Abstract
Central bank communication is considered a useful tool through which monetary authorities keep the public informed about its current and future intentions regarding monetary policy decisions and other expected changes in the conduct of the monetary policy. Historically, central banks publish quantitative estimates of several macroeconomic variables. Such quantitative measures are used to study the current economic situation and predict the future movement of the economic indicators. This practice is changing as several analyses on monetary policy are carried on through textual analysis tools of the available text data from central banks. Analysing narratives of the central banks to understand the state of the economy and predict fluctuations in economic measures is considered an important supplement to the overall economic analysis. In this thesis, I study the communication strategies of the three major central banks, the Bank of England, the European Central Bank, and the Federal Reserve of the US, through different methods of text analysis applied to the speeches of the officials of these central banks for the period from 1998 to 2018. The work of this thesis can be divided into two major parts. The first part provides a thorough overview of the literature on central bank communication through text analysis methods. The second part presents, through three different papers, stimulating empirical analyses that aim to complement the untapped areas of text analysis applications in central bank communication.With regard to the first part, I conclude that the growing literature on text analysis applications in central bank communication finds that text data are an essential source of information that can shape inflation expectations, influence the allocation of resources by investors in financial markets and provide more insights into financial stability, especially during turbulent times. In the second part, three contribution papers share a methodological application that involves Natural Language Processing (NLP) techniques, specifically the word frequency methods, sentiment analysis through a dictionary approach, and topic modelling combined with timeseries analysis. The three main conclusions from the three papers respectively are: (1) central bank communication is increasingly used as a tool to discuss economic activity with the general public openly, but the intensity of use of such a tool differs across banks and economic states; (2) the information derived from the speeches of the central banks improve forecasts of retail sales, money growth, unemployment rate and return and volume of the major financial indexes. The forecast improvement is more pronounced after the recent global financial crisis; (3) by using the quarterly forecast data provided by the Survey of Professional Forecasters and through the sentiment analysis of the ECB’s speeches, it is found that the growth and unemployment estimates become more accurate mainly during time-contingent guidance. In contrast, forecast convergence is particularly noted during data-driven guidance/state-contingent guidance. Change in the tone of the speeches mainly improves the accuracy of growth estimates and yields to lower standard deviation of point forecast estimates for growth and unemployment forecasts.
| Date of Award | 1 Jun 2023 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Fotios Papailias (Supervisor) & Georgios Chortareas (Supervisor) |