This paper provides a new Bayesian approach for models with multiple c
hange points. The centerpiece of the approach is a formulation of the
change-point model in terms of a latent discrete state variable that i
ndicates the regime from which a particular observation has been drawn
. This state variable is specified to evolve according to a discrete-t
ime discrete-state Markov process with the transition probabilities co
nstrained so that the state variable can either stay at the current va
lue or jump to the next higher value. This parameterization exactly re
produces the change point model. The model is estimated by Markov chai
n Monte Carlo methods using an approach that is based on Chib (1996).
This methodology is quite valuable since it allows for the fitting of
more complex change paint models than was possible before. Methods for
the computation of Bayes factors are also developed. All the techniqu
es are illustrated using simulated and real data sets. (C) 1998 Publis
hed by Elsevier Science S.A. All rights reserved.