# Microfinance: The multi-billion-dollar aid sector that we’re just starting to learn about

GiveWell attended the Microfinance Innovation & Impact Conference via Board member Tim Ogden and part-time employee Stephanie Wykstra. (Our full-time staff could not attend as we are all in India). Our main takeaways:

• The people involved with this conference, and the nonprofits working with them, are producing a lot of promising, interesting, rigorous and informative work on how microfinance affects the poor and how these effects might be improved.
• While microlending is the form of microfinance that has been most funded, celebrated and scaled up over the decades, it appears to be the least proven and least-well understood in terms of its impact on the poor. In many ways it seems that good information on how (and to what extent) it’s helpful is just starting to become available.

Content presented at the conference
There is much abstract agreement about the importance of learning from good research, but we feel that there are few concrete examples of groups actually doing it. A high concentration of those examples could be found at this conference. We have written before about how impressed we are with the work of the Poverty Action Lab and Innovations for Poverty Action (groups behind this conference), and that continues to be the case.

The research has already been summarized in a large number of posts by others (see the roundup by GiveWell Board member Tim Ogden). Our quick take:

Microsavings, microinsurance, and “ultrapoor programs” are clearly promising. High-quality studies found significant improvements on metrics like income, food security, and children’s ability to attend school when these programs – which differ markedly from the loans most people associate with “microfinance” – were rolled out. Programs included accounts intended to help people pre-commit to saving (more); intensive promotion of difficult-to-sell insurance products (more), and “ultra-poor programs” that target people with extremely low incomes and offer direct benefits such as cash/asset transfers and education, rather than financial products (more).

Unfortunately, we know of no clear vehicles through which individual donors can fund these sorts of programs. We’ve written before about the frustrating situation of seeing a good program without a good funding vehicle, and called for major funders to consider this problem. There do appear to be similarities between the “ultrapoor program” and the work of GiveWell-recommended charity Village Enterprise Fund.

The impact of microlending is, at this point, not very well understood. A new high-quality study in Morocco found no positive social impact overall, though it did find that people with existing agricultural businesses appeared to invest more/differently in their businesses (more). These results were similar to those observed in the only previous high-quality study of traditional microlending, which took place in a very different environment (Hyderabad, India) and was summarized last year by David Roodman.

There was significant discussion of the risks that people might be overborrowing (something we are quite concerned about, and something that there is apparently still little information to assess). And new data from the Philippines reinforced a point we have emphasized in the past – that microloans are more likely to finance consumption, and less likely to finance entrepreneurial investment, than most donors suppose. From Philanthropy Action’s summary:

Around 46 percent of clients used new debt to pay down old debt, and 28 percent used funds for a single large household purchase. In all, 39 percent of the money goes unaccounted for, 15 percent is used to pay down other loans, 9 percent goes into the household and 37 percent is invested in the business. Curiously, Karlan’s study got at these results by asking clients the same questions in a number of different ways. Not surprisingly, when the banks ask clients why they want the funds only 2 percent of clients admit potential uses other than the business, even after the funds are given. Only when an independent surveyor asks indirectly do any relevant percentage of people admit to non-entrepreneurial uses for credit.

Studies are finding interesting potential ways to improve the impact of microlending. One possibility is better targeting (as discussed above, microlending appears to affect different people differently). Two other possibilities raised at the conference, both with encouraging preliminary evidence behind them, were the implementation of grace periods for repayment (more) and financial education focusing on “rules of thumb” rather than formal accounting (more).

Learning how microlending works – better late than never
It appears to us that an enormous amount of philanthropic capital has gone into microfinance. Last year we looked at data from 2008 and found that $11.7 billion had gone to microfinance funding of some kind, with$2.2 billion being grants. For context, total official aid earmarked for health was about $3 billion disbursed /$8 billion committed in the same year (data from Aid Data).

My greatest fear about microfinance is that all (or nearly all) of that money is chasing a good story rather than a good program.

