Dynamic Copula Toolbox 3.0

Functions to estimate copula GARCH and copula Vine models.

Description

Updates from version 2.0:

1. The marginal GARCH models are estimated from the toolbox functions (without the use of the econometrics/GARCH toolbox of MATLAB).
2. Hansen's Skew t distribution for the margins is supported.
3. Asymptotic standard errors are computed (Godambe info. matrix)

• Acknowledgements: This submission has inspired the following:; fitparp function
• Required Products: Optimization Toolbox; Statistics Toolbox
• MATLAB release: MATLAB 7.7 (R2008b)

[Stee Huang]

Dear author,

Thanks for your nice work.
However, I find some problems when using this toolbox.
The wrong messages are follows:

??? Undefined function or method 'hessian_2sided' for input arguments of type 'struct'.

Error in ==> CalcStErrors at 39
Hnum = hessian_2sided(MyFunc,theta,data,spec,'fmincon'); % calculates the Hessian

Error in ==> fitModel at 185
                   [derivatives, RobVCV, VCV, hessian, RobStE]=CalcStErrors('CopulaGARCHLogL',
                   parameters, data, grad, hessian, spec, 'fmincon');

I truely hope you can deal with the above problem.

[Manthos Vogiatzoglou]

Dear Stee

Sorry for the inconvenience. I just forgot to add the hessian_2sided function to the toolbox. I will fix it tomorrow. Thanks for the rating and comments

[Paula]

Dear author,
Excellent work! It has helped me a lot.
I just would like to mention two minor difficulties I had:
1) I was not able to estimate a vine copula sequentially using the menu. I had to use the function “SeqFitCopVine.m”.
2) I could not find “hfuncJC.m” in the third version of the toolbox. I think it’s missing. I had to copy this from version 2.0.
Thanks for your work.

[Paula]

Dear author,
Just correcting my first comment above: I was not able to estimate a vine copula sequentially using “modelspec” (and the menu) and “fitModel” after that. I had to use “SeqFitCopVine.m” instead of “fitModel” and this is not specified in the Tutorial.
Thanks again.

[wenhua yu]

nice work! this program help me a lot~~~

[Behdad]

 

[charles yang]

Dear Author,
Thanks for your work, now I meet a problem with FitModel.m as:
[parameters, LogL, evalmodel, GradHess, varargout] = fitModel(spec, out1, fminunc)
??? Error using ==> fminunc at 156
FMINUNC requires two input arguments.

I am still confusing with the data input way, because I already transformed the stock return into empirical CDF through empirical.m in out1 and just put out1 instead of data in the FitModel.m function, you see the error comes out as FMINUNC requires two input arguments. So could you show me how to input the data in a correct way?

[Yin]

Dear author,
When I ran this code , I met this problem. I tried to find out the reason, but failed. The data here is one stock return. I want to fit a GARCH(1,1)model.
>> spec = modelspec(data)
>> fitModel(spec,data,'fminunc')
??? Operands to the || and && operators must be convertible to logical scalar
values.

Error in ==> FromCons2Unc at 41
        if max(max(x))> a || min(min(x))<-a

Error in ==> RescaleParameters at 116
            [uncmparams, d1mp, d2mp] = FromCons2Unc(consmparams,.5,3);

Error in ==> fitModel at 86
                theta0 = RescaleParameters(theta0, 2, spec);

[Yin]

I found out by myself ,sorry to disturb u.
Anyway, it is really a good and helpful toolbox!!!
Thanks for the Author!