Computation of Multidimensional Conditional Distribution Function

Abstract

An important method of forming univariate time series models is by means of conditional distributions. For a vector having a power-normal distribution of dimension k, it is already possible to find approximately the conditional distribution of the last element in the vector. This paper extends the previous result on univariate conditional distribution to conditional distribution in the light of the values of the initial k-m elements in the vector. The extended results can be used to form multivariate time series models. The new numerical procedure offers an easy way to obtain satisfactory results.