Standard errors may be unstable in VARMAX AR(1)

45 views Asked by At

I ran a state space VARMAX with an AR(1) disturbance model with stochastic volatility using statsmodels, and while it attempts to converge, I receive two warnings when the results are displayed.

[1] Covariance matrix calculated using the outer product of gradients (complex-step).

[2] Covariance matrix is singular or near-singular, with condition number 9.69e+20. Standard errors may be unstable.

What do the warnings written upon VARMAX summary mean? Does it heavily affect my model performance and how to resolve it?

I suspect multicollinearity but given that I run big data VARMAX with related variables chances were to be present.

Can I still go ahead with the model under these conditions? I'm mostly interesting to analyze the time varying volatility and the functions Impulse responses of the model. Additionally, what does the second warning mean?

0

There are 0 answers