Ph.D. in Economics, London School of Economics
MSc. in Economics, London School of Economics
B.A. in Economics, Sun Yat-sen (Zhongshan) University
Dr. Wu is Associate Professor of Economics and of Management and Strategy at the University of Hong Kong. He is also a research fellow of the Centre for Economic Policy Research (CEPR, London) and a non-residential Fellow of the 21 Century China Center at UC San Diego. Prior to HKU, he was Assistant Professor of Finance and Business Economics at the Marshall School of Business, University of Southern California. His primary research interest is media economics, in which he has studied how social media affects economic performance, policy making, and consumer behavior in China. Another stream of his research focuses on creative production in the digital economy and the welfare implications of AI-guided economic activities. Recently, he has collaborated with data scientists to apply cutting-edge big data methods to various areas in the social sciences. His work has been published at top economics, management, and statistics journals, including the American Economic Review, Review of Economics and Statistics, Economic Journal, Journal of Economic Perspectives, Journal of the American Statistical Association, Management Science, and Organization Science.
- Economics of Organization and Strategy (PhD, Master)
- Managerial Economics (MBA)
- Causal Inference (Undergraduate)
- Big Data Economics (Undergraduate)
- Media Economics, Organizational Economics, Development Economics, Chinese Economy, Big Data and Computational Social Science
- Competition, Contracts, and Creativity: Evidence from Novel Writing in a Platform Market (with Feng Zhu), Management Science, forthcoming.
- Intentional Control of Type I Error Over Unconscious Data Distortion: A Neyman-Pearson Approach to Text Classification (with Lucy Xia, Richard Zhao and Xin Tong), Journal of the American Statistical Association, 116(533): 68-81, March, (2021).
- Media Bias in China (with Bei Qin and David Strӧmberg), American Economic Review, 108(9): 2442-76, September, (2018).
- Incentive Contracts and the Allocation of Talent, Economic Journal, 127(607): 2744-2783, December, (2017).
- Authority, Incentives and Performance: Evidence from a Chinese Newspaper, Review of Economics and Statistics, 99(1): 16-31, March, (2017).
- Why Does China Allow Freer Social Media? Protests versus Surveillance and Propaganda (with Bei Qin and David Strӧmberg), Journal of Economic Perspectives, 31(1): 117-40 (2017).
- Organizational Structure and Product Choice in Knowledge Intensive Firms, Management Science, 61(8): 1830-1848 (2015).
- Knowledge, Communication and Organizational Capabilities (with Luis Garicano), Organization Science, 23(5):1382-1397 (2012).
- Hong Kong GRF Grant (No. 17500321), PI, 2021-2024
- European Research Council Advanced Grant, Co-PI, 2018-2022
- USC Marshall Outlier Research Fund, PI, 2016-2018
- USC Greif Entrepreneurship Research Award, 2014
- Vincent Cheng Scholarship (LSE), 2007-2010
- Chevening Scholarship (British Council), 2003-2004
This article addresses the challenges in classifying textual data obtained from open online platforms, which are vulnerable to distortion. Most existing classification methods minimize the overall classification error and may yield an undesirably large Type I error (relevant textual messages are classified as irrelevant), particularly when available data exhibit an asymmetry between relevant and irrelevant information. Data distortion exacerbates this situation and often leads to fallacious prediction. To deal with inestimable data distortion, we propose the use of the Neyman–Pearson (NP) classification paradigm, which minimizes Type II error under a user-specified Type I error constraint. Theoretically, we show that the NP oracle is unaffected by data distortion when the class conditional distributions remain the same. Empirically, we study a case of classifying posts about worker strikes obtained from a leading Chinese microblogging platform, which are frequently prone to extensive, unpredictable and inestimable censorship. We demonstrate that, even though the training and test data are susceptible to different distortion and therefore potentially follow different distributions, our proposed NP methods control the Type I error on test data at the targeted level. The methods and implementation pipeline proposed in our case study are applicable to many other problems involving data distortion. Supplementary materials for this article, including a standardized description of the materials available for reproducing the work, are available as an online supplement.