The World Bank has committed itself to twin goals: eliminating extreme poverty by 2030 and boosting shared prosperity, measured as the income of the bottom 40 percent in any given country. This recently inspired a post by Nancy Birdsall arguing that median income would be a better measure of shared prosperity, and another post by Lant Pritchett arguing that the extreme poverty goal is too narrow, which sparked comments by Martin Ravallion and others.
My view on those intriguing issues is that the train has already left the station. The question of what the goals are has been settled, and the question we are now pondering within the World Bank is what it means to “operationalize the goals.” We understand that projects should be prioritized in terms of how much they contribute to these goals. But how?
I see two ingredients to thinking through this question. First, what is the expected causal chain for how a particular project will contribute to the goals? Second, what is the evidence for the links on that chain?
In approaching these questions, it’s useful to consider a typology of projects. Here is one simple possible typology, in terms of the causal chain of impact on the extreme poor or bottom 40 percent:
- Direct impact
These projects reduce extreme poverty or boost the bottom 40 percent by directly targeting individuals or households. The clearest case in this category is a cash transfer.
- Growth impact
At the other end of the spectrum are projects that are thought to raise overall economic growth, which may pull people out of extreme poverty and boost the income of the bottom 40 percent. An example would include technical assistance to improve the business environment.
- Intermediate impact
In between the two extremes are projects which target neither individuals nor the economy as a whole. Localized infrastructure projects would fall in this category.
For “direct impact” projects, the causal chain is short and it’s clear the kind of evidence that could be brought to bear—studies based on household or individual surveys.
For “growth impact” projects, we can first ask what is the evidence that a particular project will boost economic growth. Next we can consider the evidence that growth will reduce extreme poverty or boost the bottom 40 percent, where the particular country’s track record could be taken as a rough guide.
The thornier cases are those in the “intermediate impact” bin, for which the causal chain is often longer. For example, consider a highway project in a major city where few bottom 40 percenters live. The project can be expected to boost the city’s growth, which will boost nationwide growth through many channels.
A concern I have heard repeatedly about trying to operationalize the goals is “you can tell a nice story about any project.” This is exactly why it’s crucial that the thinking be focused on evidence. Many possible kinds of evidence could be used across a range of rigor including studies based on experimental and quasi-experimental designs where feasible, cross-country growth regressions, case studies, synthetic controls, and even economic theory.
If the evidence is weak that a particular project will help achieve the goals, we should ask ourselves if the project really should be a priority for the World Bank. Certainly, we want to encourage innovative approaches—which will necessarily include those without firm evidence. But in those cases, the proposed project should include a plan to generate evidence on its impacts, so that we fully exploit the opportunity to learn.
All of this is very much an open question in the World Bank. What are your views on how the Bank should operationalize the goals?