In this context, it’s hard for us to see how one can argue that good evaluation is too expensive. The total budget of Innovations for Poverty Action that year was just under \$6 million (data from NCCS).

• Andrew on November 5, 2010 at 8:13 pm said:

Do you expect your findings from the conference will affect the position of any of the economic empowerment charities in your ranking? (Specifically, you mentioned that VEF’s model appeared more promising according to the findings presented at the conference.)

• J. S. Greenfield on November 8, 2010 at 10:20 am said:

So if GiveWell has long been and continues to be impressed by Poverty Action Lab and Innovations for Poverty Action, at least the latter is a charitable organization that accepts donations, and you think a relatively small amount is currently spent on such work (suggesting the possibility of room for further funding), how come you haven’t evaluated it, as a charity?

• Thomas Cutting on November 8, 2010 at 8:58 pm said:

I found this post and your many links to be a useful summary of the Conference’s presentations and findings. What really struck me was your remark that “microfinance may be chasing a good story rather than a good program.” A friend of mine spent time in Latin America participating in microlending. He believes that the work he did for his clients had positive long-term effects, but still admits that the vast majority of the marketing of the small organization involved emotional appeal rather than numbers and figures (granted these are difficult to obtain.) Despite a lack of hard statistical data, I still cannot help but feel that microfinance plays an important, helpful role. Nonetheless the studies as to the overall effectiveness of microfinance coupled with the suicide reports out of India are rather depressing. While the Conference may have come to premature dire conclusions, it should serve as a catalyst to more research in the near future.

Tamara Straus of the Stanford Social Innovation Review mentions that Compartamos Banco will be providing data to Dean Karlan of Yale in an attempt to better understand the effects of microlending. Are you aware of what specific questions he will be attempting to answer or the extent of the study for that matter? I believe that cultural and gender specific questions must be raised. Additionally, I think an in depth investigation into the support programs provided as well as trials as to the feasibility and effectiveness of several new innovations (grace periods and “rules of thumb” education to name a few) could revolutionize the industry. The market for these loans is there; MFI’s just need to focus on tailoring their products to fit the specific needs of their clients. This new push for research should provide the type of guidance these institutions require. It is difficult to believe that it has taken this long to investigate the effects of the microfinance industry, then again a lack of feedback and poor results reporting are unfortunate issues affecting the philanthropic sector at large.

J.S. Greenfield, to date we’ve considered IPA to be outside our “scope”, as it is research rather than direct aid. For purely practical reasons we’ve confined ourselves to one sector at a time. However, we’re now broadening our reach and planning to evaluate at least some areas of research in the near future. We also see that IPA will soon be launching a “proven impact fund” with more of a focus on direct aid. So I think an evaluation of IPA is coming in the fairly near future.

Thomas, thanks for the thoughts. I don’t know the details of what Prof. Karlan is investigating in this case.

Andrew, good question. At this point I don’t think our rankings will be affected.

As I stated, we can’t find any appropriate vehicles for the programs with the most encouraging results (savings, insurance, ultrapoor-focused programs). There are similarities between ultrapoor-focused programs and VEF, but considering the strength of the results and the strength of the similarity, I don’t think there’s enough there to push VEF up in the rankings. I believe that many more high-quality studies of ultrapoor-focused programs are currently underway and if the results are consistent, VEF may be promoted at that point.

As for the microlending-focused organizations, our position on microlending is still pretty similar to what it’s been since we started investigating them – i.e., we think the fact that people repay their loans with interest is encouraging, but nobody has observed welfare effects big enough to show up in a formal high-quality study. The main change is that we now see more “evidence of absence” of these large short-term welfare effects than “absence of evidence” (what we saw before).

Our feelings about the organization, as distinct from its program, carry a heavy weight in our rankings; our views on an approach would generally have to shift pretty substantially for rankings to change.

Ranking across sectors is one of the fuzziest and most difficult decisions we make and I’m interested in your and others’ input on this.