Continued from Part I and Part II, these are my thoughts from my recent visit to two of our top charities in Africa.

Some of what I saw and discussed prompted me to rethink our frameworks for evaluating certain kinds of programs:

• Vaccinations. We’ve taken the “vaccination coverage rate” as a reasonable proxy for lives changed, since the evidence base for vaccines is so strong. But of course, “vaccination coverage rate” describes how many children received vaccines, not how many received functional and correctly administered vaccines. I was somewhat concerned that VillageReach staff found several vaccines in refrigerators that had “gone bad,” and I was glad to hear that VillageReach is considering adding an indicator to its information system to track how often this happens. The strong macro-level track record of vaccines (causing major drops in mortality at the country level, not just in carefully controlled trials) is some comfort here.
• Microfinance. We’ve been concerned about the possibility that clients are taking out loans against their own best interests, and have largely pictured “coercive” versions of this problem: loan officers pressuring clients to borrow more than they should, clients getting themselves into debt cycles, etc. A very interesting anecdote from a staffer raised a more subtle version of this concern: clients may be losing money on their loans without knowing it. The anecdote given was about a particular woman who was literally selling goods for the same price she had bought them for, making the problem obvious. It could, however, be a much more subtle problem for other clients – given high interest rates, potentially transportation costs, etc., it could take quite a bit of calculation and careful accounting even to know whether the business one is running with a loan is in fact operating at a profit or a loss. (And since many families may have several sources of income, a loss might not be noticed if accounting isn’t careful.)
• Cash transfers.. Our position has been that cash transfers can be assumed to be doing some good if they are successfully targeted to poor people in an area, something that may be difficult. It struck me that in certain areas (such as the village I saw with VillageReach), poverty targeting may not be much of a challenge at all (since everyone anywhere near the area is extremely poor); on the other hand, in these kinds of areas gifts of cash or livestock may be of very limited use (note the missionaries’ claim that village people receiving pensions for military service were largely spending them on alcohol).
• Social business. I was impressed that I constantly saw Vidagas canisters throughout my trip – in hotels, stores, even the missionaries’ truck. (Vidagas is a “social business” started by VillageReach; it delivers gas, and was started in order to address the challenge of consistently powering refrigerators to keep vaccines at the appropriate temperature.)Our position on social business has been that such a business should not be considered a success until it has demonstrated either an actual profit (not just sales covering unit costs) or demonstrable social impact along the lines of what we look for from nonprofits. On reflection, I think that in certain cases there is room for more middle ground here. There are certain areas where the mere fact of selling something for a non-trivial price would seem to indicate a certain success in filling a need, even if not all costs are covered. Of course, it all depends on the area – subsidized sales may make a lot of sense where infrastructure and access to markets is poor, but in urban areas it could serve simply to “crowd out” private supply and/or enrich middlemen.

  I don’t regret our skepticism of social business to date. It has always been more important to us to avoid “false positives” (i.e., recommendations of organizations that are not impactful) than to avoid “false negatives” (i.e., failures to recommend organizations that are impactful). And I have not seen any “social enterprise investment” fund put together the case I’d need to see, even using the “middle ground” roughly sketched out above. But I do want to keep thinking about how to recognize the good social businesses may be accomplishing without being overly credulous.

These visits very much made the activities of our top charities feel more “real” to me.

To this point, the work we’ve done on international aid has felt very abstract. That’s not a reason not to act/give based on it, but in many ways the situations we’re analyzing are so different from what I see every day that it can be hard to believe that the charities are helping real people in the way our analysis suggests they are.

Much of what I saw on the trip was, in fact, consistent with what I expected. To a large degree, it made the research “come to life.” I saw people and areas that really are at a level of poverty that I’ve never seen in the U.S.; I talked to staff about the details of what they’re doing, and to some degree saw them doing it; and I felt, very tangibly, how the work they’re doing can make a difference.

