Observations of the USAID Evidence Summit – Day One, December 12

I was a speaker at this two-day meeting in Washington, DC last week. It was designed to educate USAID staff in the Bureau for Economic Growth, Education & Environment (E3) and the Bureau for Food Security (BFS) about the current evidence of impacts of microfinance, ultra-poor programs, value chain development, and mobile solutions. The subtitle of the Evidence Summit was

From Microfinance to Inclusive Market Development.”

While there will be videos and summaries on MicroLINKS, I would like to share what I took away from this very interesting meeting. I will do one post for each of the two days of presentations and discussion, and a third post to share my own presentation at the meeting.

Shari Berenbach, Director of E3’s Microenterprise & Private Enterprise Promotion (MPEP) Office (starting Dec. 17, she is now executive director of the African Development Foundation), gave an excellent and comprehensive review of the history of microfinance from the early 80s to the present, weaving in the emergence of business development services, then value chain development and the graduation pilots to deepen the outreach (inclusion) of the ultra poor and the use of mobile phone technology for financial transactions. This provided excellent perspective for the talks that followed. Shari’s history attempted to tie together these four areas of concentration of the MPEP Office but there was still a lingering sense of disconnect between them. Altogether, these four areas are supposed to constitute “inclusive market development” – which was described as the “mix of financial services and enterprise support services.”

Throughout this meeting, it was clear that USAID intends to focus on reaching the poor, even the poorest; there was no equivocation about the priority target population, from Administrator Raj Shah, who opened the Summit, on down. At one point, the meeting participants were asked to choose among four descriptions of what motivates USAID to focus on inclusive market development. The one that got by far the most votes: “their approaches that have reached the poorest have not linked them to markets.” However, a few USAID staff questioned whether they have reached the poorest.

Thomas Jayne on the spectrum of the poor

Thomas Jayne of Michigan State University elucidated the spectrum of smallholder farmers in Africa in relation to efforts to link them with value chains in agricultural products all the way to the global market. First, Jayne displayed a pie chart showing the livelihoods of the global poorest. I was able to capture only these figures, which are quite revealing: 610 million smallholder farmers, 180 million microentrepreneurs, and 4-500 million casual laborers. The implication drawn was that agriculture and employment are far more important than non-farm microenterprise in any plan for broad-scale poverty reduction.

Then Jayne showed a table which became the focal point for many discussions during the two days. The data were for smallholder farmers in eastern and southern Africa. These were divided into “Large Sellers” of agricultural products to the markets, “Small Net Sellers, “Autarkic” farmers (which I gather means purely subsistence or balance of buying and selling of ag products), and “Buyers/Net Buyers.” The table compared size of farms and many other features representing their resource base, skills, and participation in the market. “Large Sellers” were 2.2% of Malawi smallholders and 27% of Kenyan smallholders. For Zambia, these data really stood out:

        • Top 50% of maize sales from 2% of the farms
        • Rest of the maize sales from 30% of the farms
        • Growing but not selling maize 67% of the farms

Then Jayne compared the assets and incomes of these three groups. The 67% had an average of 1.1 ha (compared to around 10 ha for the 2% group), $756 annual household income from non-farm activities and $57 from farm activities (much, much lower in both categories than the 2% and 30%). There is a positive correlation of farm incomes and non-farm incomes at all three levels, but the proportion of non-farm income increases as poverty increases and landholdings decrease. These poorer households are being “pushed out” of agriculture (increasing not inheriting any land at all) rather than being “pulled out” by alternative livelihood opportunities. Moreover, the subsidies supposed to benefit such small smallholders are captured primarily by the better-off smallholders. They suffer two challenges: asset constraints (land, skills, knowledge, tools, inputs) and low productivity of their assets. Both together foil attempts to bring them into value chains to the external markets. The implication was that value chain development, as we know it, will not include these small smallholders until their assets and productivity can be improved by the tried-and-true (but “not sexy”) interventions: R&D for crop improvements, extension services, ag credit, input supply systems, farm-to-market roads, producer co-ops, etc., specifically for these smallest rural households.

Throughout the rest of the two days, the question was repeated: What can help these small smallholders, the 67% of rural households in Zambia (perhaps less in other countries but still the majority)?


