Essays on Monetary Policy Shocks and Firm Strategic Response: Applications of Machine Learning Methods

The business environment is changing rapidly due to various factors, such as COVID-19 pandemic, financial crises, and governmental policy changes. Therefore, firms must have the ability to be agile and adapt quickly in order to survive and succeed in such an environment (Alchian, 1950). However, much remains unknown, such as how firms strategically respond to shocks, how effective their responses are, and how to quantify their agility. To fill this gap, my dissertation examines the strategic responses of firms around the world to monetary and regulatory policy changes originating from central banks and financial market regulators such as the Securities and Exchange Commission (SEC) in U.S.A. I apply the latest machine learning methods to analyze textual data to examine my research questions.
The first essay studies the stock reaction to monetary policy shocks in the Eurozone and shows that by issuing management guidance before the ECB announcement, young firms can reduce their exposures by as much as 35 percent. The study is relevant for policymakers as it responds directly to Janet Yellen’s call for more research to understand the role of firm heterogeneity in explaining the influence of monetary policy on the economy (Yellen, 2016).
The second essay studies the corporate agility concept – ability to respond quickly and effectively to business environment changes. While this ability is crucial for firms’ success, it is difficult to measure and use in quantitative research. We apply machine learning techniques and develop a reliable and replicable measure of agility. Our measure is one of the first in finance and it can enhance academic research about the implications of corporate agility on corporate governance, corporate finance, and asset pricing. We also use this measure to examine how agile firms manage exposure to monetary policy uncertainty – a significant form of threat.
The third essay studies how firms use voice to disclose information and how investors react to it. Using machine learning methods to measure passive voice in 10-K filings, I find that to moderate investor reactions, firms strategically avoid using passive voice to disclose bad news and that investors overreact negatively to passive voice after 10-K release. The paper contributes not only to the strategic disclosure literature but also to textual analysis literature in finance and accounting by introducing a new dimension to analyze textual data, namely passive voice, which has less measurement error and contain novel information about future financial performance.