Loading...

Using Student’s t Autoregressive (STAR) to Model Financial Variables of Iran

Fekrazad, Amir | 2011

462 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 41840 (44)
  4. University: Sharif University of Technology
  5. Department: Management and Economics
  6. Advisor(s): Souri, Davoud
  7. Abstract:
  8. Time series of asset returns display specific regularities such as bell-shaped distribution, leptokurticity and volatility clustering. Economists have made continuous efforts to develop models that explain these patterns and can be used to predict the return and the risk of holding an asset. These efforts can be classified into 3 eras: Bachelier Era (1900-1960) in which the random walk model was developed for speculative prices. Mandelbrot Era (1960-1980) in which the normality assumption was replaced with the Pareto-Levy family of distributions which are flexible enough to justify leptokurticity and infinite variance. And finally, the Dynamic Volatility era in which the focus was on conditional variance and during which the ARCH family of models were born and developed.
    In works done in the area of financial econometrics in Iran, the ARCH family is used extensively but these models are subject to some critics. For instance, the conditional mean and conditional variance are specified separately, even though they are the first two moments of the same conditional distribution. In addition, the functional form of the conditional variance is ad-hoc. In this thesis, Student’s t Autoregressive (STAR) is introduced as a replacement for ARCH family. This model is built based on Probabilistic Reduction approach and effectively deals with the theoretical problems of ARCH models. We then model the weekly return of the Euro/Rial exchange rate and the weekly return of the Golden Coin using STAR. Misspecification tests show the adequacy of STAR, and that STAR is at least as good as GARCH(1,1) for modeling these variables
  9. Keywords:
  10. Turbulence ; General Autoregressive Conditional Heteroskedastic (GARCH) ; Autoregressive Method ; Probabilistic Reduction ; Student T Distribution

 Digital Object List

 Bookmark

No TOC