David Roodman gave his now-standard presentation of his terrific book, Due Diligence. Dean Karlan and Jonathan Morduch spoke pretty much off the cuff, and since I was next, I found it difficult to pay close attention to their presentations, since I was mentally gearing up for my own, which was the last of the panel. Dean talked about recent research (but not the Compartamos RCT, which is not quite ready yet for public airing). He made an important observation about the importance of research to capture not just longer term outcomes, which recent research has not captured, but also immediate outcomes (which I interpreted to include mundane things like uptake of loans and use of loans). Jonathan focused on the financial diaries work in Portfolios of the Poor, especially the variability of income and how the poor use a wide variety of financing tools and transactions, mostly informal, to smooth consumption. Formal microfinance institutions add to rather than replace this mix of financial tools.

Then it was my turn. My presentation will be the next post here. The gist is that the emerging theory of change for microfinance, drawing on evidence from recent and older research, is that microfinance serves the poor primarily by providing an additional, more rule-bound and reliable tool(s) for supporting the resilience of the poor by smoothing consumption and managing financial shocks due to health and other common causes. The following Q&A was very interesting. Because I was in the thick of it, I forget the details, but two stick in my mind.

One person (Kate McKee or Shari Berenbach) asked about the relevance of microfinance for the poorest smallholders in Zambia, given they do not seem to have the necessary assets to be candidates for agricultural value chain inclusion. I suggested that given the evidence that microfinance is primarily serving to smooth consumption, microfinance may be more relevant to their situation than to the needs of those larger smallholders who are candidates. I also suggested that microfinance for enterprise (including value chain participation) probably should be emphasizing narrow targeting of true entrepreneurs (be they farmers or non-farm) rather than massive outreach, and also provision of business development services.

Another question, from Wendy Abt (DAA for E3 and moderator of this panel), asked if we need a new microfinance business model for supporting resilience of the poor. I suggested that the current, common, massive-outreach business model of microfinance is actually better suited for supporting resilience than for supporting enterprise development (including value chain development), and I repeated my thought about more narrowly targeting and adding non-financial services. Wendy asked if that means everything in microfinance is just fine after all. I said, “Not at all.” MFIs need to be much clearer about the purpose of their products and to improve product suitability for the different purposes through the kind of market research represented by financial diaries and similar methods.

Graduation Pilots

The first day finished with a panel on the Graduation Pilots (Ultra Poor Programs) in ten countries (funded by Ford Foundation and CGAP). Alexia Latortue and Syed Hashemi (via video) described the programs, which offer a sequenced package of household-level interventions. Nate Goldberg presented IPA’s impact research results for four of the ten pilots, and Jonathan Morduch described the impact research results from the SKS pilot in South India. Results from three of the pilots were very positive (well, much more evidence of positive impacts in multiple dimensions than seen in RCTs on microfinance programs), one was so-so (Honduras), and one (South India) showed no impacts at all.

Jonathan raised the question of external validity of these studies: Are the positive results (or lack thereof) likely to be found in different populations and when the program is offered by different supporting institutions? He asked also to what extent the graduation pilot interventions merely displace other activities that the beneficiaries would have engaged in. For example, in South India the lack of impact evidence was probably due to the fact that wages for casual farm labor (the usual employment for the kind of women and households served by the graduation pilots) were rising rapidly due increasing demand in the program area, which means the control group was benefiting as much as the treatment group (which was engaged in livestock production that made it difficult to engage in casual labor as well). Context matters! And so does durability of treatment effects. In the year post-program, the women had mostly sold off their livestock and returned to casual labor, which shows the importance of monitoring impacts well after the treatment ends.

In the subsequent Q&A, Alexia posed the question of what modifications could be made to the intervention package to reduce the cost per household (which for the pilots is in the hundreds of dollars) and still be effective for the household. What could the program do with one-half the money per household or even one-tenth the money per household? Jeff Ashe suggested that savings group approaches offer an effective alternative at a tiny fraction of the cost per household. My tentative take-away is that the impact results are very encouraging, but not so positive and lasting that I can yet say the evidence of impacts justifies the cost per household. A persuasive theory for how these programs can be scaled is still lacking. Yet the research seems to confirm what common sense tells us, which is that a combination of interventions from different sectors of development, properly sequenced and delivered, is much more effective than single-sector, simple alternatives, like microfinance services alone.