C
CPPPS 中国综合评价指数
China Panoramic Public Performance Score
中国榜 · 智库 China · Think Tanks

智库评价Think Tank Evaluation

规模 + 质量 + 创新 + 贡献 四维对数加权模型 · 全国主要智库 24 家 · 排名更新日期：2026-10-20Four-dimensional log-weighted model (Scale + Quality + Innovation + Contribution) · 24 major national think tanks · Last updated: 2026-10-20

综述Overview

A think tank is a carrier of national soft power in the "marketplace of ideas", a bridge connecting academic research and public decision-making. The level of a country's think tanks largely determines its strategic foresight capability and the degree of modernization of its governance. The Third Plenary Session of the 18th CPC Central Committee first proposed "strengthening the building of new-type think tanks with Chinese characteristics", and in 2015 the General Office of the CPC Central Committee and the General Office of the State Council issued the "Opinions on Strengthening the Building of New-type Think Tanks with Chinese Characteristics", officially launching the pilot program for national high-end think tanks—thus making think-tank evaluation an important window for observing a country's governance capacity and knowledge-production capacity.
智库（Think Tank）是"思想市场"上的国家软实力载体，是连接学术研究与公共决策的桥梁。一个国家的智库水平，很大程度上决定其战略前瞻能力与治理现代化程度。中国共产党十八届三中全会首次提出"加强中国特色新型智库建设"，2015 年中办、国办印发《关于加强中国特色新型智库建设的意见》，正式开启国家高端智库建设试点——智库评价由此成为观察国家治理能力与知识生产能力的一个重要窗口。
However, the quality of a think tank is extremely hard to quantify: it differs from an enterprise (which has revenue and profit) and from a university (which has papers, patents and disciplines); its core outputs are "ideas" and "influence"—research reports, policy recommendations, agenda-setting capability on public issues, and the social externality after its advice is adopted in decision-making. Internationally common think-tank rankings mostly rely on expert peer review and subjective reputation scoring, with insufficient transparency and poor reproducibility, and rarely incorporate "serving national decision-making", the core mission of think tanks with Chinese characteristics.
然而，智库的优劣极难量化：它既不同于企业（有营收、利润），也不同于大学（有论文、专利、学科），其核心产出是"思想"与"影响"——研究报告、政策建议、公共议题的议程设置能力，以及建言被决策采纳后的社会外部性。国际通行的智库榜单多依赖专家同行评议（peer review）与主观声誉打分，透明性不足、可复现性差，且鲜少纳入"服务国家决策"这一中国特色智库的核心使命。
This index abandons the practice of "reproducing public ranking positions" and returns to underlying public data to compute its own ranking, building a four-dimensional evaluation framework of "Scale—Quality—Innovation—Contribution": Scale characterizes the volume of talent and outputs, Quality characterizes the academic and social influence of outputs (including per-capita efficiency), Innovation characterizes layout in emerging fields and methodological breakthroughs, and Contribution characterizes the actual effect of advice being adopted and serving national decision-making (a distinctive CPPPS dimension generally absent from international rankings). The four dimensions are range-standardized and then weighted and summed, with large-span indicators taken in logarithmic form to curb head monopoly.
本指数放弃"摘抄公开榜单名次"的做法，回归底层公开数据自算排名，构建"规模—质量—创新—贡献"四维评价框架：规模刻画人才与成果的体量，质量刻画成果的学术与社会影响力（含人均效率），创新刻画新兴领域布局与方法论突破，贡献刻画建言被采纳、服务国家决策的实际效果（CPPPS 特色维度，国际榜单普遍缺失）。四大维度经极差标准化后加权求和，大跨度指标取对数以抑制头部垄断。
Evaluation results: the Chinese Academy of Social Sciences, the Chinese Academy of Sciences, the Chinese Academy of Engineering, the Development Research Center of the State Council and the Party School of the CPC Central Committee (National Academy of Governance) rank in the top five. Overall, comprehensive national-level think tanks and academic think tanks with complete institutional structures, huge talent and output volumes, and deep influence rank ahead; university-affiliated think tanks known for "small scale, high innovation and strong influence" and private think tanks focused on niche fields also occupy important positions by virtue of per-capita efficiency and innovation indices. The ranking data are publicly known approximate values, marked "subject to official figures", permanently free and public, and are provided for research and reference only.
评价结果：中国社会科学院、中国科学院、中国工程院、国务院发展研究中心、中央党校（国家行政学院） 位列前五。整体看，建制完整、人才与成果规模宏大、影响力深厚的综合性国家级智库与学术型智库居前；以"小规模、高创新、强影响"见长的高校智库与聚焦细分领域的民间智库亦凭借人均效率与创新指数占据重要位次。本排名数据为公开知识近似值，标注"以官方为准"，永久免费公开，仅供研究与参考。