(As an aside, I’ve had the opposite experience with site visits to U.S. charities. I’m not sure why. The U.S. visits were definitely more “staged” while the international visits had a lot of wandering and improvisation; in addition, the U.S. charities tend to address less tangible problems, and it was often hard to connect the charities’ theories of their own value-added with what I was seeing.)

It was frustrating to say “no” to kids rubbing their stomachs and asking for money, and to see so many people who seem like they could benefit greatly from things that are pretty basic – though not necessarily easy to deliver. The bottom line is that while I’ve pushed to make my actions consistent with my beliefs, my beliefs about the importance of international aid carry a little more emotional weight now, and I feel more emotionally motivated to give and to give well. I would recommend a similar trip to anyone who intellectually accepts the importance of international aid, but is having trouble getting behind it emotionally.

Continued from Part I, these are my thoughts from my recent visit to two of our top charities in Africa.

Diverting skilled labor looks like a real concern.

The COO of SEF stressed that one of SEF’s biggest challenges is human resources (i.e., continually finding good people to staff it). I can easily see how this would be. As I mentioned in Part I, I found that the nonprofits I visited were employing capable, impressive people with a combination of local background and well-above-average educational credentials and command of English.

On one hand, seeing these staffers made me feel good about the organizations we were recommending. At the same time, it highlighted one of the most universal and hardest to evaluate concerns we have about nonprofit work: diversion of skilled labor from other potentially productive pursuits.

Adding to this concern was a general impression I got (reinforced by Leah from VillageReach) that nonprofit jobs are among the best-paying and most prestigious jobs for African locals. It looks like we have a situation where:

• Many of the people hired by nonprofits could also be potentially very helpful to their communities if they were doing for-profit work.
• They work instead for nonprofits, partly because nonprofits are out-bidding the for-profits for their services.
• Within a for-profit framework, there is often (not always and never perfectly) a connection between the value of a job and the salary, which creates a (imperfect) tendency for talented people to end up in roles where they can do more good.
• I have no sense of how (or whether) nonprofits are attempting to calibrate salaries and value, and I fear that they could be “overpaying for” (and thus misusing) local talent simply because they want the best people available and they have the donor-supplied funds to get them.

More on this idea in a future post. Though we have no great methods for quantifying the losses from “diversion of labor,” we do believe that this concern reinforces the importance of demanding that nonprofits be accomplishing as much good as possible and not merely some good.

Getting basic info about people’s standard of living seemed fairly straightforward.

I understand that estimating people’s incomes can be a very complex endeavor, but in the areas I visited, it seemed possible to get a sense very quickly for how “poor” one area was relative to another. I asked basic questions at the village level: where the nearest water source was, who was responsible for maintaining it, where the nearest school was, what the school fees were, etc. I walked around and observed how many of the dwellings were made of mud vs. concrete . And when talking to individual clients, I asked straightforward questions like “Do you have a TV?”, “Do you have electricity?”, “What do you eat?” and “When was the last time you had a fever and what did you do about it?” Answers were fairly consistent in a given area, but varied dramatically across charities (more below).

Throughout our investigations into international aid, I’ve been frustrated by the fact that most charities seem either unable or unwilling to produce data on clients’ standards of living. Because I don’t tend to trust stylized stories, and I haven’t had what I consider credible data on standards of living, I’ve constantly felt very unclear on who is being helped and how. I now find it less likely that this problem stems from prohibitive costs of data collection; I find it more likely that it stems from (a) the fact that donors rarely (if ever) ask for data on clients’ standards of living; (b) the possibility that some charities may not want to reveal that their clients are anyone but the “poorest of the poor” (even when their clients are still quite poor).

The three areas I visited were very different in terms of standards of living.

