Abstract

This paper considers the application of the generalized autoregressive conditional heteroscedasticity process in the estimation of volatility in the Kenyan exchange rates. A quasi-maximum likelihood estimation procedure is used and asymptotic properties of the estimators given. Exploratory data analysis performed indicates the returns are heavy tailed. It is found that the estimated model fits the exchange rates return data well.

Key words: Volatility, exchange, returns, autoregressive, heteroscedasticity, likelihood, quasi, maximum, estimator.