Forecasting volatility in Chinese and Hong Kong stock markets.
This paper analyses the forecasting performance of historical volatility models and GARCH-class models of Shenzhen component index, Shanghai composite index and Hang Seng index at weekly and daily frequency under both symmetric and asymmetric loss functions. Under symmetric loss functions exclude Th...
| Main Author: | |
|---|---|
| Format: | Dissertation (University of Nottingham only) |
| Language: | English |
| Published: |
2011
|
| Online Access: | https://eprints.nottingham.ac.uk/25232/ |
| _version_ | 1848792942127349760 |
|---|---|
| author | Wu, Ming |
| author_facet | Wu, Ming |
| author_sort | Wu, Ming |
| building | Nottingham Research Data Repository |
| collection | Online Access |
| description | This paper analyses the forecasting performance of historical volatility models and GARCH-class models of Shenzhen component index, Shanghai composite index and Hang Seng index at weekly and daily frequency under both symmetric and asymmetric loss functions. Under symmetric loss functions exclude Theil-U and HR, results suggest that historical volatility models provide a much better forecast than GARCH-class models both in weekly and daily frequency. Under asymmetric loss functions historical volatility models, especially moving average and random walk, outperform, GARCH-class models. EGARCH, TGARCH and GARCH (3, 1) are found to provide the worst forecast. EGARCH and TGARCH cannot fit the in-sample data. However Theil-U and HR tests suggest that EGARCH and TGARCH have a better forecast on the direction of change of future volatility. And historical volatility models provide the worse forecasts based on HR. Furthermore, the results also suggest that assuming different distributions for errors and utilizing different ARIMA models for return will affect the volatility forecasts accuracy. Assuming a student t distribution will generate a large forecast error than a normal distribution. Furthermore, under normal distribution higher orders of ARIMA will increase the forecast accuracy and under student t distribution higher order of ARIMA will decrease the forecast accuracy. |
| first_indexed | 2025-11-14T18:52:25Z |
| format | Dissertation (University of Nottingham only) |
| id | nottingham-25232 |
| institution | University of Nottingham Malaysia Campus |
| institution_category | Local University |
| language | English |
| last_indexed | 2025-11-14T18:52:25Z |
| publishDate | 2011 |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | nottingham-252322018-01-31T00:34:31Z https://eprints.nottingham.ac.uk/25232/ Forecasting volatility in Chinese and Hong Kong stock markets. Wu, Ming This paper analyses the forecasting performance of historical volatility models and GARCH-class models of Shenzhen component index, Shanghai composite index and Hang Seng index at weekly and daily frequency under both symmetric and asymmetric loss functions. Under symmetric loss functions exclude Theil-U and HR, results suggest that historical volatility models provide a much better forecast than GARCH-class models both in weekly and daily frequency. Under asymmetric loss functions historical volatility models, especially moving average and random walk, outperform, GARCH-class models. EGARCH, TGARCH and GARCH (3, 1) are found to provide the worst forecast. EGARCH and TGARCH cannot fit the in-sample data. However Theil-U and HR tests suggest that EGARCH and TGARCH have a better forecast on the direction of change of future volatility. And historical volatility models provide the worse forecasts based on HR. Furthermore, the results also suggest that assuming different distributions for errors and utilizing different ARIMA models for return will affect the volatility forecasts accuracy. Assuming a student t distribution will generate a large forecast error than a normal distribution. Furthermore, under normal distribution higher orders of ARIMA will increase the forecast accuracy and under student t distribution higher order of ARIMA will decrease the forecast accuracy. 2011-09-23 Dissertation (University of Nottingham only) NonPeerReviewed application/pdf en https://eprints.nottingham.ac.uk/25232/3/edissertation_2.pdf Wu, Ming (2011) Forecasting volatility in Chinese and Hong Kong stock markets. [Dissertation (University of Nottingham only)] (Unpublished) |
| spellingShingle | Wu, Ming Forecasting volatility in Chinese and Hong Kong stock markets. |
| title | Forecasting volatility in Chinese and Hong Kong stock markets. |
| title_full | Forecasting volatility in Chinese and Hong Kong stock markets. |
| title_fullStr | Forecasting volatility in Chinese and Hong Kong stock markets. |
| title_full_unstemmed | Forecasting volatility in Chinese and Hong Kong stock markets. |
| title_short | Forecasting volatility in Chinese and Hong Kong stock markets. |
| title_sort | forecasting volatility in chinese and hong kong stock markets. |
| url | https://eprints.nottingham.ac.uk/25232/ |