分层梯队Tiers

A++ 前三Top 3 A+ 四至十4th–10th A 十一至二十11th–20th B 其余Rest ■ 政府型 ■ 学术型 ■ 高校型 ■ 民间型

24 家智库排名Ranking of 24 Think Tanks

排名智库综合得分规模质量创新贡献研究人员成果建言采纳梯队
1中国社会科学院Chinese Academy of Social Sciences学术型85.53100.083.160.694.0400018000280A++
2中国科学院Chinese Academy of Sciences学术型69.8767.248.5100.074.020003000180A++
3中国工程院Chinese Academy of Engineering学术型60.8458.140.684.868.78002000160A++
4国务院发展研究中心Development Research Center of the State Council政府型60.5535.648.460.6100.0300900320A+
5中央党校（国家行政学院）Central Party School (National Academy of Governance)政府型58.5268.061.030.368.79005000160A+
6中国宏观经济研究院Chinese Academy of Macroeconomic Research政府型51.4449.648.330.374.04002500180A+
7北京大学国家发展研究院National School of Development, Peking University高校型50.0116.461.390.937.48090080A+
8清华大学国情研究院Institute for Contemporary China Studies, Tsinghua University高校型43.739.159.678.831.36080070A+
9中国国际经济交流中心China Center for International Economic Exchanges民间型42.9727.748.348.547.41501200100A+
10中国现代国际关系研究院China Institutes of Contemporary International Relations政府型41.4538.539.530.355.73001500120A+
11上海社会科学院Shanghai Academy of Social Sciences学术型41.171.940.624.224.41100700060A
12全球化智库（CCG）民间型37.9214.750.769.720.480100055A
13中国国际问题研究院China Institute of International Studies政府型34.0521.636.224.251.7130900110A
14中国人民大学国家发展与战略研究院National Academy of Development and Strategy, Renmin University of China高校型34.0514.652.742.424.470100060A
15商务部国际贸易经济合作研究院Chinese Academy of International Trade and Economic Cooperation政府型33.7834.228.69.159.32501300130A
16中国科学技术发展战略研究院Chinese Academy of Science and Technology for Development政府型33.7318.425.154.542.712080090A
17复旦大学发展研究院Fudan Development Institute高校型30.8418.547.739.416.190110050A
18中国财政科学研究院Chinese Academy of Fiscal Sciences政府型30.4427.326.50.062.61801100140A
19中国（海南）改革发展研究院China Institute for Reform and Development民间型30.4415.638.136.431.39090070A
20盘古智库Pangoal Institution民间型28.37.742.560.66.06080040A
21广东省社会科学院Guangdong Academy of Social Sciences学术型27.5544.733.112.116.1350250050B
22上海国际问题研究院Shanghai Institutes for International Studies学术型26.4514.928.524.237.410070080B
23察哈尔学会Charhar Institute民间型19.411.830.948.50.05060035B
24中国南海研究院National Institute for South China Sea Studies政府型17.035.719.618.224.47050060B
表中"规模/质量/创新/贡献"为四维子得分（0–100）；"研究人员"为研究人员数（人，含在编科研与研究人员）；"成果"为近五年研究成果数（篇/部，含研究报告、专著、核心论文）；"建言采纳"为近五年政策建言被采纳或进入决策的项数。此外模型还纳入高级专家数（院士/学部委员/正高）、成果被引/转载次数、研究创新指数三项底层数据。以上均为公开知识近似值，以官方为准。
In the table, 'Scale/Quality/Innovation/Contribution' are four-dimensional sub-scores (0–100); 'Total Market Cap' is an approximate total market capitalization of the index constituents (or the market it represents) in trillion USD; '10Y Annualized' is an approximate annualized return over the past 10 years. The model also incorporates four underlying data points: number of listed companies / constituents, average daily turnover, share of technology stocks, and annual equity financing. CSI 300 overlaps in constituents with the Shanghai and Shenzhen indices, so its financing amount and market cap are approximated under their respective scopes and are for reference only. All of the above are public approximate values, subject to official figures.
In the table, 'Scale/Quality/Innovation/Contribution' are four-dimensional sub-scores (0–100); 'Researchers' is the number of research staff (persons, including in-house research and research personnel); 'Outputs' is the number of research outputs over the past five years (papers/works, including research reports, monographs, and core-journal papers); 'Adopted Proposals' is the number of policy proposals adopted or entered into decision-making over the past five years. The model also incorporates three underlying data points: number of senior experts (academicians / academic-division members / full professors), citation/reprint counts of outputs, and a research innovation index. All of the above are public approximate knowledge values, subject to official figures.

