More Mysteries of Savings

A lot of progress has been made in understanding the savings behavior of poor households over the last few years. A raft of new studies are beginning to appear that  promise to advance our understanding further.

But thus far, the new studies are providing as many new mysteries as answers. David Roodman does a good job of exploring the mysteries presented by one of the first of these studies—a trial of a commitment savings product in Malawi. In summary, the researchers found that access to a commitment savings product helped increase savings balances—but NOT in the commitment savings account.  Sometime soon I’ll be blogging about a new paper from Pascaline Dupas and Jonathan Robinson that similarly suggests that a commitment device works even though it doesn’t require much commitment.

The mysteries aren’t limited to commitment devices though. Last week I attended a presentation by Dina Pomeranz of Harvard Business School at the Center for Global Development (CGD) in Washington. She presented a study she and colleagues had conducted in urban and peri-urban Chile looking at ways to increase savings. Pomeranz et al. basically tested two different plausible mechanisms for increasing savings against a baseline of access to a typical savings account. One test is of a dramatically higher interest rate. The baseline account, offered to the control group, is a savings account that paid .3% real interest with no fees. One treatment group was given access to an account that paid 5% interest (yes, 16x higher than the baseline). A second treatment group was invited to participate in a group modeled on Alcoholics Anonymous but with the intention of encouraging group participants to save (rather than resist an addiction) but had access to only the standard (.3%) account.

You may not be surprised to hear that Undersavers Anonymous (UA), as the researchers call it, seems to work quite well to increase savings. People offered the chance to participate in the group made roughly twice as many deposits and had a savings balance nearly twice as high as the group with the basic, low-interest account.

But there is a mystery in how UA is working and what it is actually doing. It seems clear that the UA treatment helped people resist temptation and overcome inertia to get to the bank and put some money away. But savings balances for UA members leveled out very quickly, meaning the balances didn’t continue to grow. That suggests that savers have a set amount they desire to save for emergency purposes and don’t feel a pressing need to save more than that. More curiously, a similar leveling out was seen in the control group, but at approximately 50% lower balance than the UA participants. And the “set point” doesn’t seem to converge at all. Why and how did the UA treatment change people’s savings set point? Why is the control group able to overcome anti-savings behaviors despite the lack of support but not able to continue to build savings balances over time and close the gap—all of which appears to come from additional saving in the first few months for UA participants—on the UA group?

Another surprising finding is that apart from a few outliers, the high interest savings account seems to have had no impact on increasing savings balances. In fact, that understates it. People with access to this very high interest account didn’t even transfer their existing savings balances to the high interest account. You might assume that this means that the participants didn’t understand the offer—and you would be partly right. In conjunction with the high interest rate account, participants in the study received a one hour training class on compounding interest and what a difference interest rates can have on savings balances over time. Still, even after the class, participants didn’t seem to really understand the differences between high interest and low interest accounts.

That’s yet another disappointing finding for financial literacy efforts. One of the possible reasons I’ve heard proposed that financial literacy training doesn’t seem to have much impact is that few good financial options are available to those who receive the training. In other words, it won’t help much to understand compound interest if there isn’t an interest bearing account available to you. But in this case, the training was given and the interest bearing account was immediately available. And still, it doesn’t seem to make any difference.

Overall, my take on the study is that it reinforces the idea that the barriers to saving are primarily behavioral and not financial—and thus the interventions likely to have the most impact shouldn’t be focused on financial incentives but on methods of defeating behavioral barriers. That’s very useful information for savings banks in terms of cost control. The message is don’t use limited resources on boosting returns but focus them on marketing and support services.

Personally, I’m very interested in the mystery of the differing savings balance set points and why neither group was continuing to accumulate savings.