Time-varying conditional Johnson Su density in Value-at-Risk methodology

Abstract

Value-at-Risk (VaR) is a standard method of forecasting future losses in a portfolio of financial assets. An alternative method of estimating VaR using time-varying conditional Johnson SU distribution is introduced in this paper, and the method is compared with other existing VaR models. Two estimation procedures using the Johnson distribution are developed in the paper: (1) the joint estimation of the volatility; and (2) the two-step procedure where estimation of the volatility is separated from the estimation of higher parameters, i.e., skewness and kurtosis. Empirical analyses of the two procedures are illustrated using data on foreign exchange rates and the Philippine Stock Exchange index. The methods are assessed using the standard forecast evaluation measures used in VaR models. Modeling procedures where estimation of higher parameters can be integrated in VaR methodology are introduced in the paper. 

JEL classification: C22, C58, G12, G32