|
Abstract:
|
The skewed generalized t (SGT), the exponential generalized beta of the second kind (EGB2), the inverse hyperbolic sine (IHS), the skewed generalized error (SGED), and the g-and-h distributions are flexible parametric probability density functions (pdf's) which can accommodate a wide variety of distributional characteristics. This paper explores possible advantages and disadvantages associated with the g-and-h distribution: though it has been noted for its relatively large skewness-kurtosis space, the lack of a closed-form pdf for the g-and-h distribution complicates traditional estimation methods. Monte Carlo simulations are used to compare the relative performance of maximum-likelihood, method of moments, and quantile methods, and empirical daily, weekly, and monthly stock return data are used to compare the alternative models and estimation methods.
|
