does not say aid is bad because it helps poor kids go to school. It says there is a pass-through, some portion of aid captured by the seller, and the argument is over how big that portion is and where it shows up.

The theory rests on three links, and a good argument names them.

1. Demand is subsidized. A family's willingness to pay for college is limited by what it can raise from savings, income and what a bank will lend a teenager with no credit history. Federal loans remove much of that limit. The loan does not care what the degree is worth, and it does not ask whether the major will ever produce a paycheck that covers the payment.
2. Supply is limited. There are only so many seats at a selective school, only so many hours of faculty time, only so many dorm beds. Schools cannot expand like a pizza chain adding locations, and many do not want to, because scarcity is part of what they are selling.
3. The price signal is muted. In an ordinary market, a rising price tells buyers to look elsewhere. When the buyer pays with a loan that comes due ten years from now, the sting of the price is postponed. A seller who sees a muted signal does not have to compete as hard on cost.

Put the three together and you have a market in which buyers have more money, sellers cannot easily add capacity, and nobody feels the price when it rises. You do not need a conspiracy of college presidents. You only need people responding to incentives, which is what people do. The old lesson in I, Pencil and the price system, how nobody plans your breakfast cuts in both directions: prices carry information, and a subsidy that hides the price hides the information with it.

A toy model, labelled as one

Here is a hypothetical, built only to show the mechanism. It is not data.

Suppose a college has 1,000 seats in its entering class and could fill them all at a price of $20,000 a year. Now suppose the government raises the amount that every student can borrow by $5,000 a year, and suppose this raises what each family is able to pay at that school to $25,000. The college has no more seats. Demand just got easier to satisfy. A price of $25,000 will still fill the class.

If the college moves to $25,000, it collects an extra $5,000 times 1,000 students, which is $5,000,000 a year. Every extra dollar of aid, in this stylized case, was captured by the seller. The student got a larger loan and a larger bill, and ended up exactly where she started, with a larger debt.

Real colleges are not that simple, and that is the point of the next section. The model shows the direction of the force. Whether the force is large or small in the real world is an empirical question, and honest people argue about it.

The strongest objections

If you only hear the case for the hypothesis, you are being sold something. Here are the best counterarguments, stated the way a thoughtful critic would state them.

Objection one: state funding fell. Public colleges got a large share of their money from state appropriations, and when state budgets got squeezed, especially in recessions, legislatures often cut higher education before other line items. Schools replaced the lost dollars with tuition. In that story, tuition rose because the state stopped paying, not because Washington started. This is a serious argument and it has real support. Public universities enroll the majority of American college students, and the move from taxpayer funding to tuition funding is one of the largest shifts in higher education finance in living memory. Any honest account of rising prices has to include it.

Objection two: the cost of instruction rose. Economists William Baumol and William Bowen explained in the 1960s why services that depend on skilled labor, such as teaching, string quartets and surgery, get relatively more expensive over time. A string quartet takes the same four musicians the same number of minutes to play Beethoven that it did in 1800. Wages in the rest of the economy rose with productivity, so the quartet has to pay more to keep its players. A classroom works the same way, they argue. Add the cost of administration, compliance, technology, student services and the expectation of amenities, and prices rise without any help from a loan program.

Objection three: sticker price is not what people pay. Colleges run something like a discount airline in reverse. They publish a high list price and then give grants and scholarships to most students. The average amount a student actually pays, the net price, grew much more slowly than the sticker price in many years. If aid were simply captured as higher tuition, the objection runs, we should see net prices explode as well. A rise in sticker price can reflect a school's attempt to sort wealthier from poorer students more than a general rise in what everyone pays.

Objection four: causation runs both ways. Colleges got more expensive, so Congress raised loan limits to keep up. Policy chased prices, rather than leading them. Loan caps are often raised after complaints that students cannot afford school. If that sequence is true, then aid and tuition rise together without aid causing the rise.

Each of these deserves a direct reply. They are the reason a flat "aid causes tuition" claim fails in an argument with a prepared opponent.

The reply, point by point

Here is how the careful version of the Bennett case answers.

1. State cuts and federal aid are not rivals. They can both be true. If a state cuts its subsidy and a school raises its price to make up the gap, the school needs buyers who can absorb the increase. Federal loans are what make that possible. A school facing a funding cut in a world without easy credit would feel strong pressure to cut costs. The state cut sets the school's need, and easy credit supplies the means to meet it by raising prices.
2. Baumol's disease explains a trend, not a ratchet. It predicts that costs in labor-intensive services rise relative to other goods. It does not by itself predict that spending on administration and amenities would grow alongside, or faster than, spending on teaching, and researchers who track institutional budgets have documented growth in exactly those categories. The cost-of-instruction argument has trouble explaining price increases at institutions that added non-teaching staff rather than faculty.
3. Net price versus sticker price cuts both ways. Yes, discounting means the published number overstates what most people pay. But a school can only discount if it can charge somebody the high price, or if it is subsidized by those who do. A high sticker price funded by loans from families who qualify for little aid is exactly the channel the hypothesis predicts. And many families, middle-income families in particular, receive far less aid than the lowest-income families.
4. The two-way street is real, and it proves the problem. If Congress keeps raising limits because prices rose, and colleges keep raising prices because limits rose, you have a feedback loop. That is not a reason to stop worrying about the loop.

None of these replies settles the matter. They show why the hypothesis is plausible. What decides it is evidence.