算法详细论述Methodology

一、智库评价的依据与理论来源

A think tank is a specialized institution in a modern national governance system that "produces ideas and influences decisions". This algorithm's evaluation of think tanks rests on three groups of grounds. First, policy grounds—the deployment of "strengthening the building of new-type think tanks with Chinese characteristics" at the Third Plenary Session of the 18th CPC Central Committee and the national high-end think-tank building orientation established by the "Opinions on Strengthening the Building of New-type Think Tanks with Chinese Characteristics", which lists "serving the decisions of the Party and the state" as the primary function of think tanks. Second, theoretical grounds—the classic frameworks of modern think-tank research (Dickson's definition of "think tanks", Rich's study of the politics of think tanks and policy expertise, and McGann's cross-national comparison of global think tanks), which generally take "research output", "policy influence", "independence" and "public communication" as the core variables of think-tank capability. Third, evaluation-methodology grounds—the four iron rules followed by this index (public weights, public algorithm, public data sources, reproducibility), and the CPPPS four-layer indicator framework (objective, reputation, expert, contribution).
智库是现代国家治理体系中"生产思想、影响决策"的专门机构。本算法对智库的评价，建立在三组依据之上：其一，政策依据——十八届三中全会关于"加强中国特色新型智库建设"的部署与中办、国办《关于加强中国特色新型智库建设的意见》确立的国家高端智库建设导向，将"服务党和国家决策"列为智库的首要功能；其二，学理依据——现代智库研究的经典框架（Dickson 对"智库"概念的界定、Rich 对智库与政策专业知识之政治的研究、McGann 对全球智库的跨国比较），普遍以"研究产出""政策影响""独立性"与"公共传播"作为智库能力的核心变量；其三，评价方法论依据——本指数所遵循的四条铁律（权重公开、算法公开、数据源公开、可复现），以及 CPPPS 四层指标框架（客观、声誉、专家、贡献）。
Based on the above grounds, this algorithm deconstructs think-tank capability into four measurable dimensions:
基于上述依据，本算法将智库能力解构为四个可测度的维度：
  1. 规模（Scale）：研究人员数量、高级专家数量、研究成果数量——反映智库的智力资源体量与知识生产规模。
  2. 质量（Quality）：成果被引/转载次数（学术与社会影响力）与人均成果效率——反映成果的真实影响力与生产效率，剔除"堆量不增效"的泡沫。
  3. 创新（Innovation）：研究创新指数（新兴领域布局、研究方法论突破、跨学科交叉）——反映智库在前沿议题上的前瞻能力与开风气之先的能力。
  4. 贡献（Contribution）：政策建言被采纳、进入决策的项数——反映智库"服务国家决策"的实际效果，是中国特色新型智库区别于国际智库榜单的核心维度。

