Why FIGARCH Estimation May Return No Statistics

Summary

The document addresses a reported difference in output from FIGARCH estimation: one specification displays coefficients and inferential statistics, while another displays only coefficients. The accepted response explains that the latter estimation did not converge under quasi maximum likelihood estimation. When the optimizer cannot find stable parameter estimates, the estimation fails and associated statistics such as standard errors and t-statistics are unavailable.

The practical lesson is that missing output can indicate an estimation failure rather than a display setting problem. The response notes that quasi maximum likelihood estimation does not always produce a result. It does not offer troubleshooting steps, alternative starting values, software guidance, or advice for assessing convergence, so the explanation is limited to identifying the likely cause in the stated example.

Key ideas

• A failed FIGARCH estimation can leave coefficient statistics unavailable.
• The response attributes the missing statistics to non-convergence under quasi maximum likelihood estimation.
• Quasi maximum likelihood estimation does not guarantee stable parameter estimates for every specification.
• The document does not provide software instructions or methods for resolving convergence problems.

Tags

# Disappear Standard Error in OxEdit/G@rch6 package


# Disappear Standard Error in OxEdit/G@rch6 package












Hellow everyone, I'm new here. Please instruct me to do something.

My problem is when I run FIGARCH(0,d,1), OxEdit still show me a matrix with variable names, coefficient, s.e, t-stat... like this

But when I try FIGARCH(1,d,2) it show nothing but coefficient parameters, like this

So please instruct me how to show the other stats.

Furthermore, please instruct me how to run G@rch6 package in R via function GarchOxFit, because this function is no longer supported but Mr. Brian in this topic show that R can run G@rch6.

I much appreciate your help. Thank you.

## Answer by Malick (score 1, accepted)

https://quant.stackexchange.com/a/24788

Regarding your first question, as it is written in your example, the FIGARCH(1,d,2) estimation fails due to "no convergence" : ie the quasi Maximum Likelihood Estimation Method fails to obtain stable parameters via the maximization of the likelihood and so you can't get parameters for this specification (neither other stats). The QMLE method does not always get result.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.