Ridder, Geert; Montfort, Kees van - 1998
We show that a sufficient condition for the identification ofall parameters of the censored regression model with astochastic and unobserved threshold is that the errors are jointlynormally distributed. Exclusion restrictions are not needed.