Einflussfaktoren der Informationsverarbeitung von Nachrichten und die resultierende Geschwindigkeit der Aktienpreisreaktion
Zusammenfassung der Projektergebnisse
This project built a unique, comprehensive dataset of US online company news from March 2015 till November 2016 to examine in much more detail how investors process company information and how such information gets diffused via the financial news markets. The project is intended to provide the basis of multiple studies. The first one uses a subset of news about company earnings announcements to examine the existence of financial news market segmentation. It proposes that financial events frequently leave room for interpretation, allowing news outlets to differentiate and target audiences with different levels of financial sophistication and dispositional optimism. The study develops a probabilistic model to infer these unobservable audience characteristics from earnings announcement coverage and finds economically significant heterogeneity. Furthermore, consistent with this heterogeneity reflecting differences in earnings news interpretations, a larger difference in audiences exposed to an earnings announcement is associated with significantly higher trading volume and return volatility after the announcement. Robustness tests give further support. For example, the model used to infer audience characteristics can also be used to predict which websites will cover an earnings announcement as well as the tone and the style of the resulting writing. The study furthers our understanding on information diffusion in financial markets in three ways. First, while other research has provided compelling evidence of slant in political news coverage being driven by political belief heterogeneity, it is not clear ex-ante whether similar mechanisms are at work in the financial news market. In contrast to political news, financial information provides much less room for interpretation and disagreement, so that the existence of segmentation in today's competitive online news market with its non-local scope and low barriers to entry is an open question. The answer is highly relevant though for regulations aimed at efficient diffusion of information in financial markets. The study provides evidence of sizable and predictable outlet-specific variation in earnings news coverage and trading volume patterns that is consistent with the financial news market catering to differences in beliefs rather than reducing them—adding further pieces to the puzzle of a “more complete theory of the role of the media in financial markets”. Second, since target audience characteristics are typically unobservable, the study further contributes to the emerging literature employing structural estimation approaches to accounting and finance settings. The developed model offers a flexible, robust approach to inferring target audiences of news outlets from multiple observable actions in large scale but noisy data. This approach can also be applied to other settings, such as inferring heterogeneous analyst or CEO characteristics. Third, given that news coverage guides investor attention, the documented significant variation in news audience characteristics and its relation to trading volume provides important insights for future theories of investors' information acquisition strategies and firms' disclosure strategies. Future follow-up studies will build on these results to develop and test those implications.
