One of the ways we are irrational is our preference for instant gratification, something the behavioral economists refer to as hyperbolic discounting. This leads to all sorts of common human behaviors, such as procrastination. Doing some work now feels really high cost, but it seems like it will be less of a chore in the future. Of course, when that future time rolls around, it again feels high cost. So those of us that are irrational fall into a cycle of procrastination. Its probably fair to say we all procrastinate to some degree, so we are all irrational.
The same logic also relates to repaying a loan. The benefit of whatever we purchase with a loan right now seems high (instant gratification) and the cost of repaying it in the future seems low. Then the future comes and it's time to repay that loan. But there is something else we want to purchase in the new right now, and the cost of repaying the loan is once again high.
Let me be clear, this isn't true of all loans. Access to credit can also provide real value. It can help people invest in themselves, like student loans. It can help people deal with some of the cashflow problems of real life, like if your utility bills are due before your paycheck comes in. And, it can help you get something of real value before you have saved enough for the whole thing, like the car you need to take your kids school.
Sometimes loans can provide real economic value, and sometimes they play into some of our all-too-human irrationalities. There is some very interesting economic literature trying to figure out when loans are helpful and when they hurt.
A really interesting book I am currently reading, How Other Half Banks by Mehrsa Baradaran discusses one controversial type: payday loans. Here's a quick little primer on how payday loans work. A person in need of a loan goes to the lender and they show a recent paystub They write a check, post-dated for after their next paycheck. Then they get some upfront cash. In general, these loans are expensive, and going to people who aren't that well off (but not the worst off among us, as you need a paycheck and a checking account to get one).
The book describes that these loans cost between $10 and $30 per $100; "a typical two-week payday loan with $15 per $100 fee equals an annual percentage rate (APR) of about 400 percent". The borrower can roll over the loan if they need to, and extend they pay back period, which they also do frequently. The books describes that over 80% of payday loans are extended for more than 14 days, 50 percent are in a sequence of at least ten loans, and the average payday lending customer is indebted for 199 days.
One could imagine a world where however expensive, it is rational. The quick hit cash allows some one to pay their rent, or deal with a medical emergency. But they could also be dangerous, if people are getting loans now for consumption, believing it will be easy to pay them back later but are consistently underestimating the pain involved in paying it back later.
How would we figure this out? It's pretty hard, but Brian T. Mezler, a finance professor at Kellog School of Management at Northwestern, has an interesting approach that he describes in "The Real Cost of Credit Access: Evidence From The Payday Lending Market." Imagine you could run an experiment, where you randomly let some people who want to take out loans do so and prevent others. Then you measure who is in better financial position after the loans. Mezler figured out a way to use some survey data and a natural experiment to pretty much do just that.
First, he had to figure out what it meant to be in better financial position. He used The Urban Institute's National Survey of American Families. He looked at question about financial hardship: whether families had postponed any medical care, had difficulty paying bills, cut out meals, or moved due to finances. Strictly speaking, learning if families who took out loans experienced these these things doesn't tell us whether they are better or worse off. It is conceivable, that whatever the purchased with the loan, provided greater utility than these things, even if down the road they had any of these behaviors. But, I think its fair to say thats extremely unlikely, as these are basic necessities.
Second, he needed an experiment, but its not like he could force loans on some people and randomly deny loans to other. It turns out some states ban loans and other states, providing some variation. However, that's still not quite enough for a good experiment, because presumably policy in each state in some form responds to the preferences of the constituents. A state where the population really likes payday loans (and whatever the consequences of them are) is less likely to ban pay day loans than a state where the population dislikes them. And maybe there is some latent feature that makes people attracted to payday loans and more likely to display the behavior described in the outcome metrics.
Instead, he looks only at the population in states with banned payday loans, and uses distance to the border as a proxy for the experiment. Those closer to the border have easier access to loans, and are similar the group that in an experiment we randomly assign to get the payday loan. Those further from the border have less access, and are like the group we would prevent from getting the loan. Again, this is not perfect, because it possible that people who have a strong preference for payday loans (and the things that come with it) would choose to live closer to the border. But, it seems pretty darn reasonable to me to assume this isn't happening.
The results are downright depressing. In measured academic economics speak, Mezler states "for some low income households, the debt service burden imposed by borrowing inhibits their ability to pay important bills." People are more likely to experience each measure of financial hardship, and across the board the results are statistically significant. He finds that living in within 25 miles of the border with a payday loan allowing state is associated with 5% increase in the likelihood of having difficulty paying a bill, for families making between $15,000 and $50,000.
Later in the paper, he does some more math to try to back out exactly what this means the impact of each loan is (remember, not everyone in this group takes out a loan). Making some reasonable assumptions, he finds that a family taking out payday loan is 62% more likely to have difficulty paying bills. Similarly, he finds a person taking out a payday loan has 32% increase in the likelihood postponing a drug purchase. And remember, because of the nifty natural experiment, it appears the loans caused these changes!
From everything I can tell, this was some solid research. The paper describes multiple specifications and variations of the analysis. These include substituting a probit model instead of the linear probability model, using distance to border instead of a singular 25 mile cutoff, looking at differences in counties where people commute across borders more frequently, and a difference and difference approach which also controls for the before-and-after effects of when payday loans started appearing. The results are all quite consistent.
Most interestingly, he does two falsification tests. These are cases where he runs the same analyses, on a set of data where we would not expect to see the same result. If we do, then there is reason to believe that there is something about living near the border of one these states, and not the payday loans themselves, that are the problem. First, he does the same analysis in period of time before payday loans were common in any state. Second, he does the same analysis for families making more than $50,000, of which a very small percentage ever take out payday loans. In both cases, we do not see the increased financial hardship. This supports the hypothesis that is the loans themselves that are the problem.
Enough summarizing the paper, I have two things to say about it. First, I thought this was a fantastic paper. Interesting data, simple and elegant design for analysis, and very well written. After 5 years in graduate school, I am convinced reading academic papers (particularly econometric papers) is a skill that needs to be learned. But this one is so neatly put together that I think most educated people could understand it. So if you ever wanted to try reading an econometrics paper for fun, this isn't a bad choice. I would mock, but apparently I think reading and summarizing them is fun.
Second, I think the analysis is pretty darn suggestive that some (but not all!) loans cause more harm than good. I would add, this is particularly true when the places are joint payday lenders and rim shops (not in the paper). And I don't believe its because certain people are unsophisticated, but because human psychology is damn hard to overcome!
No comments:
Post a Comment