We Need to Keep Learning about Overindebtedness

It is my privilege to close out this blog series which has been a wonderfully rich exploration of one of the most complex and consequential topics in microfinance. I’m tempted just to sit back and say – I like what Rich wrote about the challenge of definitions and how Jessica illustrated the difficulty of drawing the line of “acceptable sacrifice” and the way Jacco laid out the responsibilities of various stakeholders. In fact, that’s largely what I propose to do, while taking up the overall theme of learning from research.

This series demonstrates the wide field of research emerging on overindebtedness. Several of the contributors to the series are the first movers who have established tentative hypotheses and tested research methodologies. We need more such studies to inform product design and policy making. Milton is skeptical that more knowledge will lead to better outcomes, and he does have a point. It’s quite possible that the only way to learn and take the steps necessary to produce a better outcome is to plunge a market into a crisis. Even better, however, would be to use the results of research. Although the studies profiled here are a huge step forward, it is still early to use them for making decisions: it is not yet clear how their findings will hold up in other countries and different settings.

We need to know more about clients. In the absence of microfinance, or in addition to microfinance, how much debt and what kind of debt do people have? What sacrifices come up in their lives, how often, and what do they think about them? As Sarah asks, what are the consequences of default for clients? Does the picture Jessica drew from interviews in Ghana hold up in Ecuador, Morocco, or Nepal? This kind of understanding is fundamental.

We need to look more carefully at the interactions between microfinance institutions and their clients, examining how MFI practices play out in lives of clients on an everyday basis. Stuart posed two specific hypotheses: if a lender needs clients to borrow continually, it incentivizes overselling and overindebtedness, and that insistence on immediate, in-full repayment drives some clients to begin bicycling loans. These hypotheses deserve to be tested in many circumstances. Another hypothesis I would like to see examined is that clients with both savings and credit are less likely to become overindebted than those with only credit.

These are the kinds of explorations that should shape microfinance methodology and product design. They may lead to very significant rethinking of both. I worry when practices like the ones Stuart cites are considered business as usual by a large swath of providers in a given country. In Bangladesh and India it is difficult to identify specific bad players, since a number of bad practices have been used by many. His examples signal that a solution that’s good for clients will require the difficult task of changing the practices of many providers across a short time. As Rochus points out, such change can happen quickly when regulators respond vigorously to a crisis as they did in Bolivia in 2000 and (as Gabriel notes) in South Africa in 2002. Both those countries were blessed with exceptionally competent regulators.

Getting more operational, a study of Bosnia study carried out for the European Fund for Southern Europe by the Microfinance Center in Warsaw helped give empirical backing to what is probably the number one rule of thumb (and time-tested MFI experience) regarding overindebtedness: the debt service to household income ratio. Since this ratio is often the backbone of both credit underwriting methodologies and client protection regulation in many countries, the more we understand it, the better. A deep body of data would help develop the ability to answer questions such as these: How does the debt service ratio correspond with default? Or with clients perceptions about acceptable sacrifice? What is the best cut-off for credit approvals? What about the cutoff for reckless lending?

The Bosnia study provided strong evidence about another number – cross-indebtedness, or the number of loans from different institutions per customer. In Bosnia, indebtedness across two providers was fine; defaults jumped noticeably with three, but not to a level that would be hard for lenders to handle. At four loans, however, the likelihood of default jumped up steeply and it was astronomical from five on. Data from credit bureaus in Peru shows an identical pattern. Could this be the start of an iron law? It would be terribly convenient, because number of multiple loans is such a straightforward indicator, and currently microlenders are not very secure about the rules they use regarding cross-indebtedness. Where credit bureaus exist and cover microfinance, this kind of data is easily assembled, and it would be a great service to the industry for someone to examine the data across a number of countries. One caveat: the figures from both Bosnia and Peru mainly represent individual loans. Things may look different with group loans, especially those with highly constrained loan sizes, as Sanjay has pointed out regarding South Asia.

At the most aggregated level, the University of Zurich study Jacco, Sarah and Rochus reported on began to test market-level indicators to provide an early warning about overheating. The model has been built using data from the past, but it remains to be seen whether it can provide useful predictive value. If so, it can be especially useful for investors and regulators.

I have been examining all this information from the perspective of the Smart Campaign, which is developing methods to certify financial institutions on whether they apply the Client Protection Principles, including the first principle on preventing overindebtedness. I want to know whether the lender policies and procedures the Smart Campaign’s assessors look at accurately reveal whether a lender is pushing its clients into too much debt. The Smart Campaign’s assessment procedures already use most of the indicators that have just been mentioned, especially the debt service ratio, the policies regarding cross-borrowing, and the loan officer incentive package.

The assessments also check that lenders are not doing some of the things Gabriel warns about, like abusive collections and excessive penalty fees. Assessors also consider the enabling environment (presence of unique I.D.s, credit bureaus) and market conditions (level of competition, market penetration). If conditions press lenders toward disbursement of more and more loans while providing little borrower information or regulatory oversight, responsible practices by the provider will be trumped by the market-based incentives to lend more. Renso and Scott indicate that Smart Campaign endorsers are not significantly stricter than non-endorsers about cross-indebtedness; I hope and expect that to change as the Campaign deepens from awareness raising to implementation and finally to certification.

To end, a call to action. As Jacco says, everyone has a role to play in preventing overindebtedness: researchers in discovering the most valuable indicators of overindebtedness, providers in using research to inform policy and being willing to change practices, investors in rewarding good providers and avoiding bad ones, and regulators in prescribing safe market conduct and prudent lending rules.

Finally, clients. It is only right to close this series by turning to clients and their literacy, as Shameran highlighted. Karuna reports that in a sample of clients in Karnataka, higher financial literacy was associated with less repayment stress. No client can fully trust the advice a lender gives about how much debt is sustainable. Clients have a responsibility to make their own realistic judgments about how much debt they can sustain, and lenders have a responsibility to assist clients to understand their limits.