Resumen

Despite the long established role of information in the stock price formation as proposed by the efficient market hypothesis (EMH), how news affects the financial market is still a debatable topic. Using an innovative approach, this dissertation addresses the effect of news in financial markets by using a large news database which approximates the size, diversity, and complexity of information investors receive daily. The consideration of high-volume information poses computational challenges in research while at the same time provides an opportunity to better understand the news-financial returns relationship. Earlier studies analyzed this relationship using predetermined variables or news sentiment which can be information restrictive. The information considered in the two studies in this dissertation is not pre-selected but determined by the methodology. The first study establishes the relationship between news and returns of different financial indexes in which information is translated into numerical signals through a Bayesian approach called multinomial inverse regression method (MNIR). This method is capable of translating large quantities of information in news into a quantitative news index that is more comprehensive than what has been used as information proxy in the existing literature. As an input in the autoregressive model, the news index proves to perform well having explanatory power over the returns of different financial indexes which indicates that the methodology captures the information relevant to price formation. The second study brings the analysis of news-financial returns relationship into a more specific social issue - the climate change - which, as the EMH implies, should affect stock prices if it is relevant to the market. Results show that climate change is persistent in the news universe which indicates that it is a relevant news topic. This is supported by the non-zero contribution of climate change related trigrams (CCRTs) in the constructed news index. However, the sample does not show an increasing trend of the relative daily presence of CCRTs which signals that news is unlikely the source of the perceived increasing climate change awareness. This also finds the salient CCRTs present during good and bad days of the market and highlights the presence of topics related to fuel and energy, emission, climate change per se, disaster, and fiscal policy. As shown in these two studies, this dissertation is significant because of the consideration of a unique and large news database in addressing the fundamental information-returns relationship and for being able to translate such information into a news index capable of explaining various financial indexes. Moreover, this dissertation analyzes in granular manner the role of climate change news in financial markets and thus provides better understanding of the news and financial returns relationship.