Impact of central bank speeches

Abstract

This thesis examines the impact of central bank communication on market return The central bank communication is proxied by speeches delivered by the respective governors chairpersons The tone from central bank speeches is extracted and analyzed with respect to its impact on the stock market returns across a varied array of economies The impact of market returns is analyzed independently as well as in the presence of existing tone variables Similar exercise has been attempted in the past Lee et al 1991 Neal and Wheatley 1998 Baker and Wurgler 2006 Baker and Wurgler 2007 Schmeling 2009 and Szu et al 2015 An important distinction in this study is with respect to the process of tone quantification from central bank communication One of the major questions in tone quantification is how many word s should be considered as a batch primary unit The previous studies in related literature have ranged the same from 8220 one word at a time 8221 bagof words approach Tetlock 2007 Loughran and McDonald 2016 to ngram analysis n words at a time Li 2008 Tetlock et al 2008 However all the approaches in this array of one to n words have their share of critics due to the presence of multiple adverbs and adjectives with a corresponding word For example the phrase 8221 not so good financial year 8221 will lead to incorrect tone quantification for one word to ngram n 2 approach as the corresponding noun phrase 8220 financial year 8221 has three adverbs adjectives Also the number of adverbs adjectives can vary widely in any text including central bank communication In this work we suggest an alternate solution to this issue by identifying a sentence as a base unit of tone quantification Also we assign appropriate weights to adverbs adjectives 8220 valence shifters 8221 henceforth to ensure each cluster of noun phrases is quantified correctly within a sentence To the best of our knowledge this is the first attempt at the use of valence shifters to financial text newline newline