A characterization of random-coefficient AR(1) models (Q582792)

A random coefficient (RC) AR(1) process \(X_ t\) is defined by \(X_ t=\Lambda_ tX_{t-1}+\epsilon_ t\) where the i.i.d. random variables \((\Lambda_ t,\epsilon_ t)\) are independent of \((X_{t-i})_{i\geq 1}\). The transition probability T(.)(x) is given by \[ E(f(X_{t- 1})| X_{t-1}=x)=T(f)(x). \] The author gives a characterization of RCAR(1) processes based on the transition probabilities and shows that the transition probabilities can be used to find a solution of the equation \[ (\partial g/\partial t)(t,x)=-g(t,x)+\int g(t,\lambda x+u)d\eta (u),\quad g(0,x)=f(x). \]