At least since former World Bank President Jim Wolfensohn coined the phrase “knowledge bank,” there have been periodic efforts to strengthen evidence-based policymaking at the World Bank. They have focused overwhelmingly on the supply of knowledge–with a steady stream of “flagship reports.” The World Bank has invested in better data, more rigorous research, systematic reviews, impact evaluations, and increasingly sophisticated analytics to inform its operations. The most recent reorganization aims to create a “knowledge bank.”
Yet previous reorganizations and rhetoric have not consistently translated into improved research quality or greater development impact. High-quality evidence often fails to shape policy choices, lending priorities, or institutional reforms in low- and middle-income countries—or even within the bank itself.
What is missing from this conversation is the demand for knowledge.
Evidence-based policy does not emerge simply because good evidence exists. It emerges when institutions are structured so that decision-makers (1) want to know, and (2) are rewarded for using knowledge. The history of places like Bell Labs illustrates that insight production depends at least as much on institutional demand for understanding as on the technical ability to generate it.
Lessons from Bell Labs: Why knowledge thrives where it is needed
How did institutions like Bell Labs consistently produce transformative knowledge? The answer is not that they hired unusually brilliant people or funded research indiscriminately. It is that Bell Labs sat inside an organization—AT&T—that needed deep understanding to solve real, operational problems and was willing to tolerate long horizons, uncertainty, and intellectual risk to get it.
Knowledge production must be disciplined by demand: research must matter because it feeds into decisions that senior leaders care about. And, crucially, researchers must be rewarded not for producing board-approved projects, but for achieving measurable and reasonable outcomes that matter in the client country.
This contrasts sharply with how evidence often functions in development institutions. At the World Bank, knowledge is frequently treated as an input into lending rather than as a driver of strategy. Analytical work is produced, but its uptake is ad hoc: often incidental, contingent, or politically constrained.
The result is what we might call the “knowledge paradox”: external stakeholders and clients consistently report valuing the World Bank’s knowledge more than its finance, yet internal incentives overwhelmingly favor lending volumes, project preparation, and disbursement.
Supply is not the constraint
Given its size, the World Bank likely has an absolute advantage in development knowledge relative to most other institutions, including universities and think tanks. The problem, thus, is not necessarily a lack of data, methods, or intellectual capacity.
Instead, the constraint lies in how knowledge is demanded, selected, and used.
Three features stand out.
First, in budgeting and staffing decisions, knowledge work is largely residual. Lending preparation and supervision come first; whatever remains goes to analytics. This sends a clear signal about what the institution values.
Second, quality control and contestability are uneven. When analytical work is closely tied to lending or government relations, there are strong incentives—often subconscious—to avoid inconvenient findings, controversial topics, or politically sensitive conclusions. If a piece of research calls into question some aspect of the lending program, it is scrutinized, criticized, and sometimes even suppressed.
Third, much of the World Bank’s knowledge is aimed exclusively at governments, even when the binding constraint is political rather than technical (i.e., the government is the problem). In such cases, evidence that remains confidential or narrowly circulated cannot shift incentives, build coalitions, or empower citizens.
These patterns reflect a deeper issue: knowledge demand is weakly articulated. There is little systematic process asking: what do we most need to know to relax the binding constraints to development in this country? Or, which pieces of evidence, if widely understood, could change political incentives?
Government failure and the political economy of knowledge
Several of today’s development challenges are primarily problems of government failure. Energy subsidies, weak service delivery, regulatory capture, and low state capacity persist not because policymakers lack technical understanding, but because political incentives reward the status quo.
In this context, the demand for knowledge must extend beyond technocrats. Evidence needs to inform public debate, media scrutiny, and citizen accountability. Transparency, benchmarking, and comparative data—when designed well—can shift norms and expectations.
Yet this role for knowledge sits uneasily within the bank’s operational model. Publishing politically salient analysis can strain country relationships. Lending incentives can discourage confrontation. As a result, potentially transformative knowledge can be softened, reframed, delayed, or never produced at all.
This is not a supply failure. It is a failure to create institutional demand for uncomfortable knowledge.
Toward a demand-driven Knowledge Bank
What would it mean for the World Bank to take demand seriously?
First, the bank would start with questions, not products. Country strategies would explicitly identify the most consequential policy and institutional distortions—and commission knowledge work that would help reform them, regardless of political comfort.
Second, in contrast to the usual rhetoric that the value of the World Bank is that it “bundles” knowledge and finance, knowledge would increasingly be decoupled from lending, especially where government failure is the binding constraint. Bundling knowledge and finance compromises the quality of each. The knowledge work is constrained by the lending program; and lending does not take advantage of all the knowledge that is available, inside and outside the bank. Having an independent research group that is protected by strong quality control and publication norms and remains independently steered would enhance credibility and impact.
Third, dissemination would be treated as part of the core task, not an afterthought. If evidence is meant to shift incentives, it must be intelligible, accessible, and contestable by non-experts.
Finally, internal incentives must change. Staff who produce influential knowledge—whether or not it leads directly to a project—should be rewarded. Managers should be evaluated on whether they achieve impact, not just whether they move money.
Wanting to know more
The World Bank does not suffer from a lack of evidence. It suffers from a lack of institutional hunger for knowledge that challenges, disciplines, and occasionally disrupts.
Bell Labs thrived because understanding was indispensable to action. If the World Bank is serious about being a Knowledge Bank, it must cultivate the same condition: not just the ability to produce evidence, but the necessity of using it.
Until then, improving the supply of knowledge will continue to yield low returns. The harder—and more important—task is to build demand.
This blog has been reprinted from CGD Blog with permission from the authors.
*About the authors:
- Shanta Devarajan is Professor of Practice at Georgetown University and Non-Resident Fellow at the Center for Global Develeopment. He is Chair of the Economics thematic group of the 1818 Society.
- Eeshani Kandpal is a senior fellow and research manager at the Center for Global Development. Prior to this, she was a Senior Economist in the Development Research Group of the World Bank.
Disclaimer
Member’s blog posts reflect the views of the author(s), drawing on prior research or personal experience. Freedom of expression is an essential part of the 1818 Society’s culture. The 1818 Society® is a nonpartisan, independent organization and does not take institutional positions.
KEYWORDS knowledge, Knowledge Bank, World Bank
