Anti-poverty programs must often start with a rather obvious step: finding poor people. This may seem like a simple exercise, particularly in countries with high poverty rates. However, consider the potential pitfalls of some common approaches: a simple asset test (questions like “do you have a bicycle? A cow? A radio?”) may be easy to game; a complex survey tool assessing consumption may yield accurate data, but be prohibitively expensive; receiving input from local leaders may result in cronyism. Ultimately, governments, NGOs, and donors must balance multiple objectives – accuracy, cost, enforceability, and community perception – when formulating pro-poor targeting criteria.
At GiveDirectly, we have identified the majority of our recipients by asking a simple question: is his/her home made out of organic materials (e.g., mud and thatch)? Data suggest that housing materials are a strong proxy for poverty in these geographies; housing is less vulnerable to manipulation and can be easily verified; and community members understand the criteria (and have, in fact, suggested it independently during focus groups).
Despite all of these advantages, we launched a project earlier this year to experiment with new targeting techniques for the simple reason that organic housing will not be an appropriate targeting criterion in all settings. With the objective of increasing our capacity to rapidly identify poor recipients in any setting, we have initiated the search for more widely applicable (“universal”) targeting methods.
So far, these include: 1) having our staff rate photographs of recipients and their homes on a poverty scale; 2) using simple points-based surveys focused on verifiable criteria (e.g., asking whether someone is a widow or has a toilet); and 3) experimenting with various forms of community-based targeting (i.e. communities nominating their own members in an open meeting facilitated by our staff).
We will be modifying aspects our standard enrollment process over the next few months to pilot each of these methods, while assessing their cost, accuracy, and perceived fairness. We’ve already run into some interesting operational challenges: organizing community nomination meetings in areas that are strongly divided along clan lines and figuring out how to prevent staff from ranking a disproportionate number of households as “very poor.”
We look forward to solving these problems and keeping you updated on our progress.