• Small Enterprise Foundation (SEF) clients: I visited two villages, one in the Microcredit program (SEF’s original program) and one in the Tšhomišano Credit Program (targeted more directly at the poorer people in a village). In both villages, at least half the buildings I saw were made of concrete, and everyone I spoke to reported convenient access to running water, electricity, a fairly well-stocked local market, and public transportation to larger cities. Living spaces appeared fairly cramped (they were larger than in the other areas I visited, but when I asked who slept where it quickly became clear that there wasn’t much space per person); clients reported eating meat “only when they could afford it.”
• VillageReach clients: infrastructure was much, much worse in these areas. The town of Macomia, where we spent the night, had no running water and no electricity except for generators; it took hours to reach (in a truck) from Pemba, which I believe was the closest area with reliable electricity and running water. The one village we visited took over an hour (of alert driving on very bad roads) to reach from Macomia, and the only concrete structures I saw there were the health center, a closed shop, and the school. I was told that other nearby villages were even harder to reach (in some cases impossible in a truck) and that access to water was a major problem. In terms of both standard of living and life opportunities, these areas appeared fundamentally worse than SEF areas.
• Soweto: I took a quick tour through a poor area of Soweto (urban). It was generally filthy (literally strewn with trash) and extremely crowded, with tiny steel shacks next to each other. It seemed to me like a much more unpleasant place to live than either of the other two areas, although on the flip side, people in Soweto appeared to have access to public transportation, electricity, good schools, etc. as they were very close to much wealthier residences.

One of the reasons Small Enterprise Foundation stood out to us is that it appears more diligent about targeting the poor than other organizations. Even so, its clients – while poor – appear to be substantially better off (in fundamental infrastructure-related ways, not ways that can be attributed to program effects) than VillageReach’s clients. This doesn’t make me less supportive of SEF (it’s largely consistent with my existing suspicion that microfinance clients are rarely if ever the poorest of the poor), but it’s an important thing to keep in mind that I feel better informed about now than before.

Are you looking to help people in the worst situation, and with the most basic needs, possible? Or are you interested in helping people who are better off to begin with, in the hopes that a little assistance might go a longer way with them? To me there’s no clear right answer, but it’s a decision donors are likely making constantly without knowing it.

More thoughts coming in Part III.

I previously posted “raw data” (pictures, audio, notes) from my recent visit to two of our top charities in Africa. The next few posts will give my thoughts from the trip.

First, a note on representativeness. I was only in Africa for two weeks; I was a complete outsider; I certainly don’t think that anything I saw “proves” anything about the programs or areas I was looking at. In many cases what I saw (and what I discussed with staff) prompted me to discuss and think harder about issues I’d already thought about a little. So as I share thoughts from the trip, think of these as thoughts that were partly inspired by what I saw and discussed, not as “things I’ve learned.”

In fact, I was hesitant to visit the field too early because I was afraid that I would form a vivid picture of how things work based on what I saw, and that it would be difficult to imagine how differently things could work in other settings (and even on other days). From this point on, I am definitely going to have a little trouble thinking about microfinance without picturing what I saw at Small Enterprise Foundation (SEF), for example. I think that when dealing with multinational charities that work in a huge variety of settings, it is best to get most of our information by reading the observations and analysis of others.

With that said, here are some thoughts.

I was impressed with the staff of the two nonprofits I visited.

In many of my conversations with nonprofit staff, I feel like I’m being sold a story, people are telling me what they think I want to hear, etc., which makes me instinctively somewhat distrustful. I can honestly say that I felt none of this during my interactions with SEF and VR staff, and that includes the lower-level staff. They were straightforward with me about challenges and concerns. Most acknowledged that there are reasons to worry about whether they’re being effective, and did not seem interested in downplaying concerns or exaggerating successes. And most seemed to me to be quite intelligent, knowledgeable, and reasonable about the work they were doing.

These two organizations had already been identified as outstanding before I visited, and I would have ranked them among the very best organizations in terms of “straightforward, no-nonsense interactions with staff” even before I went (the other nonprofits I’d put in this category are Against Malaria Foundation, Population Services International, and Stop Tuberculosis Partnership). However, it’s possible that people who are working in the field in program roles tend to be better (more direct) communicators with GiveWell than people in fundraising roles, and I’m very curious as to what impression I would have come away with if I had done a similar visit to a charity we have a lower opinion of.

Getting pictures, audio and video was not a problem.

