A think tank is an organisation that researches public policy and tries to influence it. Think tanks publish reports, brief officials, testify before legislatures, hold events and appear in the media. Thousands exist worldwide, from global names such as RAND, Brookings and Chatham House to small specialist institutes with a handful of researchers. They also include government-affiliated research centres in China, Russia and elsewhere.
Their influence is real but hard to measure, and their independence is often questioned. This guide explains how think tanks work, which models exist and what research says about what makes them effective.
Four models
| Model | Typical funding | Examples | Strength / risk |
|---|---|---|---|
| Academic (“universities without students”) | Endowments, foundations, donations | Brookings, Chatham House | Depth and credibility / can be slow and distant from decisions |
| Contract research | Government and agency contracts | RAND | Rigorous, decision-focused / agenda set partly by clients |
| Advocacy | Donors aligned with a viewpoint | Many US policy institutes, on all sides | Fast, effective in debate / perceived as partisan |
| Party or government-affiliated | State or party budgets | German party foundations, Chinese government and Party-school think tanks | Direct access / limited independence |
Most real organisations mix these models. RAND, founded in 1948 out of a US Air Force project, set the template for contract research. It brought systems analysis, game theory and cost-effectiveness methods into policy. Its techniques, such as the Delphi method for structured expert forecasting, spread far beyond defence.
China’s “new-type think tanks”
In 2015 the Chinese government issued guidelines on building new-type think tanks with Chinese characteristics (中国特色新型智库). It then designated a group of national high-end pilot think tanks, drawn from government research centres, academies, universities, the military, Party schools and a few enterprises and social organisations. The policy encourages think tanks to support decision-making through internal reports (内参), policy evaluation and public diplomacy.
Chinese research on think tanks often asks questions familiar elsewhere. How independent should a think tank be from government? How should policy research products be “traded” between researchers and officials? Why do some think tanks fail to deliver useful advice? A recurring answer is that closeness to power gives access but risks telling decision-makers what they want to hear.
How think tanks influence policy
Influence rarely comes from a single report changing a decision. More often it works through several channels:
- Agenda setting. Making an issue visible and framing how it is discussed.
- Ideas in waiting. Developing proposals that are ready when a crisis or a new government creates a “policy window”.
- People. Staff move between think tanks and government (the “revolving door”), carrying ideas and networks with them.
- Evidence and evaluation. Providing data, models and assessments that officials lack the time or capacity to produce.
- Convening. Bringing together people who otherwise would not meet, often in confidential settings such as off-the-record discussions.
- Second-track diplomacy. Unofficial dialogues between experts from countries whose governments are not talking.
Measuring influence
Measuring think tank influence is notoriously difficult, because policy outcomes have many causes. Common proxies include media mentions, citations in legislative and government documents, testimony, web traffic, social media reach and academic citations. Each proxy is incomplete. Media visibility rewards advocacy think tanks, and academic citations reward academic ones.
The best-known ranking, the University of Pennsylvania’s Global Go To Think Tank Index, relied largely on peer nominations. It was criticised for being a reputation survey rather than a measure of impact, and it stopped after its 2020 edition. Evaluation now leans towards case studies: tracing how a specific idea moved from a report into a law or programme, and what else contributed.
Why advice fails
Research on failed policy advice identifies a handful of recurring causes:
- Capture. Funders, clients or political patrons shape conclusions, explicitly or through self-censorship.
- Scientific limits. Overconfidence in models, cherry-picked evidence, or recommendations that go beyond what the evidence supports.
- Timing and format. Excellent analysis that arrives after the decision, or that is too long for a busy official to read.
- Groupthink. Organisations that hire people with similar views and rarely test alternative hypotheses.
The remedies resemble good intelligence practice: separate evidence from recommendations, state uncertainty, use peer review and red teams, and be transparent about methods.
Transparency and independence
Because think tanks seek to influence public policy, their funding matters. Rating initiatives such as Transparify have scored think tanks on whether they disclose who funds them and how much. Funding transparency does not guarantee independence, but its absence makes it impossible for readers to judge potential conflicts of interest. Good practice includes publishing donors above a threshold, having clear policies on donor influence over research, and disclosing when a report was commissioned.
Think tanks and AI
Generative AI affects think tanks in two ways. It can speed up literature reviews, data analysis and drafting, which helps small institutes compete. It also changes how their work is found. Officials and journalists increasingly ask AI assistants for summaries rather than reading reports. Think tanks whose work is clearly structured, openly accessible and well cited are more likely to show up in those answers. Opaque PDFs are less likely to.