Conditional characterizations of multivariate distributions (Q1110216)

Let (X,Y) be a bivariate random varialbe with \(X=^{d}Y\). Denote the family of conditional distributions of X given Y by \(\phi\) (x\(| y)\). Knowledge of \(\phi\) (x\(| y)\) determines the joint distribution of (X,Y) in such a setting. This observation provides an alternative proof of a bivariate normal characterization first proved by \textit{M. Ahsanullah} [ibid. 32, 215-218 (1985; Zbl 0609.62019)]. Many analogous characterization results can be enumerated. Some examples are provided. Multivariate extensions are discussed. For example it is shown that a random vector \(\underset \tilde{} X_ n=(X_ 1,...,X_ n)\) has a multivariate normal distribution, if and only if \(X_ 1,X_ 2\) belong to a location-scale family and for \(i=2,3,...,n\) the conditional distribution of \(X_ i\) given \(X_{i-1}=x_{i- 1},...,X_ 1=x_ 1\) is \[ N(\beta_ i+\sum^{i- 1}_{j=1}\alpha_{j,i-1}x_ j,\quad \sigma^ 2_ i),\quad for\quad all\quad real\quad x_ 1,...,x_{i-1}. \] Also, we show that a conjecture of M. Ahsanullah concerning multivariate normality of \(X_ n\) is incorrect.