TY - THES T1 - On the implications of recent advancements in information technologies and high-dimensional modeling for financial markets and econometric frameworks A1 - Schmidt,Alexander Y1 - 2019/12/16 N2 - Around the turn of the millennium, the Organization for Economic Co-operation and Development (OECD) published an article, which summarizes the organizations expectations towards technological developments of the 21st century. Of particular interest to the authors are innovations in the area of information technology, highlighting their far-reaching impact on, amongst others, the financial sector. According to the article, the expected increasing interconnectedness of individuals, markets, and economies holds the potential to fundamentally change not only the flow of information in financial markets but also the way in which people interact with each other and with financial institutions. Looking back at the first two decades of the 21st century, these predictions appear to have been quite accurate: The rise of the internet to a platform of utmost relevance to industries and the economy as a whole profoundly impacts how people nowadays receive and process information and subsequently form, share, and discuss their opinions amongst each other. At the financial markets around the globe trading has become more and more accessible to individuals. Less financial and technical knowledge is required of retail investors to engage in trading, resulting in increased market participation and more heterogeneous trader profiles. This, in turn, influences the dynamics in the financial markets and challenges some of the conventional wisdom concerning market structures. In this context, the interdependencies between the media, retail investors, and the stock market are of particular interest for practitioners. However, the changed dynamics in the flow and exchange of data and information are also highly interesting from a researchers perspective, resulting in entire branches of the academic literature devoted to the topic. While these branches have grown in many different directions, this doctoral thesis explores two specific aspects of this field of research: First, it investigates the consequences of the increased interconnectedness of individuals and markets for the dynamics between the new information technologies and the financial markets. This entails gaining new insights about these dynamics and assessing how investors process certain company-related information for their investment decisions by means of sentiment analysis of large, publicly available data sets. Secondly, it illustrates how an advanced understanding of high-dimensional models, resulting from such analyses of large data sets, can be beneficial in re-thinking and improving existing econometric frameworks. Three independent but related research projects are presented in this thesis that address both of the aforementioned aspects to give a more holistic picture of the implications that the profoundly changed flow and exchange of data and information of the last decades hold for finance and econometrics. As such, the projects (i) highlight the importance of carefully assessing the dynamics between investor sentiment and stock market volatility in an intraday context, (ii) analyze how investors process newly available, rich sources of information on a firms environmental, social, and governance (ESG) practices for their investment decisions, and (iii) propose a new approach to detecting multiple structural breaks in a cointegrated framework enabled by new insights about high-dimensional models. The first original work of this doctoral thesis aims at closing an existing gap in the behavioral finance literature by taking an intraday perspective in assessing the relationship between investor sentiment and stock market volatility. More precisely, the paper titled "The Twitter myth revisited: Intraday investor sentiment, Twitter activity and individual-level stock return volatility", which is joint work with Simon Behrendt, takes a closer look at the dynamics of individual-level stock return volatility, measured by absolute 5-minute returns, and Twitter sentiment and activity in an intraday context. After accounting for the intraday periodicity in absolute returns, we discover some statistically significant co-movements of intraday volatility and information from stock-related Tweets for all constituents of the Dow Jones Industrial Average (DJIA). However, economically, the effects are of negligible magnitude, and out-of-sample forecast performance is not improved when including Twitter sentiment and activity as exogenous variables. From a practical point of view, this chapter finds that high-frequency Twitter information is not particularly useful for highly active investors with access to such data for intraday volatility assessment and forecasting when considering individual-level stocks. Inspired by this first research project, the second original work presented in this thesis keeps its focus on sentiment analysis in the context of the financial markets. Titled "Sustainable news - A sentiment analysis of the effect of ESG information on stock prices", it investigates the effect of ESG-related news sentiment on the stock market performance of the DJIA constituents. Relying on a large data set of news articles that were published online or in print media between the years of 2010 and 2018, each articles sentiment with respect to ESG-related topics is extracted using a dictionary approach from which a polarity-based sentiment index is calculated. Estimating autoregressive distributed lag models reveals significant effects of both temporary and permanent changes in ESG-related news sentiment on idiosyncratic returns for the vast majority of the DJIA constituents. According to the models results, one can assign the stocks to different groups depending on their investors apparent predisposition towards ESG news, which in turn seems to be linked with a stocks financial performance. The last original work presented is then concerned with the second aspect of this doctoral thesis - the question of how our enhanced understanding of the increasingly high dimensional datasets that occur in practice can produce new solutions to familiar problems in econometrics. The paper "Multiple structural breaks in cointegrating regressions: A model selection approach", which is joint work with Karsten Schweikert, introduces the least absolute shrinkage and selection operator (lasso) as a tool for consistent breakpoint estimation. In this paper, we propose a new approach to model structural change in cointegrating regressions using penalized regression techniques. First, we consider a setting with fixed breakpoint candidates and show that a modified adaptive lasso estimator can consistently estimate structural breaks in the intercept and slope coefficient of a cointegrating regression. In such a scenario, one could also perceive our method as performing an efficient subsample selection. Second, we extend our approach to a diverging number of breakpoint candidates and provide simulation evidence that timing and magnitude of structural breaks are consistently estimated. Third, we use the adaptive lasso estimation to design new tests for cointegration in the presence of multiple structural breaks, derive the asymptotic distribution of our test statistics and show that the proposed tests have power against the null of no cointegration. Finally, we use our new methodology to study the effects of structural breaks on the long-run PPP relationship. KW - Social Media KW - Ökonometrie KW - Twitter KW - Strukturbruch KW - Aktienmarkt CY - Hohenheim PB - Kommunikations-, Informations- und Medienzentrum der Universität Hohenheim AD - Garbenstr. 15, 70593 Stuttgart UR - http://opus.uni-hohenheim.de/volltexte/2019/1680 ER -