Off-campus WSU users: To download campus access dissertations, please use the following link to log into our proxy server with your WSU access ID and password, then click the "Off-campus Download" button below.
Non-WSU users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Access Type
WSU Access
Date of Award
January 2022
Degree Type
Dissertation
Degree Name
Ph.D.
Department
Economics
First Advisor
Liang Hu
Abstract
Exchange rate dynamics is a central topic in international finance. However,exchange rate returns are known to be difficult, if not impossible, to forecast using readily available information. Stylized facts may provide insights to the behavior of exchange rate returns and volatility. For example, their returns have fat tail distributions, while the volatility exhibits high persistence. Though both returns and volatility tend to increase during bear markets and financial crises, research shows that the correlation between volatility is stronger than returns, meaning that volatility tends to be more predictable.
In this dissertation, we examine two topics in exchange rate forecasting:directional forecastability and volatility forecastability. For the first topic, we use Logistic Regression, the Ridge Classifier, Support Vector Machines, and an ensemble of the three classifiers to forecast the one-day ahead directional change for seven bilateral exchange rates and evaluate their relative out-of-sample performance. We also propose a trading strategy guided by each classifier and use market data to test this strategy. We find that Ridge Classifier is the strongest performers based on both statistical and economic criteria. Furthermore, it is possible to earn a positive profit with our trading strategy even during recessions, especially when we include a momentum indicator (TRIX) and a volatility proxy (true range) in the specification.
For the second topic, we compare the performance of models that usehigh-frequency data to models that use daily data in forecasting exchange rate volatility. Additionally, we compare the performance of long-memory models to short-memory models. Finally, we explore the impact of recent events, such as the Brexit and the COVID-19 pandemic, on exchange rate volatility. Our findings suggest that high-frequency data can improve forecasting accuracy. Moreover, the long-memory models tend to outperform the short-memory models. Finally, we discover a spike in volatility during the COVID-19 pandemic, as well as an increase in forecasting error. Nevertheless, these large shocks to exchange rate volatility at the start of the pandemic were of short-memory nature.
Recommended Citation
Tarchick, John, "Topics In Exchange Rate Forecasting: Directional Change And Volatility" (2022). Wayne State University Dissertations. 3769.
https://digitalcommons.wayne.edu/oa_dissertations/3769