Files
Abstract
This dissertation is composed of two essays that explore the literature of corporate finance and risk. The first chapter is about expectation management in corporate finance. Expectation management, e.g., via guiding analyst forecasts downwards to achieve positive earnings surprises, is a common practice adopted by corporate management. This paper investigates Twitter as a mechanism for expectation management. CEOs have been increasingly using Twitter to provide information, either about the firm or their personal lives. Using a hand-collected sample of CEO Twitter usernames, I quantify the sentiment of CEOs’ Tweets prior to their firms’ earnings announcements. I find that tweet sentiment and abnormal stock returns on tweet days are positively correlated, suggesting that the market incorporates the information contained in CEOs’ tweets. Furthermore, negative pre-earnings tweet sentiment is followed by downward analyst earnings forecast revisions and positively predicts the likelihood of meeting earnings expectations. The second chapter is about risk management, a core competence of financial institutions and techniques for quantifying market risk that are central to the process of managing risk. Value-at-risk (VaR) and expected shortfall (also called conditional VaR) are two mostly widely used measures of financial risk. However, both measures have some drawbacks. For example, VaR ignores losses beyond a designated threshold and also fails to satisfy mathematical principles characterizing coherent risk measures while expected shortfall fails to have the elicitability criterion deemed essential to backtesting. These deficiencies have motivated interest in other risk measures. Expectile, which was introduced in the context of linear regression, has been revealed as a reasonable risk measure to offset the weaknesses of VaR and expected shortfall. In this paper, using special characterization of the expectile as the minimizer of a suitable discrepancy function, we propose a method to construct a coherent posterior distribution for the expectile and thus provide the full picture of the expectile of the distribution of financial losses through combining the information from both the data and its prior belief. The asymptotic consistency of the posterior is established to support its validity in practice. Some numerical examples are provided to illustrate our method.