二、模型的数学结构

Each underlying indicator is first unified into the 0–1 range via min-max standardization, and large-span indicators (researchers, senior experts, research outputs, citation counts, adopted advice) first take the natural logarithm to curb head monopoly:
各项底层指标首先经 min-max 标准化统一至 0–1 区间，大跨度指标（研究人员、高级专家、研究成果、被引次数、建言采纳）先取自然对数以抑制头部垄断：
σ(x) = (x − xmin) / (xmax − xmin)  ；  对规模/影响类指标先取 ln(x)
The four-dimensional sub-scores are defined as follows:
四维子得分定义如下：
规模 S = 0.35·ln(研究人员)′ + 0.30·ln(高级专家)′ + 0.35·ln(研究成果)′ 质量 Q = 0.70·ln(被引/转载)′ + 0.30·ln(1+人均成果)′ 创新 I = 创新指数′ 贡献 C = ln(建言采纳)′
Here ′ denotes the 0–1 value after min-max standardization. The four dimensions are weighted and summed, and mapped to a 0–100 composite score:
其中 ′ 表示经 min-max 标准化后的 0–1 值。四维按权重加权求和，映射为 0–100 综合得分：
综合得分 = 100 × [ 0.25·S + 0.30·Q + 0.20·I + 0.25·C ]
The weight design embodies the orientation of "emphasizing quality and contribution": Quality accounts for 30% (the influence of ideas is the foundation of a think tank), Contribution for 25% (serving decision-making is the first function of think tanks with Chinese characteristics), Scale for 25% (talent and outputs are the foundation), and Innovation for 20% (foresight determines the long-term vitality of a think tank). The four weights sum to 1, ensuring the scores are normalized and comparable.
权重设计体现"重质量、重贡献"的导向：质量占 30%（思想的影响力是智库的根本），贡献占 25%（服务决策是中国特色智库的第一功能），规模占 25%（人才与成果是基础），创新占 20%（前瞻性决定智库的长期生命力）。四项权重之和为 1，保证得分归一可比。

三、指标与权重依据

指标方向处理权重维度依据
研究人员数量正向对数规模 0.35规模智力资源体量；Dickson 智库定义
高级专家数量正向对数规模 0.30规模顶尖人才储备（院士/学部委员/正高）
研究成果数量正向对数规模 0.35规模知识生产规模；研究报告/专著/论文
成果被引/转载正向对数质量 0.70质量学术与社会影响力；引用即认可
人均成果效率正向对数质量 0.30质量生产效率；剔除"堆量不增效"
研究创新指数正向线性创新 1.00创新新兴领域布局、方法论突破
政策建言采纳正向对数贡献 1.00贡献服务国家决策；《意见》首要功能