People I spoke to never objected to being recorded and were usually (with some exceptions) happy to have their pictures taken, sometimes even insisting on it. Children particularly enjoyed being photographed (for example, see this video of me taking pictures as well as the photos from my trip to the village).

This surprised me somewhat (arguably it shouldn’t have) only because I feel like I’ve seen relatively little use of multimedia to monitor, evaluate, and report on programs. For example, I’ve been told many times that I “have to see a program in action” to be sold on its effectiveness; yet now I wonder why the charities that feel this way aren’t posting large amounts of real-time, unedited footage to give America-based donors as much of the experience as possible.

Charities do often produce heavily edited videos and photos, but we see little value in such productions as evidence (or as monitoring/evaluation tools) because it is so difficult to distinguish observation from editorial.

We’ve written before that we see a lot of potential value in “qualitative evidence” that is presented systematically and transparently, but we rarely see this happening.

Connecting with clients – culturally and even linguistically – appeared to be a fairly significant challenge, and the way I believe most nonprofits deal with it points to the importance of systematic monitoring and evaluation.

Any American charity working in the developing world ultimately has to connect people (donors and clients) who speak different languages and come from very different cultures. If the charity is even of moderate size (i.e., working in more than a few village, as even the relatively small charities I visited do), it also has to manage operations beyond what upper management can observe directly. It seems to me that the usual approach to this challenge is to have several degrees of separation between upper management and the people doing work in the field.

• The CEO of the Small Enterprise Foundation does not speak the local languages, and the COO (originally from Croatia) says he has learned to understand quite a bit but still cannot speak them. The lowest-level staff, development facilitators (similar to “loan officers”), tend to have similar backgrounds to the clients and to speak the local languages well, but this of course means that their background is very different from that of upper management and donors (note that the development facilitator I spoke with has very limited English). In in-between roles, there are some employees who are better able to “bridge the gap” (such as the staffer who translated for me on day two). Given this situation, it isn’t surprising that SEF’s management process is heavily dependent on systematic collection, auditing and analysis of key metrics (as I discussed with the COO).
• VillageReach doesn’t employ as many people, but it also has people in major roles who are American and need help from translators to communicate with local staff (during my visit, we had a translator traveling with us partly to help with communications between Leah, from the Seattle office, and Durao, a local VillageReach employee).
• I did see one instance of an alternate approach: literally sending Americans to live among clients, learn their languages, etc. This was the approach taken by the missionaries who helped us out when VillageReach’s vehicle broke down in Mozambique. However, my impression (which they confirmed) is that this approach (which I imagine presents problems and challenges of its own) is relatively unusual even among missionaries and is essentially unheard of among nonprofits focused on humanitarian aid.

None of the above observations should come as a surprise, but to me they highlight the importance of formal, systematic monitoring/auditing/evaluation. We often focus on the benefits of monitoring/evaluation for donors, but in situations like those described above it also seems like they are essential for conducting any kind of meaningful organizational management. I have trouble seeing how an organization that conducts no formal data collection and auditing can even run a program of any meaningful size. I would certainly be curious to see how operations work within some of the charities we have found to be less data-oriented.

I think it’s also important to note how difficult translation and communication can be. For example, during my first day with SEF, my communication with clients required several steps: I would ask a question in English, SEF’s CEO would rephrase it so that the development facilitator could understand it, the development facilitator would ask the question in the local language and relay the answer back in English, and finally the CEO had to rephrase the English again so I could follow it. On my second day, I was with someone who was very strong in both languages, but you can hear how much work he put into translating my fairly short and basic-seeming questions. He explained that, in addition to culture-based difficulties with translation, he was being very careful with wording because clients very much want to tell donors what they think the donors want to hear.

We have long felt that survey data is most useful for extremely concrete, factual questions: “What did you eat yesterday?” is more useful than “What do you normally eat?” is more useful than “Did this program help you?” is more useful than “How much did this program help you?” More on this idea at this post on Philanthropy Action (co-maintained by GiveWell Board member Tim Ogden).

More thoughts from the trip coming in Part II.