This is a crosspost from Consultative Group to Assist the Poor (CGAP).
There was plenty of encouraging information shared at the recent “Reaching the Poorest 2012” meeting, convened by CGAP and the Ford Foundation. Together with my fellow researchers, I was among the panelists who presented the findings of well over five years worth of randomized control trials evaluating the impact of the Graduation Model. The projects that were evaluated in Bangladesh, Pakistan, Honduras, and India showed an impact on the livelihoods of the poorest that were targeted. The results were mostly heartening – they showed that Graduation Program participants typically improved their food security, stabilized and diversified income, and increased their assets.
The benefits we’re seeing in the lives of the poorest are big and important. The results are strong evidence that the Graduation Model can work. (We’ll know even more in a couple years when we have full results from seven pilots, with more sites and longer term results to see if results sustain themselves.) One of the most intriguing, and I believe important, results is the simplest: happiness went up in the two sites where “happiness” was measured (Honduras and West Bengal). As part of the surveys to measure the impact of the program on their livelihoods, participants were also asked a series of questions on their general level of happiness and mental health. Often we talk about consumption and income as a measure of wellbeing.
But sometimes the story is more complicated: income may go up, but leisure activity down. Consumption may go up now, but at the expense of future consumption (e.g., through borrowing). This makes it more complicated to assess welfare, without simply assuming that individuals choose what is best for them, so a change is by definition good as long as the prior choice was still available to the participant.
“Happiness” gets at the issue more directly, albeit in a less tangible way (e.g. we show people five simple drawings of faces, from very sad to very happy, and ask them to identify which one corresponds to their overall personal satisfaction, and we also ask more specific yet still quite broad questions, such as whether they never, sometimes, or often cry or feel sad).
But the program did not always work. Not everything came up roses. Some programs (Pakistan and West Bengal) had more impact than others (Honduras), and some had little to no impact (Andhra Pradesh), compared to their respective control groups. But there is no such thing as a failed pilot, unless one chooses to ignore the results rather than draw learning from them.
The upside to a rigorous evaluation is that even the failures in terms of lack of impact are winners if properly documented. We then learn something about what not to do. But that is only if we can learn something about why it didn’t work. Was it the underlying context, or something about the implementation? And likewise even the biggest successes in terms of tons of impact are not winners, if we do not learn why they worked.
Although a simple “it worked” answer could tell us to scale up in that context, we would be a bit tongue-tied in prescribing the solution to other settings without some more information about why it was working.
Furthermore, simply knowing “it worked, and here are some important contextual conditions” may tell us when and where to scale-up, but would miss out on important questions: was this the most cost-effective solution to the problem? For that, one needs to build in tests from the beginning that help tease out the relevant importance of different components. What does 1+1 equal? It is not hard to put forward reasonable theories and hypotheses in which it equals 1 (two solutions crowd each other out), 2 (two solutions have nothing to do with each other, so their impact is additive) or 3 (two solutions have synergies, thus when done together are even better than the sum of the parts).
The Ghana pilot is focused on questions just like these: how much bang for one’s buck do we get merely by transferring assets? How important is the savings component?
Maybe all of the components of the graduation model are necessary. Maybe the big change occurs when people gain hope, and for that the full set is needed. If part of being in the poverty trap is not seeing a pathway out of poverty, then a holistic approach like the one taken by the Graduation Program might be doing far more than a series of sequenced inputs and services – it gives the poorest the opportunity to develop the self-confidence needed to take control of their lives. We hope further research will help unpack this black box a bit more. Stay tuned!