How to Falsify the Bennett Hypothesis

The Bennett hypothesis asserts that colleges and universities absorb increases in federal student aid and pass them back onto students as higher tuition charges. Put differently, the hypothesis states that the quantity of higher education is inelastic (insensitive) to its price, so higher-education institutions collect subsidies to students as rents instead of expanding enrollments. Importantly, the hypothesis does not in itself explain why colleges and universities absorb federal student aid (contrary to what its namesake might have believed); it merely offers the mechanism. I regularly cite the hypothesis as a guide to understanding the relationship between the federal-student-loan program and law school costs and debt.

As stated, the Bennett hypothesis is easy to falsify: Find examples where increases in federal student aid did not correspond to higher tuition charges, or show that those subsequent higher charges were due not to the government’s intervention but other factors. Time sequence, correlation, and non-spuriousness. There is a contentious literature on this topic.

Via TaxProf, I see that a research paper, “An Empirical Examination of the Bennett Hypothesis in Law School Prices” (“Empirical Examination”), attempts to test the Bennett hypothesis on private law-school tuition costs. The paper’s author, Robert Kelchen of Seton Hall University (not the law school), argues that he finds no empirical evidence of the Bennett hypothesis’ effects. I believe “Empirical Examination” arrives at unsound conclusions because it mischaracterizes the Bennett hypothesis, and it insufficiently addresses the dynamic history of legal-education financing since 2005.

(I note that “Empirical Examination” is published via AccessLex, which is related to Access Group, the erstwhile private student-lending organization that financed many law students’ legal educations before the advent of Grad PLUS loans.)

Mischaracterizing the Bennett Hypothesis

“Empirical Examination” poses two research questions that mischaracterize the Bennett hypothesis:

(1) Did tuition/fees or living expenses for law school students increase at a faster rate following the creation of the Grad PLUS program in 2006 and the expansion of income-driven repayment in 2007?

(2) Did the student debt burden of law school graduates increase at a faster rate following the creation of the Grad PLUS program in 2006 and the expansion of income-driven repayment in 2007?

These questions appear to assume that the Bennett hypothesis is disproven by discovering a lower rate of increase in law-school costs and student borrowing, as though law-school cost growth can go on indefinitely. This is unscientific. “Empirical Examination” cites no formulation of the Bennett hypothesis discussing growth rates in costs and borrowing, I know of none, and I don’t think any would be correct.

Rather, the way to test the hypothesis empirically is to take away Grad PLUS loans from students at some law schools and not others. If costs and borrowing stay the same at the Grad PLUS-less law schools, then that tends to discredit the hypothesis. In fact, “Empirical Examination” cites a study that conducted a similar test of for-profit colleges and found “some support” for the Bennett hypothesis. For law schools, the closest test case is Charlotte Law School, which lost its access to federal loans earlier this year. There is no Charlotte Law School anymore, and while this may or may not be related to federal loans, it is congruent with the Bennett Hypothesis.

The History of Law-School Lending Is Consistent With the Bennett Hypothesis

“Empirical Examination” understands correctly that around 2004, law students could borrow money from the federal government and private lenders. However, thanks to a rapid series of changes, law students could borrow all of their cost of attendance plus living expenses from the government without any private lenders (whose loans became mostly nondischargeable).

A theoretical examination of Grad PLUS loans with regard to the Bennett hypothesis would compare these changes to a hypothetical baseline without them. For example, one could try to find similar situations today in which a lender would offer an unsecured consumer loan of around $100,000 at 7 percent interest for three years. I lack the finance background to perform such an estimate, but intuition suggests the answer is not good for skeptics of the Bennett hypothesis.

To illustrate, in 2005, Seton Hall University School of Law charged $32,620 for full tuition to full-time students, and law students could borrow only up to $18,500 in federal loans, leaving them to generate the extra $14,120 (43 percent of the cost). Last year, Seton Hall charged $52,022 to full-time students. Under the pre-Grad PLUS loan system, law students would need to cover $33,522 (64 percent of the cost). I doubt private lenders would be so willing to cover more than $100,000 in unsecured loans after three years of legal education, especially given law students’ repayment prospects.

Other Weaknesses in ‘Empirical Evidence’

There are other problems with how “Empirical Examination” explores the Bennett hypothesis and legal education. For one, its focus on the increase in the rate of charging and lending ignores the fact that demand in legal education has plummeted. Last year, Seton Hall received only 1,387 full-time applications, about half as many as in 2007 (2,638). The Bennett hypothesis addresses the supply-side of education, assuming demand is constant, but when demand falls on its own, then you have law schools teetering financially. As a result, assuming demand for legal education had been constant for the last decade, it’s possible that the rate of increase in cost and debt would have continued at their pre-Grad PLUS loan pace nonetheless.

“Empirical Evidence” also discusses how many law schools funnel their revenue to parent universities. This phenomenon, certainly greatly diminished today, is also consistent with the Bennett hypothesis. If law schools were not absorbing tuition as rents, then these transfers would be unsustainable—again, assuming demand is constant. Similar arguments can be made for zero-sum tuition discounting and the dysfunctions in the law-student transfer market.


To conclude, “Empirical Examination” does not challenge the Bennett hypothesis as it applies to law schools. It mischaracterizes the hypothesis as predicting an increased growth rate in costs and borrowing, which implicitly assumes that growth in those measures is natural when they should reach a limit at some point. “Empirical Examination” does not address facts in the history of legal-education finance tending to show the Bennett hypothesis is correct.



  1. “The Bennett hypothesis asserts that colleges and universities absorb increases in federal student aid and pass them back onto students as higher tuition charges.”

    This makes me wonder whether we should be talking about the Bennett Hypothesis at all, since it appears to address only the influence of credit on the change (upwards) in tuition, and does not address the changing composition of how tuition is financed (i.e. paying cash vs. borrowing).

    Let me say that another way. Does the Bennett Hypothesis address the role of credit availability in preventing deflation in tuition cost?

    [If memory serves, Liberty Street Economics was pretty clear back in 2015 that fewer and fewer people are paying any cash towards tuition, and more and more people are debt-financing their own or their childrens’ educations in ever-increasing amounts.]

    It does not remotely pass the smell test to assert that the DOE distributing 150 billion to 200 billion dollars in subsidies each and every year to accredited institutions DOES NOT affect what those institutions can and do charge for their ‘services.’ DOE is the only game in town.

    JP Morgan Chase won’t do it. Wells Fargo won’t do it. Citibank won’t do it. Banks won’t make these loans, because even in the total absence of equal, fair, legal-economic rights for borrowers, these loans perform worse than NINJA mortgages did during the crash. Way worse.

    This is a disaster for the lives of tens of millions of people. It is the very picture of predatory lending. No matter how hard we try to isolate the damage to borrowers, the broader swath of society will not escape the fallout.

    Uncle Sam, and a screwed job market, harasses kids to take gigantic loans that neither they nor the lender can afford. The proof that the lender cannot afford it, is that the DOE takes earmarks out of Treasury auctions to make these loans, adding every penny of federal student loans originated to the national, public debt from ‘go.’ Our ‘repayments’ to DOE are not allocated to pay off or pay down that national debt, and even if they were, we wouldn’t break even.

    This is fucked. This is THE problem of the age. Bad actors extend bad loans for the purpose of enriching a few and even when said loans utterly fail to perform, the government refuses to allow the financial system to clear them, in order to preserve a fucked up system and prevent reform. ‘Extend and pretend,’ is what Varoufakis called that. But, there’s a limit. There’s always a limit.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s