Thursday, May 05, 2011

More than good intentions

Suppose you are walking down a street by a lake on your way to a meeting, and if you miss the meeting you will lose two hundred dollars. You see a child drowning in the lake. Do you have an ethical obligation to stop and jump in and save the child, even though it will cost you two hundred dollars? Most people say yes.

Don't you then also have an ethical obligation to send two hundred dollars right now to one of many organizations delivering aid to the poor, where it can save a child's life? Most people say no, or at least, they don't cut that check.

Linkchildren in an IDP camp in Northern Uganda, 2007

This example comes from Peter Singer, a utilitarian philosopher at Princeton. Dean Karlan and Jacob Appel use the hypothetical scenario to introduce the central dilemma of their new book, More than Good Intentions. The idea makes most of us uncomfortable - sending a check off to an organization requires a great deal of trust - how do we know that the organization is actually going to save the child? There's a big leap between the personal rescue and the checkbook.

When I first traveled to Uganda in 2006, I had the vague notion that I wanted to work in international development. I arrived in Gulu for the first time and my jaw dropped at the roundabouts lined with signs for NGOs - psychosocial support, HIV/AIDS work, conflict resolution, infrastructure, agricultural programs, demobilization, reintegration of child soldiers - the list goes on. I was moved by what I saw. I was appalled by the poverty, the conditions at IDP camps, the blank gaze of an 11-year-old boy who had just escaped captivity. I was deeply saddened, and felt a strong sense of responsibility (like the "walking-by-the-child-drowning-in-the-lake" example).

walking by the lake, sanyu babies home, kampala 2006

The super-saturation of NGOs and donors seemed positive. Everyone was passionate about helping Northern Uganda, and hundreds of millions of dollars in aid were pouring from the US, Europeans, and other donors. It was like a giant do-gooder party.

When I actually started working in Northern Uganda, I quickly became one of the cynics. Once a champion of Jeffrey Sachs, I started to think along the lines of Dambisa Moyo - who in Dead Aid posits that aid is actually making things worse. There are very clear reasons why I started One Mango Tree as a business. It created jobs and income for people in Northern Uganda. There's nothing muddy about that, and the impacts are obvious. I wasn't as interested in the trial-and-error (and keep funding even if it fails) approach that development seemed to take.

Randomized Control Trials: Asking the Right Question

In More than Good Intentions, Karlan and Appel take a step back from the aid debate and suggest that measurement is the only clear way forward. Serious testing and analysis can tell us what works and what doesn't, and if we put efforts only into what works (and abandon what doesn't), then aid really should make a difference, right? They dive right into this possibility, using randomized control trials (RCTs) to answer some big questions:

Why would a perfect loan candidate choose not to take a loan?
When you loan someone money for something specific (and urgent!), why do they buy a TV?
Why don't people use bed nets? Will they use them if they're free?
Why won't people pay for chlorine tablets for clean drinking water?
Why do people go to herbalists and avoid hospitals?

The answers are behavioral, and Karlan and Appel argue that most development programs operate on classical economic assumptions - leaving very important behavioral factors out of the equation. RCTs provide answers to the questions of why people do what they do. By surveying both individuals receiving the program and a control group who does not, researchers can get to the bottom of what works. Simple before-after analysis is not enough.

The Takeaways

The implications of Karlan and Appel's work are clear - support research to back up effectiveness of aid programs. If you're going to spend aid money, then spend money on a system to monitor its effectiveness - and do it right (I supposed this is where the ever-present M&E comes in). And if it ain't workin, then STOP funding it.

One Mango Tree's Village Savings & Loan Association

For a business venture, One Mango Tree does a lot of research. We collected baseline surveys for all of our staff, and re-visit the data each year. We collect cost information to re-assess our wages annually - can our staff still support their households on their One Mango Tree earnings? More than Good Intentions made me wonder if we could do more.
  • We could survey tailors in the local marketplace as a control group to see how our staff are faring compared to tailors in the market
  • We've come across problems where women stop coming to work once they earn about $100 in any period - RCT could help us get to the bottom of this
  • We could apply the research on savings - working within the One Mango Tree VSLA structure to help women set goals and restrict them from withdrawals until they reach the goal amount
Cutting a $200 check to save a child isn't enough. If we're to take poverty and development seriously, then we need rigorous testing to find out what works pump funding to those areas. A few listed by Karlan and Appel:
  • Microsavings (over microcredit)
  • Reminders to save - to combat messaging to buy
  • Prepaid fertilizer sales
  • Deworming
  • Chlorine dispensers for clean water
  • Remedial education in small groups
  • Commitment devices

No comments: