AI is changing how financial information and research data are produced, assembled, and analyzed. I study what this means for the beliefs researchers measure and the decisions people make.
My single-author paper develops a method to detect whether survey respondents use AI to answer questions. I show that AI can create artificial consensus: recorded responses cluster more tightly because respondents draw on a shared algorithmic source, not because their underlying beliefs have converged.
Artificial Consensus: AI and Survey Measurement (Single-Author)
Central banks, governments, industry, and investors use surveys to measure economic conditions that are otherwise difficult to observe. AI can distort survey measurement at three stages: by replacing respondents, assisting them, or analyzing the resulting data. I study all three channels by fielding an online survey based on the Survey of Consumer Expectations (SCE), benchmarking answers from fifteen AI models, and analyzing SCE microdata. A browser-side answer-origin measure distinguishes between AI-assisted humans, agentic tools, and autonomous bots, revealing that the dominant respondent-side problem is assistance rather than substitution. Among 1,294 respondents, 31.4% leave evidence of AI involvement, 4.8% disclose it, and only 0.4% are autonomous bots. AI-generated answers reproduce a plausible normal-times center of inflation expectations but collapse their dispersion, while AI-led cleaning disproportionately removes genuine upper-tail SCE responses. With independent respondents, tighter answer distributions generally signal more convergent beliefs; with a shared algorithmic source, the same pattern can arise without belief convergence. AI-induced homogenization can make surveys appear more precise while biasing inference.
Working Papers
Roy Sorting: Climate and Status Quo Strategies (with Samuel Drempetic, Andreas G.F. Hoepner, Christian Klein and Adair Morse)
Do advisors respond to investors’ preferences? (with Julia Eckert, Anne Kellers, Christian Klein and Bernhard Zwergel)
Published Papers
Cauthorn, T., Klein, C., Remme, L., & Zwergel, B. (2023). Portfolio benefits of taxonomy orientated and renewable European electric utilities. Journal of Asset Management, 24(7), 558-571.
Cauthorn, T., Dumrose, M., Eckert, J., Klein, C., & Zwergel, B. (2023). Rating changes revisited: New evidence on short-term ESG momentum. Finance Research Letters, 54, 103703.
Rochell, J., Cauthorn, T., Höck, A., & Zwergel, B. (2020). Drivers of Socially Responsible Investments Across Europe. Credit and Capital Markets–Kredit und Kapital, 53(4), 493-512.
Policy Work
Consileon Frankfurt, EB – Sustainable Investment Management (EB-SIM), Universal Investment Group, University of Kassel (2025). Impact Investment Anlegerstudie
Klein, C., Cauthorn, T., Eckert, J., Kellers, A., Zwergel, B., & Jürgens, I. (2024). Does MiFID II enable private investors to invest sustainably?
Arnold, J. L., Cauthorn, T., Eckert, J., Klein, C., & Rink, S. (2023). Let’s talk numbers: EU Taxonomy reporting by German companies.