Markov switching regression - singular covariance
Posted: Thu Jul 27, 2017 11:23 am
Hi,
I am trying to do Markov switching regression, but every time I set my initial regime probabilities to ‘estimated’, I get a warning in my output saying: “Singular covariance – coefficients are not unique”. When I set them to ‘ergodic’, there’s no problem. I am estimating an equation of the following form: y c x1 x2 y(-1) with regime specific error variances. The rest of the settings are kept at their default values. Does somebody have an idea what might be the issue? (I’m using eviews 9 btw). Thanks in advance!
Bert
I am trying to do Markov switching regression, but every time I set my initial regime probabilities to ‘estimated’, I get a warning in my output saying: “Singular covariance – coefficients are not unique”. When I set them to ‘ergodic’, there’s no problem. I am estimating an equation of the following form: y c x1 x2 y(-1) with regime specific error variances. The rest of the settings are kept at their default values. Does somebody have an idea what might be the issue? (I’m using eviews 9 btw). Thanks in advance!
Bert