Dear DYNARE team,
I’ve been trying to re-estimate my model with the latest versions of DYNARE but enountered singularity problems when using mode_compute=6. On the other hand, these problems do not occur when using version 4.4.2.
Best,
Peter
Which type of singularity problems?
initial_estimation_checks:: The forecast error variance in the multivariate Kalman filter became singular.
initial_estimation_checks:: This is often a sign of stochastic singularity, but can also sometimes happen by chance
initial_estimation_checks:: for a particular combination of parameters and data realizations.
initial_estimation_checks:: If you think the latter is the case, you should try with different initial values for the estimated parameters.
Error using initial_estimation_checks (line 143)
That means the problem has been there before as well. But Dynare 4.4 automatically switched to the univariate Kalman filter, hiding the fundamental problem. You can revert to the old behavior with the
use_univariate_filters_if_singularity_is_detected=1
option, but this is strongly discouraged.
I also share this concern. 4.4.3 is way slower than 4.4.1 in McMC convergence when I use mode compute 6. It takes double time in my computer. I tried several computers. Same issue.
@pb2415 What exactly do you mean? Does the MCMC take longer for the same number of draws? Or does the mode_compute=6 take longer?