Document Type

Article

Journal/Book Title/Conference

Economics Research Institute Study Paper

Volume

4

Publisher

Utah State University Department of Economics

Publication Date

1997

First Page

1

Last Page

42

Abstract

Exchange rates commonly exhibit periods of stability punctuated by infrequent, substantial adjustments. Statistically, this generates empirical distributions of exchange rate changes that have high peaks, long tails, and, sometimes, are asymmetric. Existing time-series estimation methods do not account for these characteristics satisfactorily. This paper introduces a more general GARCH model, based on the exponential generalized beta (EGB) family of distributions, which can accommodate most nonnormal characteristics of data, including leptokurtosis, skewness, and high peakedness, and yet remains tractable for estimation. Applied to daily U.S. dollar exchange rate data for six major currencies, the GARCH-EGB2 model uniformly outperforms conventional time-series models of exchange rate volatility.


Download

Share

COinS