The RATS Software Forum

Matrix P in MSVarSetup is a nstates-1 x nstates transition matrix. With three regimes, e.g., the procedure provides estimates and standard errors for the following:
Am I right that to get the nstates x nstates matrix, one needs to construct the last row as follows?


What about the standard errors, how can these be computed?

That's correct. You can use SUMMARIZE if you want to compute standard errors. See, for instance, the MHHP737.RPF example (Martin, Hurn and Harris textbook examples), which does a function of the transition probabilities in a 2 regime case.