Archive for May, 2014

Bill Gates has a Project Syndicate piece up that, in the context of discussing Nina Munk’s book The Idealist, argues in favor of Jeffrey Sachs’ importance and relevance to contemporary development.

I’m going to leave aside the overarching argument of the piece. Instead, I want to focus on a small passage that, while perhaps a secondary point to Gates, strikes me as a very important lesson that he fails to apply to his own foundation (though to be fair, this is true of most people working in development).

Gates begins by noting that Sachs came to the Gates Foundation to ask for MVP funding, and lays out the fundamental MVP pitch for a “big push” of integrated interventions that crossed health, agriculture, and education sectors that Sachs was selling:

[Sachs’] hypothesis was that these interventions would be so synergistic that they would start a virtuous upward cycle and lift the villages out of poverty for good. He felt that if you focus just on fertilizer without also addressing health, or if you just go in and provide vaccinations without doing anything to help improve education, then progress won’t be sustained without an endless supply of aid.

This is nothing more than integrated development, and it makes sense. But, as was predicted, and as some are now demonstrating, it did not work. In reviewing what happened in the Millennium Villages that led them to come up short of expectations, Gates notes

MVP leaders encouraged farmers to switch to a series of new crops that were in demand in richer countries – and experts on the ground did a good job of helping farmers to produce good crop yields by using fertilizer, irrigation, and better seeds. But the MVP didn’t simultaneously invest in developing markets for these crops. According to Munk, “Pineapple couldn’t be exported after all, because the cost of transport was far too high. There was no market for ginger, apparently. And despite some early interest from buyers in Japan, no one wanted banana flour.” The farmers grew the crops, but the buyers didn’t come.

But then Gates seems to glide over a really key question: how could a smart, well-intentioned man miss the mark like this? Worse, how could a leading economist’s project blow market engagement so badly? Gates’ throwaway argument is “Of course, Sachs knows that it’s critical to understand market dynamics; he’s one of the world’s smartest economists. But in the villages Munk profiled, Sachs seems to be wearing blinders.” This is not an explanation for what happened, as telling us Sachs suffered from blinders is simply restating the obvious. The real issue is the source of these blinders.

The answer is, to me, blindingly obvious. The MVP, like most development interventions, really never understood what was going on in the villages targeted for intervention. Sure, they catalogued behaviors, activities, and outcomes…but there was never any serious investigation into the logic of observed behaviors. Instead, the MVP, like most development interventions, was rife with assumptions about the motivations of those living in Millennium Villages that produced these observed activities and outcomes, assumptions that had little to do with the actual logic of behavior. The result was interventions that implicitly infantilized the Millennium villagers by providing interventions that implicitly assumed, for example, that the villagers had not considered the potential markets for new and different crops/products. Such interventions assume ignorance as the driver of observed behaviors, instead of the enormously complex decision-making that underlies everyday lives and livelihoods in even the smallest village.

To give you an idea of what I mean, take a look at the following illustrations of the complexity of livelihoods decision-making (these are from my forthcoming article on applying the Livelihoods as Intimate Government approach in Applied Geography – a preprint is here).

First, we have #1, which illustrates the causes behind observed decisions captured by most livelihoods frameworks. In short, this is what most contemporary development planning gets to, at best.

Figure 1

However, this is a very incomplete version of any individual’s decision-making reality. #2 illustrates the wider range of factors shaping observed decisions that become visible through multiscalar analysis that nests particular places in wider networks of economic, environment, and politics. Relatively few applications of livelihoods frameworks approach this level of complexity, and those that do tend to consider the impacts of markets on particular livelihoods and places.

Figure 2

While this is better than the overly-simplistic framing of decisions in #1, it is still incomplete because motivations are not, themselves, discrete. #3 illustrates the complex web of factors, local and extralocal, and the ways in which these factors play off of one another at multiple scales, different times, and in different situations.

Figure 3

When we seek to understand why people do what they do (and do not do other things), this is the complexity with which we must engage.

This is important, because were Gates to realize that this was the relevant point of both Munk’s book and his own op-ed, he might better understand why his own foundation has

many projects…that have come up short. It’s hard to deliver effective solutions, even when you plan for every potential contingency and unintended consequence. There is a natural tendency in almost any kind of investment – business, philanthropic, or otherwise – to double down in the face of difficulty. I’ve done it, and I think most other people have too.

So, what do you do? Well, we have an answer: The Livelihoods as Intimate Government approach we use at HURDL (publications here and here, with guidance documents coming later in the summer) charts an analytic path through this level of complexity. Before the usual objections start.

1) We can train people to do it (we are doing so in Mali as I write this). You don’t need a Ph.D. in anthropology to use our approach.

2) It does not take too much time. We can implement at least as fast as any survey process, and depending on spatial focus and resources, can move on a timeframe from weeks to two months.

3) It is not too expensive – qualitative researchers are not expensive, and we do not require high-end equipment to do our work.

