The influence of news and investor sentiment on exchange rate determination: new evidence using panel data in the banking sector

PhD Thesis

Dimitriadou, A. 2024. The influence of news and investor sentiment on exchange rate determination: new evidence using panel data in the banking sector. PhD Thesis Department of Business, Law, and Social Science
AuthorsDimitriadou, A.
TypePhD Thesis

Exchange rates behaviour in open economies strongly influences the country's macroeconomic policy as the extent and frequency of exchange rate changes are important indicators of the country's economic stability. Commercial banks are fairly exposed to exchange rate changes and may be directly and heavily affected. The primary goal of this study is to investigate whether exchange rates news plays a significant role in banks’ financial performance, and what other channels (factors) potentially affect the banks’ profitability. The study collected data on more than 800 US banks over the period of 21 years (1998 to 2020). Following a filtering process, 148 banks were retained, as a significant number of these institutions either declared bankruptcy or underwent mergers with larger organizations, whether in banking or investment sectors. The contribution of this study is twofold. Firstly, the investigation of the association between exchange rates news and banks' profitability, creating a net sentiment index based on the unexpected announcements of domestic currency, US dollar, and then using GMM techniques, and secondly, the examination of this net sentiment index on banks’ profitability in combination with other banking or macroeconomic factors. While the determinants of banks' profitability have been studied by many scholars, the relationship between exchange rate news and profitability has not been analyzed by anyone so far.

The analysis relies on public news categorized as favourable and unfavourable exchange rate news based on exchange rate fluctuations for 3 exchange rates. This analysis generates an index that describes the net sentiment of this news based on the characteristics of those announcements. The data of this net sentiment index is obtained from 3 basic exchange rates fluctuations per year, defining the US dollar as the domestic currency. Based on the major changes in exchange rates over time, news is classified as either positive or negative.
Using panel data for 148 US banks during the period 1998-2018 and applying the GMM method, the first goal is to find out if the unexpected exchange rate news has a negative or positive impact on the whole banking system, especially if this news affects Return on Assets (ROA), Return on Equity (ROE) or Net Interest Margin (NIM) which have been defined as measures for the profitability of banks. To do this, empirical econometric tests were performed, finding the best autoregressive model and then applying the Stepwise Forward method selected the most statistically significant variables in each model (p-value < 0.01). The panel unit root, OLS (Fixed Effect) method, and GMM method (GMM single and GMM system) two-step robust estimator, will then be applied for further analysis.

This study showed that banks’ profitability is not affected by unexpected exchange rate announcements, which automatically implies that investors underreact immediately to new information. The evidence presented in this article does not justify banking profit or debt management activities if banks react to good or bad information about the appreciation or depreciation of the dollar. Banks appear to underreact to exchange rates news as well as to information conveyed by the event. So, there is no support for the overreaction hypothesis to unexpected exchange rate news in the banking system, suing any technique. Finally, the analysis does not address whether a different explanation of behavior is based on other phenomena. It may be necessary to reinterpret the evidence in this paper. This is left as an area for future research.

KeywordsExchange rates behaviour; news; Overreaction Hypothesis; sentiment index; GMM; panel data
PublisherCollege of Business, Law and Social Sciences, University of Derby
Digital Object Identifier (DOI)
File Access Level
Output statusUnpublished
Publication process dates
Deposited22 May 2024
Permalink -

Download files

License: CC BY-NC-ND 4.0
File access level: Open

  • 78
    total views
  • 63
    total downloads
  • 12
    views this month
  • 58
    downloads this month

Export as

Related outputs

Predicting Bitcoin Prices Using Machine Learning
Dimitriadou, A. and Andros Gregoriou 2023. Predicting Bitcoin Prices Using Machine Learning. Entropy. 25 (5), pp. 1-10.
Forecasting Hospital Readmissions with Machine Learning
Michailidis, P., Dimitriadou, A., Papadimitriou, T. and Gogas, P. 2022. Forecasting Hospital Readmissions with Machine Learning. Healthcare. 10 (918), pp. 1-14.