四、数据来源与局限性

The number of researchers, senior experts, research outputs, citation/republication counts, adopted advice items and the research-innovation index used in this ranking are all approximate values based on public knowledge (institutional websites, public research reports over the years, public academic and media citation records, etc.), not item-by-item officially audited data, uniformly marked "subject to official figures" to clarify their traceable and revisable boundaries. The model has the following limitations: first, the core output of a think tank (the influence of ideas) is hard to quantify completely, and "citations/republications" and "adopted advice" are only partial proxy variables, so true influence may be under-recorded; second, a large amount of classified outputs and undisclosed internal reports of government-type and academic-type think tanks are not included in statistics, so their actual contributions may be underestimated; third, min-max standardization is sensitive to extreme values, and an individual institution's peak in one dimension may affect the overall distribution of that dimension. This ranking is provided for research reference only and does not constitute any basis for resource allocation or decision-making.
本排名所采用的研究人员数、高级专家数、研究成果数、被引/转载次数、建言采纳项数及研究创新指数，均为基于公开知识（机构官网、历年公开研究报告、公开学术与媒体引用记录等）的近似值，非逐项官方审计数据，统一标注"以官方为准"以明示其可追溯与可修正的边界。模型存在以下局限：其一，智库的核心产出（思想影响力）难以完全量化，"被引/转载"与"建言采纳"仅为部分代理变量，真实影响力可能存在漏记；其二，政府型与学术型智库的大量涉密成果、未公开的内部报告不纳入统计，其实际贡献可能被低估；其三，min-max 标准化对极端值敏感，个别机构在某维度的峰值会影响该维度的整体分布。本排名仅供研究参考，不构成任何资源配置或决策依据。

五、参考文献

  1. 《中共中央关于全面深化改革若干重大问题的决定》（2013）——加强中国特色新型智库建设。
  2. 中共中央办公厅、国务院办公厅：《关于加强中国特色新型智库建设的意见》（2015）。
  3. 国家高端智库建设试点工作启动（2015 年 12 月）——首批国家高端智库建设试点单位。
  4. Dickson, P.（1971）：《Think Tanks》——"智库"概念的奠基性著作。
  5. Rich, A.（2004）：《Think Tanks, Public Policy, and the Politics of Expertise》——智库与政策专业知识之政治。
  6. McGann, J. G.：宾夕法尼亚大学智库与公民社会项目（TTCSP）全球智库报告——跨国智库比较的方法论参照。
  7. 《中国智库报告》（历年蓝皮书）——国内智库评价的研究积累。
  8. 国家统计局：研究与试验发展（R&D）经费投入统计公报（规模类口径参照）。

说明Notes

本版采用"规模+质量+创新+贡献"四维对数加权模型（自有算法，非摘抄公开榜单名次）：规模 S=0.35·ln(研究人员)+0.30·ln(高级专家)+0.35·ln(研究成果)；质量 Q=0.70·ln(被引/转载)+0.30·ln(1+人均成果)；创新 I=创新指数；贡献 C=ln(建言采纳)。四项经 min-max 标准化至 0–1 后加权：综合得分=100×[0.25·S+0.30·Q+0.20·I+0.25·C]。大跨度指标取对数以抑制头部垄断，贡献维度（服务国家决策）为 CPPPS 特色维度。评价对象覆盖国家级高端智库、地方社科院、高校智库与民间智库共 24 家。数据为公开知识近似值，以官方为准；涉密成果与未公开内部报告不纳入统计。权重公开、算法公开、数据源公开、可复现，排名永久免费公开、永不卖榜，保留申诉与复算通道。
This edition adopts the "scale+quality+innovation+contribution" four-dimension log-weighted model (proprietary algorithm, not reproducing public list rankings): scale S=0.35·ln(researchers)+0.30·ln(senior experts)+0.35·ln(research achievements); quality Q=0.70·ln(citations/reprints)+0.30·ln(1+per-capita output); innovation I=innovation index; contribution C=ln(adopted recommendations). The four terms are min-max standardized to 0–1 and then weighted: composite score = 100×[0.25·S+0.30·Q+0.20·I+0.25·C]. Wide-span indicators are log-transformed to curb top-tier monopoly, and the contribution dimension (serving national decision-making) is a CPPPS signature dimension. Coverage includes 24 national-level high-end think tanks, local academies of social sciences, university think tanks, and non-governmental think tanks. Data are public approximate values, subject to official figures; classified achievements and unpublished internal reports are not included. Weights, algorithm, and data sources are public and reproducible; the ranking is permanently free and public, rankings are never sold, and appeal and recalculation channels are preserved.