The RATS Software Forum

dear
i wrtie code to estimate mgarch-dcc model
the code runs well
but the the result displays ,
## MAT14. Non-invertible Matrix. Using Generalized Inverse for SYMMETRIC.
## MAT14. Non-invertible Matrix. Using Generalized Inverse for SYMMETRIC.
## MAT14. Non-invertible Matrix. Using Generalized Inverse for SYMMETRIC.
## MAT14. Non-invertible Matrix. Using Generalized Inverse for SYMMETRIC.

MAXIMIZE - Estimation by BFGS
Convergence in 33 Iterations. Final criterion was 0.0000045 <= 0.0000100
With Heteroscedasticity/Misspecification Adjusted Standard Errors
Usable Observations 970
Function Value 6301.19799169

Variable Coeff Std Error T-Stat Signif
*******************************************************************************
1. CC11 0.001156334 0.000365923 3.16004 0.00157745
2. CC21 0.002291302 0.000471012 4.86463 0.00000115
3. NU 7.155382227 0.877549337 8.15382 0.00000000
4. C11 0.000001083 0.000000841 1.28759 0.19788740
5. C22 0.000004070 0.000002480 1.64107 0.10078304
6. A11 0.052042460 0.012330547 4.22061 0.00002436
7. A22 0.074079511 0.013117581 5.64735 0.00000002
8. B1 0.903048728 0.008531016 105.85476 0.00000000
9. B2 0.911806608 0.020977772 43.46537 0.00000000
10. D12 -0.017779886 0.013386615 -1.32818 0.18411737
11. D21 0.002652646 0.024390119 0.10876 0.91339360
12. D1 0.077411478 0.027000665 2.86702 0.00414355
13. D2 0.005483785 0.026736698 0.20510 0.83749142
14. A 0.008746006 0.002454924 3.56264 0.00036715
15. B 0.988348565 0.003106940 318.10996 0.00000000
16. A12 0.017238873 0.009611748 1.79352 0.07288965
17. A21 0.005378686 0.015646302 0.34377 0.73102133
is this result validity?

You're probably fine. With just 4 warnings in 33 iterations and many hundreds of function evaluations, it would seem that you might have had some problems on the early iterations, but that they got straightened out. You might put a TRACE option on and see exactly where you get the warnings. If, as I suspect, they're on the first couple of iterations, your results are OK, but you might want to examine your guess values to see if you are feeding in a singular matrix.

Hi Luxu,
Can you please share your code I have the same issue with DCC Mgarch estimation.
Thanks.