Applications of Asymmetric GARCH Models with Various Conditional Distributions:The Empirical Case of the NASDAQ Computer Index Daily Closing Returns

Open Access
Li, Ran
Area of Honors:
Interdisciplinary in Mathematics and Statistics
Bachelor of Science
Document Type:
Thesis Supervisors:
  • Jason Ryder Morton, Thesis Supervisor
  • David Russell Hunter, Honors Advisor
  • Sergei Tabachnikov, Honors Advisor
  • GARCH Model
  • Asymmetric
  • Conditional Distribution
The purpose of this honors thesis is to find an appropriate GARCH (Generalized Autoregressive Conditional Heteroskedasticity) Model for the NASDAQ Computer Index Daily Closing Returns, given a ten-year time series of closing prices. On the one hand, Standard GARCH Models are not sufficient enough, if consider the leverage effects (the volatility responds to good news and bad news differently). Instead, asymmetric GARCH Models are better, and, in particular, Exponential GARCH (EGARCH) Model is the best. On the other hand, EGARCH Models with alternative conditional distributions perform better than that with the default Normal Conditional Distribution. In particular, the Generalized Hyperbolic Distribution is found to be good fit that generate large P-values against the null hypotheses in the various tests. In conclusion, among all of the models investigated, the EGARCH Model with the Skew Generalized Error Distribution is the best.