Wholesaling Research for Development


This page in:

“Give a man a fish and he will eat for a day. Teach a man to fish and he will eat for a lifetime.” K’ung Fu-tzu (Confucius) , circa 500 BC.

   Photo: istockphoto.com
The World Bank’s analytic work can lack transparency to users—not least for those who would be affected most by the policies derived from that work. Civil society groups often suspect that the Bank dresses up advocacy as analytics. This perception stems in no small measure from the large entry costs users face in replicating and understanding the analysis. 

This concern about how we do research at the Bank—and elsewhere—can be thought of as the outcome of a traditional “retailing model.” That means researchers investigate a specific issue over a period of time and produce a research product—a paper or volume—on their findings. This is then disseminated to the public, including other researchers and policy makers. 

Much has changed since this retailing model first emerged. Dramatic changes in information technology have expanded the use of data, which makes more openness in research feasible.

It is time, then, to think about a new “wholesaling model,” under which the emphasis switches to producing the tools for others to do the research and providing open access to those tools.

There are three objectives for such an initiative. The first is to empower researchers in developing countries to do better research to inform development policy and development practice. This changes the focus of the traditional “capacity building” model from the task of “teaching the lessons from past research” to facilitating new learning in specific contexts.

Success in this endeavor will also help in the longer-term goal of expanding opportunities for a “collaborative retailing model,” under which Bank staff and academics in rich countries can work more closely with colleagues in developing countries as full peers, to the benefit of both.

The third goal is to help ensure more open and transparent policy analysis. The discipline of policy analysis—naturally dominated by economics—is increasingly sophisticated in its theories and methods. There are still long lags between the introduction of new theories and methods and their application to real-world problems. The Bank can play an important role in reducing the costs of understanding even the most sophisticated policy analysis, given that technical capabilities have increased among key stakeholders.

Through innovative uses of technology, the Bank is particularly well positioned to help increase access to data, information and experience. That way, it can better inform development policy debates and better engage direct stakeholders in those debates.
The Bank will need to become more deeply involved in producing tools to facilitate better development analysis and policy making by non-Bank researchers. These tools include public access development data, in particular survey data on households and enterprises, and surveys designs that are geared towards addressing the most important development issues. Importantly, the tools also include software applications for accessing and using existing data.

There are already some examples that point the way forward. The data tool PovcalNet is a case in point. This is an interactive software platform that provides free access to the data derived from (currently) 700 household surveys, allowing users to replicate the Bank’s global poverty counts and make their own estimates under different assumptions, such as alternative poverty lines. Before PovcalNet,  users of the Bank’s estimates of the number of people living below $1 a day ($1.25 a day at 2005 purchasing-power parity) had to take the Bank’s numbers as given, because they could not check the calculations or try different assumptions. That has changed radically.

Another example can be found in an important new product of the Bank’s research department, ADePT. This is an innovative software program designed to simplify and speed up the production of standardized tables and graphs in many areas of economic analysis, focusing on the Bank’s analytic work at country and regional levels. ADePT is a free, stand-alone program, available for download to anyone in the world.

The retailing and wholesaling models are complements, not substitutes. Useful tools for research can best be developed by researchers in the practice of solving real-world problems and writing papers on the results. This is also essential for quality assurance, since the retailed research products must pass critical peer review. And close connectivity between retailing and wholesaling functions helps ensure that the tools conform to best practice from  relevant technical literatures. Wholesale research tools of wide relevance are most likely to be produced and maintained by researchers who are themselves actively engaged in analysis that utilizes these data.

There are also a number of differences between the two models. Tool development is typically a longer-term effort, for which continuity is key to success. The work invariably involves teams, rather than individuals. Software development also requires a sufficiently large user base, which takes times to develop. It also requires constant effort in supporting, improving and updating, without which the software tools will die. It also requires support in dissemination, marketing, training and (crucially) user support.

Looking forward, the Bank’s researchers will continue to have a crucial role in developing the ideas and methods that are needed for better development policy making. Looking forward, the retailing model will remain central to the effort of the Bank’s research department.

However, we will also be exploring new ways to wholesale our ideas and methods related to the changing world we work in. In so doing, we will expand opportunities for fuller partnerships with experts in the developing world.

This new model for how we do research will combine open access to data with open access to the analytic tools used to inform policy discussions using those data. Our vision is that data, the knowledge and the solutions to development problems will ultimately be generated collaboratively by those who have most to gain from the success of those solutions.



Martin Ravallion

Martin Ravallion, Edmond D. Villani Professor of Economics, Georgetown University

Join the Conversation

Shanta Devarajan
September 30, 2010

Martin: Much as I was happy to read about the "wholesaling" model of development research, I was wondering if the metaphor of wholesale and retail trade is appropriate for development research today. The metaphor assumes that research is a good that is produced by one group of people (wholesalers and retailers) and "sold" to another group, the consumers. Yet as your blog post points out, the purpose is to influence policy that will improve the lives of poor people. How are these policies determined? It's rarely the case that a single piece of research, be it wholesaled or retailed, results in a change in policy--for good reason, because the person implementing the policy (a politician) has to answer to his or her voters, not to researchers. If politicians listen to their voters, then research should be accessible to ordinary citizens, and not just to sophisticated technocrats. In addition to disseminating research better, we should consider involving these ordinary citizens in the research itself. After all, they are the data points of our research. Why not make them part of the research process? NGOs like Pratham and Uwezo involve villagers and communities in collecting data on education outcomes. Using technology (as you suggest), we could put a nascent idea for research out to the public for discussion, and let people see (and comment on) the research as it is being done (as David Roodman at CGD is doing with his book on microfinance). We could even use cell phones and other means to learn from poor people what the important research questions are. If poor people are more involved in the process of research, they are more likely to "own" it, and thereby hold politicians more accountable for policies that are consistent with research. We need a new metaphor that goes beyond looking at research as something that is "produced" by one party and "consumed" by another.

Martin Ravallion
September 30, 2010


I think that the wholesaling research initiative is broadly consistent with your perspective. The idea is to get our researchers involved in producing the data and analytic tools needed to greatly expand the number of people in developing countries who are equipped to do their own research on key topics relevant to development. Yes, we will remain producers of research (and consumers as well). This kind of tool development is highly skill-intensive, and synergistic with many of the skills needed in the traditional retail model. We also need to continually make sure that our efforts are being tailored to the right consumers, to assure impact on development. (I like your ideas on using technology to create feedback loops, to help us keep tabs on the most pressing knowledge gaps.) But the key point here is that by a re-balancing our research efforts toward accessible tool development we will be able to turn the consumers of our products into producers themselves--producers of retail and ultimately wholesale products. That is what "wholesaling research" is all about.


Eduardo Ley
September 30, 2010

The re-balancing is most welcome. There is a dire need for accessible tools allowing clients to assess and evaluate policies in sophisticatedly simple models. Tools must be transparent so that one can understand & explain what is driving results.

Peter Goldstein
October 01, 2010

Martin: I applaud this effort, as our AudienceScapes project (www.audienescapes.org) at InterMedia is taking a similar approach of both doing our own analysis of survey-based data and also making this data available and collaborating with in-country analysts.

I am curious whether you have thought about how you might eventually measure the impact of your new model?