What the research supports

Here is where I want to be careful, because overclaiming is how you lose the argument with a prepared opponent. I will describe the research in general terms and tell you where to check, because I would rather give you something accurate than something impressive.

The research literature on the Bennett hypothesis is large, and it does not speak with one voice. Three patterns come up again and again.

First, the evidence is strongest where researchers can find a clean change in aid. When Congress changes a loan limit, or a program changes its rules for one group of schools, economists can compare institutions that were affected with institutions that were not. Studies of this kind have found tuition increases at affected institutions after aid expanded. A Federal Reserve Bank of New York staff study of changes in subsidized federal loan limits, and a number of papers on for-profit colleges, are the best-known examples. The for-profit evidence is especially interesting, because for-profit schools that were eligible for federal aid tended to charge more than similar schools that were not.

Second, the effect is not the same everywhere. Pass-through appears larger at schools with more pricing power and at schools where students rely heavily on federal credit, and smaller or harder to detect at open-access public institutions with tuition set by state governments. The Bennett claim is stronger as a statement about particular corners of the market than as a statement about all higher education.

Third, the size of the effect is disputed. Different studies, using different methods, find different pass-through rates, from small to substantial. Some find tuition effects that are close to zero in some settings. A speaker who tells you the research proves a dollar of aid produces a dollar of tuition is overclaiming. A speaker who says the research has found nothing is not paying attention.

I am not going to give you a precise pass-through percentage here. The studies differ, the numbers depend on method and sector, and I would rather send you to the source than hand you a figure I cannot defend in the comments section. If you want a number, look up the Federal Reserve staff reports and the academic papers on subsidized loan limits and on for-profit colleges, and read what each one measured.

What this does not mean

It would be easy to turn a good argument into a bad one. Four cautions.

• It does not mean aid is a mistake. A Pell Grant that lets a poor kid with talent attend a community college does something worth doing. The question is how to design aid so that it goes to the student and not the seller.
• It does not mean states are off the hook. If a legislature pulls its support and tells the school to raise tuition, the legislature owns that decision.
• It does not mean colleges are villains. They respond to the rules they were given. A policymaker who rewards the behavior has no right to be shocked when it appears.
• It does not mean every price increase is aid-driven. Some of it is the quartet problem. Some of it is state cuts. Some of it is bloat. A good conservative argument says so out loud.

What would we do about it

If the hypothesis is even partly right, the fix is not to cut aid blindly. It is to build in some price discipline. Here are the ideas conservatives and others have put on the table, with the trade-off for each.

• Cap what a student can borrow by program, not just by year. A limit tied to the typical earnings of a field ties price to value. The risk is that it rations credit to students whose programs are costly but valuable, and it needs reliable earnings data.
• Make colleges share in the loss when students cannot repay. If a school bears some of the cost of defaulted loans, it has a reason to care what its graduates earn. Critics worry it will push schools away from admitting poorer students.
• Disclose net price and outcomes in plain language. Where a program's graduates cannot service their debt, say it where families will see it. Information is cheap and it sharpens the price signal.
• Let new providers in. Many of the alternatives that would pressure prices, from short credentials to apprenticeships, depend on rules about who qualifies for federal money. The gatekeeping in Accreditation, the Gatekeeper Nobody Elected That Decides Who Gets Federal Money is part of this story.

For the mechanics of how a federal loan works and why balances grow, see How a Federal Student Loan Works and Why the Balance Keeps Growing. For how to do the arithmetic yourself before you sign, see A Worksheet for the True Cost of a College Degree Before You Sign Anything. And if the next idea in your inbox is that Washington should simply cancel the debt, Loan Forgiveness, the Strongest Case for It and Why It Still Fails explains why that treats the symptom and leaves the price alone. The department that writes the rules for all of it is the subject of What the Education Department Actually Does, and What It Cannot.

The verdict

Here is mine. The Bennett hypothesis is not a slogan and it is not a fairy tale. It is a theory with a real mechanism, a body of supporting evidence that is strongest in particular sectors, and a set of serious objections that explain a good part of the story without it. The reasonable conclusion is that federal aid expansions have probably contributed to higher prices, that they are not the whole explanation, and that the size of their contribution is still being argued by serious economists.

That is a smaller claim than the one you will hear on talk radio, and a larger claim than the one you will hear in the faculty lounge. It is also the claim you can defend, and a claim you can defend is worth more than a claim that sounds good.

We should be suspicious of any policy that sends large sums into a market with limited supply and muted price signals, then expresses surprise when prices rise. We should be equally suspicious of anyone who says it is impossible that the policy had anything to do with it. Washington will tell you with a straight face that a housing shortage pushes rents up, and then insist that subsidized demand for a scarce degree has no effect on its price. Think about that for a second.

The bottom line

• The theory is that subsidized demand, limited seats and a muted price signal let colleges capture some federal aid as higher tuition. It claims a pass-through, not a one-to-one match.
• The best objections are state funding cuts, rising costs of instruction, and the gap between sticker and net price. Know them before you argue.
• The research supports pass-through in some sectors, such as for-profit colleges and changes in loan limits, and finds smaller or disputed effects elsewhere. Check the Federal Reserve staff studies before quoting a number.
• Before you or your child borrows, do the arithmetic on net price and on the monthly payment, and ask the school what its graduates in your field actually earn.
• When a candidate promises more aid, ask who gets the money first: the student, or the school sending the bill.