Modelling and Forecasting Exchange Rate Volatility using High-frequency Data－Based on the US dollar
Abstract In this dissertation, we compare the performance of various models in predicting the USD dollar bilateral exchange rate volatility based on high-frequency data. Four exchange rates are selected, namely USD / EUR, USD / JPY, USD / GBP and USD / SEK. We also asses the forecasting performan...
| Main Author: | |
|---|---|
| Format: | Dissertation (University of Nottingham only) |
| Language: | English |
| Published: |
2018
|
| Online Access: | https://eprints.nottingham.ac.uk/54230/ |
| _version_ | 1848799025724129280 |
|---|---|
| author | Chang, Kaiwen |
| author_facet | Chang, Kaiwen |
| author_sort | Chang, Kaiwen |
| building | Nottingham Research Data Repository |
| collection | Online Access |
| description | Abstract
In this dissertation, we compare the performance of various models in predicting the USD dollar bilateral exchange rate volatility based on high-frequency data. Four exchange rates are selected, namely USD / EUR, USD / JPY, USD / GBP and USD / SEK. We also asses the forecasting performance of four models, including traditional volatility models (the intraday GARCH (1,1) model, the intraday FIGARCH model, the daily GARCH (1,1) model) and realized volatility model (the ARFIMA model). Then the out-of-sample forecast is done by using the rolling window approach. Moreover, four accuracy tests are applying to compare the forecasting ability of different models. The accuracy tests include the Mincer-Zarnowitz test, the loss functions, the DM and HLN-DM test and the SPA test. The empirical results show that the prediction performance of the intraday FIGARCH model and the ARFIMA model outperforms than other models, and the intraday FIGARCH model performs slightly superior to the ARFIMA model. Although the intraday GARCH (1,1) model performs well in some cases, it generally performs inferior to other models. Moreover, the results also indicate that capturing the long memory property of high-frequency data can effectively improve the predicting ability, and it is also found that the high-frequency USD dollar bilateral exchange rate exhibits similar characteristics to other high-frequency financial time series data.
Key Words: USD dollar exchange rates, high-frequency data, the long memory property, volatility forecasting, accuracy tests, realized volatility |
| first_indexed | 2025-11-14T20:29:06Z |
| format | Dissertation (University of Nottingham only) |
| id | nottingham-54230 |
| institution | University of Nottingham Malaysia Campus |
| institution_category | Local University |
| language | English |
| last_indexed | 2025-11-14T20:29:06Z |
| publishDate | 2018 |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | nottingham-542302022-04-29T14:10:18Z https://eprints.nottingham.ac.uk/54230/ Modelling and Forecasting Exchange Rate Volatility using High-frequency Data－Based on the US dollar Chang, Kaiwen Abstract In this dissertation, we compare the performance of various models in predicting the USD dollar bilateral exchange rate volatility based on high-frequency data. Four exchange rates are selected, namely USD / EUR, USD / JPY, USD / GBP and USD / SEK. We also asses the forecasting performance of four models, including traditional volatility models (the intraday GARCH (1,1) model, the intraday FIGARCH model, the daily GARCH (1,1) model) and realized volatility model (the ARFIMA model). Then the out-of-sample forecast is done by using the rolling window approach. Moreover, four accuracy tests are applying to compare the forecasting ability of different models. The accuracy tests include the Mincer-Zarnowitz test, the loss functions, the DM and HLN-DM test and the SPA test. The empirical results show that the prediction performance of the intraday FIGARCH model and the ARFIMA model outperforms than other models, and the intraday FIGARCH model performs slightly superior to the ARFIMA model. Although the intraday GARCH (1,1) model performs well in some cases, it generally performs inferior to other models. Moreover, the results also indicate that capturing the long memory property of high-frequency data can effectively improve the predicting ability, and it is also found that the high-frequency USD dollar bilateral exchange rate exhibits similar characteristics to other high-frequency financial time series data. Key Words: USD dollar exchange rates, high-frequency data, the long memory property, volatility forecasting, accuracy tests, realized volatility 2018-12-01 Dissertation (University of Nottingham only) NonPeerReviewed application/pdf en https://eprints.nottingham.ac.uk/54230/1/Electronic-Edissertation.pdf Chang, Kaiwen (2018) Modelling and Forecasting Exchange Rate Volatility using High-frequency Data－Based on the US dollar. [Dissertation (University of Nottingham only)] |
| spellingShingle | Chang, Kaiwen Modelling and Forecasting Exchange Rate Volatility using High-frequency Data－Based on the US dollar |
| title | Modelling and Forecasting Exchange Rate Volatility using High-frequency Data－Based on the US dollar |
| title_full | Modelling and Forecasting Exchange Rate Volatility using High-frequency Data－Based on the US dollar |
| title_fullStr | Modelling and Forecasting Exchange Rate Volatility using High-frequency Data－Based on the US dollar |
| title_full_unstemmed | Modelling and Forecasting Exchange Rate Volatility using High-frequency Data－Based on the US dollar |
| title_short | Modelling and Forecasting Exchange Rate Volatility using High-frequency Data－Based on the US dollar |
| title_sort | modelling and forecasting exchange rate volatility using high-frequency data－based on the us dollar |
| url | https://eprints.nottingham.ac.uk/54230/ |