NA for std.err

[LAI9525]

Hi Onno,

When I run fit_mfgarch() with multi.start=FALSE, I received NA for both rob.std.err and opg.std.err.
I tried multi.start=TRUE, but receive the error message
"in class(par.max.lik.nm) != "try-error" && par.max.lik.nm$maximum > : 'length = 2' in coercion to 'logical(1)'"

Can you please advice me how to fix this problem?

Best,
L

[onnokleen]

That typically means that the matrix inversion for the standard errors doesn't succeed/runs into numerical problems. Have you tried submitting the returns in % (i.e., multiplying by 100)? If that doesn't help, I cannot help you any further at this moment in time.

[LAI9525]

Onno, thanks a lot for your prompt response. I'm afraid I've already tried your solution, but unfortunately, it doesn't work in my case. Anyway, thank you.