No. 150: Large Language Models and M&A: Can ChatGPT help forecast M&A activity?

Authors

Sönke SieversSönke SieversDDDegen, DominikKJKengelbach, JensKDKim, DanielWYWang, Yiran

Abstract

This paper employs ChatGPT on firms' earnings conference calls transcripts to extract managers' so far private information regarding their sentiment and expectations concerning (future) mergers and acquisitions (M&A) activity. We aggregate these firm-level scores to a market wide ChatGPTbased M&A sentiment score (MASS) and successfully perform various construct validity checks including employing M&A experts and back testing. Our main results suggest that M&A sentiment has strong forecasting power for future M&A activity and adds additional explanatory power to established sentiment indices and fundamental drivers. The predictability holds for both in-sample and out-of-sample tests. Overall, our novel index will be useful to further understand M&A patterns, to predict future M&A activity, and for practitioners to evaluate and benchmark their own M&A activities.

To cite this article

Degen, D., Kengelbach, J., Kim, D., Sievers, S., & Wang, Y. (2024). Large Language Models and M&A: Can ChatGPT help forecast M&A activity?. TRR 266 Accounting for Transparency Working Paper Series No. 150. http://dx.doi.org/10.2139/ssrn.4862121