By estimating a multivariate MS model, can we find out what's the forcing variable that causes the regime shifts?
What about in a univariate MS model?