Research Projects Supported by HKU's High Performance Computing Facilities

¡@
Researcher:
Dr Pak-wing Fong, Department of Statistics and Actuarial Science
¡@

Project Title:
Heteroscedastic Models with Random Coefficients
Project Description:
Several research works are related to model volatility in financial and economic time series. To capture the pattern of non-constant variance in time series, random coefficients are introduced to homoscedastic time series models.
Project Duration:
3 years
Project Significance:
As the development of economy and finance in the world market has reached another milestone in recent years, traditional models cannot adequately describe varying time series in the financial market. The proposed models provide flexible forms of nonstationarity with conditional heteroscedasticity.

- Back to top- / Contents
Results Achieved:
Two research papers are published and one is under revision. Details can be found in Fong and Li (2003), Fong and Li (2004a) and Fong, An, and Li (2004b).

Some major results by HPC Cluster are shown below :

Numerical outputs by HPC Cluster (Fong and Li, 2003) :

¡@

- Back to top- / Contents
Numerical outputs by HPC Cluster (Fong and Li, 2004a)

- Back to top- / Contents
Outputs by HPC Cluster (Fong and Li, 2004b)

¡@

- Back to top- / Contents
References:
Fong, P. W. and W. K. Li, 2003. On time series with randomized unit root and randomized seasonal unit root. Computational Statistics & Data Analysis 43, 369-395.

Fong, P. W. and W. K. Li, 2004a. Some results on cointegration with random coefficients in the error correction form : estimation and testing, to appear in Journal of Time Series Analysis.

Fong, P. W., H.Z. An and W. K. Li, 2004b. A simple multivariate ARCH model specified by random coefficients, under revision, Journal of Time Series Analysis.
Remarks on the Use of High Performance Computing Cluster:
The high performance computers in the computer centre of the University of Hong Kong help a lot in supporting large scale simulation experiments. HPC Cluster can finish complicated calculations in a very short interval of time and perform stably for all intensive computation of test statistics. Fast, accurate, reliable, simple to use are the characteristics of HPC Cluster.
Email Address:
fongpw@hku.hk
- Back to top- / Contents