If you spend any time looking at student aid research, you’ll be struck by how much empirical evidence there is on the effectiveness of grants (or, more broadly, “changes in net tuition”), and how little there is in terms of the effectiveness of loans. Thus, one might be tempted to think that this means grants are effective and loans are not, but it’s a bit more complicated than that.

There are a couple of reasons why it has been difficult to conduct decent research on the effectiveness of loans. The main reason is that in the anglophone world at least, loans are ubiquitous. They are the base of the entire system. Since one cannot ethically withhold loan awards to create a no-loan counterfactual, it is very tricky to develop an experiment which would deliver useful research results. Even quasi-experiments are hard to come by because there are very few policy changes on records which involve lowering the amount of loans available without some kind of increase in grants. Measuring the effect of grants, on the other hand, is simpler: tinkering with the loan-grant mix does not present the same kind of ethical problems, provided you are increasing grants rather than decreasing them.

(There is a second problem with measuring the effect of loans, which is that loans usually are a function of need, and it’s very difficult to disentangle the effects of one from the other.)

Now on the one hand, the lack of empirical evidence isn’t exactly a huge problem for defenders of loans. Around the world, somewhere around 30 million students a year use loans to attend school, and it’s fairly clear the that majority of them would not be able to attend without them. In this sense, loans clearly “work” even if you can’t prove it.

But a couple of weeks ago Benjamin Marx and Lesley Turner of the Universities of Illinois and Maryland (respectively)  released some interesting results which, for the first time, did create some actual experimental evidence regarding loans. It took advantage of one of the weird quirks of the American loan system: while all Americans can apply for loans, not all institutions are required to tell students about the subsidized loans for which they are eligible. This has to do with some 90s-era rules about institutional eligibility for aid: in an attempt to bring down losses on student loans, the government decided to cut off all aid eligibility to institutions with high default rates. As a result, in order to preserve their students’ eligibility for Pell grants, a rather large number of community colleges (who tend to have higher loan default rates due to low completion rates) effectively decided to stop offering students subsidized federal loans as part of their aid package (this doesn’t stop many of them from borrowing in the less-generous non-subsidized loan market, of course).

Now because of this quirk, it was possible to develop an experiment which passed muster ethically. The researchers found a community college which was willing to act as a test site, the researchers randomly assigned all student aid recipients into two groups: a “treatment group” of students who were offered the loans based on their actual eligibility for federal loans and a control group who got a $0 loan offer, regardless of eligibility, in accordance with the college’s existing policy. Unsurprisingly, the incidence and average amount of borrowing was higher in the treatment group than in the control group (30% vs 23% for incidence, $1,374 vs $1,094 for amount). But what was interesting was the academic outcomes: the students who were induced to borrow earned an additional 3.7 credits per year. In doing so obtained a GPA of 2.9 compared to a GPA of 2.3 among those who received a $0 offer. And this in turn had an impact on longer-run outcomes – the borrowing students were substantially more likely to transfer to a four-year college the following year.

The point here is not that loans are better than grants: we simply don’t know what would have happened if grants had been offered instead and so we can’t compare outcomes. But what this should tell you is that even in the absence of grants, simply giving students the option to borrow allows some of them to achieve short-term financial security, which substantially increases their chances of graduating. And given that a dollar of loans tends to cost governments about twenty-five cents on the dollar (that’s the Canadian average, anyway – mileage may vary according to details of how your loan program is subsidized), that’s an important finding.

Or, to put it another way, if a government can offer four times more in loans than grants, then the real question then the question is not so much whether a dollar in loan is equal in effectiveness to a dollar in grants (it almost certainly isn’t), but whether four dollars of loans equals one dollar of grants. That’s not usually the way the question is framed, but perhaps it should be.