The proof is in the reactions we are getting from our colleagues. Here in Mali, I now have colleagues from IER and agricultural extension getting fired up about our approach as they watch the data coming in during our pilot phase. They are stunned by how much data we can collect in a short period of time, and how relevant the data is to the questions at hand because we understand what people are already doing, and why they are doing it. By using this approach, and starting from the assumption that we must understand what people are doing and why before we move to interventions, we are going to lay the foundation for more productive interventions that minimize the sorts of “surprise” outcomes that Gates references as an explanation for project failure.

There are no more excuses for program and project design processes that employ the same limited methods and work from the same problematic assumptions – there are ways to do it differently. But until people like Gates and Sachs reframe their understanding of how development should work, development will continue to be plagued by surprises that aren’t all that surprising.

While development – thought broadly as social/economic/political change that somehow brings about a change in peoples’ quality of life – generally entails changes in behavior, conversations about “behavior change” in development obscure important political and ethical issues around this subject, putting development programs and projects, and worse the people those programs and projects are meant to help, at risk.

We need to return to a long standing conversation about who gets to decide what behaviors need changing.  Most contemporary conversations about behavior change invoke simple public health examples that obscure the politics of behavior change (such as this recent New York Times Opinionator Piece). This piece appears to address the community and household politics of change (via peer pressure), but completely ignores the fact that every intervention mentioned was introduced by someone outside these communities. This is easy to ignore because handwashing or the use of chlorine in drinking water clearly reduces morbidity, nobody benefits from such morbidity, and addressing the causes of that morbidity requires interventions that engage knowledge and technology that, while well-established, were created someplace else.

But if we open up this conversation to other sorts of examples, the picture gets much more complicated. Take, for example, agricultural behaviors. An awful lot of food security/agricultural development programming these days discusses behavior change, ranging from what crops are grown to how farmers engage with markets. Here, the benefits of this behavior change are less clear, and less evenly-distributed through the population. Who decides what should be grown, and on what basis? Is improved yield or increased incomes enough justification to “change behaviors”? Such arguments presume shockingly simple rationales for observed behaviors, such as yield maximization, and often implicitly assume that peasant farmers in the Global South lack information and understandings that would produce such yields, thus requiring “education” to make better decisions. As I’ve argued time and again, and demonstrated empirically several times, most livelihoods decisions are a complex mix of politics, local environment, economy, and social issues that these farmers weigh in the context of surprisingly detailed information (see this post or my book for a discussion of farm allocation in Ghanaian households that illustrates this point). In short, when we start to talk about changing peoples’ behaviors, we often have no idea what it is that we are changing.

The fact we have little to no understanding of the basis on which observed decisions are made is a big, totally undiscussed problem for anyone interested in behavior change. In development, we design programs and projects based on presumptions about people’s motivations, but those presumptions are usually anchored in our own experiences and perceptions – which are quite often different from those with whom we work in the Global South (see the discussion of WEIRD data in psychology, for example here). When we don’t know why people are doing the things we do, we cannot understand the opportunities and challenges that come with those activities/behaviors. This allows an unexamined bias against the intelligence and experience of the global poor to enter and lurk behind this conversation.

Such bias isn’t just politically/ethically problematic – it risks real programmatic disasters. For example, when we perceive “inefficiency” on many African farms, we are often misinterpreting hedging behaviors necessary to manage worst-case scenarios in a setting where there are no safety nets. Erasing such behaviors in the name of efficiency (which will increase yields or incomes) can produce better outcomes…until the situation against which the farmers were hedged arises. Then, without the hedge, all hell can break loose. Among the rural agricultural communities in which I have been working for more than 15 years, such hedges typically address climate and/or market variability, which produce extremes at frequent, if irregular, intervals. Stripping the hedges from these systems presumes that the good years will at least compensate for the bad…a dangerous assumption based far more on hope or optimism than evidence in most places where these projects are designed and implemented. James Scott’s book The Art of Not Being Governed provides examples of agrarian populations that fled the state in the face of “modernization” efforts not because they were foolish or backward, but because they saw such programs as introducing unacceptable risks into their lives (see also this post for a similar discussion in the context of food security).

This is why my lab uses an approach (on a number of projects ranging from climate services evaluation and design to disaster risk reduction) that starts from the other direction – we begin by identifying and explaining particular behaviors relevant to the challenge, issue, or intervention at hand, and then start thinking about what kinds of behavioral change are possible and acceptable to the people with whom we work. We believe that this is both more effective (as we actually identify the rationales for observed behaviors before intervening) and safer (as we are less likely to design/condone interventions that increase vulnerability) than development programming based on presumption.

This is not to say that we should simply valorize all existing behaviors in the Global South. There are inefficiencies out there that could be reduced. There are things like handwashing that are simple and important. Sometimes farmers can change their practices in small ways that do not entail big shifts in risk or vulnerability. Our approach to project design and assessment helps to identify just such situations. But on the whole, we need to think much more critically about what we are assuming when we insist on a particular behavior change, and then replace those assumptions with information. Until we do, behavior change discussions will run the risk of uncritically imposing external values and assumptions on otherwise coherent systems, producing greater risk and vulnerability than existed before. Nobody could call that development.