Chengyu Fu

Harvard University

About Me

Welcome! I am a Ph.D. Candidate in Government at Harvard University. My main research interests cover corruption and anti-corruption, nonmarket strategies, political connections, and text-as-data. My research has received funding from American Political Science Association, Fairbank Center for Chinese Studies, Harvard Asia Center, Harvard Kennedy School, Institute for Humane Studies, Institute for Quantitative Social Science, and Weatherhead Center for International Affairs.

At Harvard, I am a Desmond and Whitney Shum Fellow at the Fairbank Center for Chinese Studies, a Graduate Student Affliate with the Weatherhead Center for International Affairs, and a Graduate Student Affliate with the Institute for Quantitative Social Science. Prior to doctoral studies, I earned my M.A. in Political Science (awarded Outstanding MA Thesis) from Duke University, and B.A. in Politics and Public Administration (with distinction) from Renmin University of China.

You can reach me at cfu@g.harvard.edu.

Background

Interests

Nonmarket strategies
Corruption and non-compliance
Political connections
Comparative corporate governance
Text-as-data

Education

Ph.D. in Government (expected 2027)
Harvard University
M.A. in Political Science (2022)
Harvard University
M.A. in Political Science (Outstanding MA Thesis, 2020)
Duke University
B.A. in Politics and Public Administration (with distinction, 2018)
Renmin University of China

Research

Chengyu Fu (2023). Political Corruption and Voter Turnout: The Effects of Perception, Experience, and Purchases. Under Review.

Link

Chengyu Fu (2022). Gender Norms and Firms' Corruption: Evidence from China. Norms, Gender and Corruption: Understanding the Nexus. Edward Elgar Publishing.

Link

Chengyu Fu (2022). Corruption and Tax Evasion: Evidence from China. Work In Progress.

Teaching

1. Comparative Politics

Department of Government, Harvard University; Harvard Extension School

Foundations of Comparative Politics [syllabus1] [syllabus2]

Teaching Fellow (with Yuhua Wang and George Soroka)

2. Chinese Politics

Department of Government, Harvard University

Government and Politics of China [syllabus]

Teaching Fellow (with Yuhua Wang)

3. Quantitative Social Science

Department of Government, Harvard University

Quantitative Social Science Methods II [syllabus]

Tutor

Data

Corruption

My dissertation draws on an original database of approximately 120,000 criminal corruption cases in China from 2001 to 2023. It covers all major corruption offenses under Chinese criminal law: bribe-taking, bribe-giving, embezzlement, misappropriation of public funds, and abuse of power. To extract structured information from these semi-structured legal documents, I developed a Large Language Model-based extraction pipeline that parses all cases into (1) more than 380,000 individual corrupt events with their corruption dates, amounts, types, and related firms, and (2) 120,000 case-level data on defendant information including demographics, position, and sentence outcomes. I link these data to anti-corruption inspections and firm-level government procurement database. To my knowledge, this is the most comprehensive transaction-level corruption dataset assembled for any single country.

Round-tripping FDI

Round-trip foreign direct investment—domestic capital that exits a country through offshore financial centers and returns disguised as foreign investment—poses significant measurement challenges for understanding the politics and economics of FDI. This dataset identifies round-trip FDI at the firm level in China using machine learning classification applied to the Foreign-Invested Enterprises in China (FIEC) database. We hand-coded ultimate beneficial ownership for a training sample of 3,300+ firms across Beijing, Nanjing, and Taiyuan, tracing ownership chains through corporate registries and commercial databases to distinguish genuinely foreign investors from mainland-controlled entities using offshore structures. Using these validated labels, we trained a random forest classifier on firm characteristics including ownership structure, registered capital, state involvement, and sector to predict round-trip status for the full population of foreign-invested enterprises. The resulting predictions enable classification of round-trip status for 25,000+ foreign-invested enterprises across three major cities from 2017-2023. This dataset enables researchers to distinguish round-trip from genuinely foreign capital in firm-level analyses of FDI behavior, policy responses, and economic outcomes in China.