Forecasting using SAS Forecast Server, SAS/ETS, and more

GARCH form=bekk

Reply
Occasional Contributor
Posts: 8

GARCH form=bekk

Dear all,

I am aiming to model volatility spillovers between two price series. As such, I need a bivariate GARCH. Volatility spillovers are usually modeled by means of GARCH BEKK. SAS gives some background on the GARCH BEKK module it has available here: SAS/ETS(R) 9.2 User's Guide The off-diagonal elements of matrices A and G  (see link to SAS user's guide for formula) capture the volatility spillovers the two price series.  E.g. see this paper: http://journal.fsv.cuni.cz/storage/1199_str_519_533_-_fedorova.pdf for how to interpret a GARCH BEKK.


I started estimating my data using:

proc varmax data=Libr.BZFD;

nloptions tech=qn maxiter=500 maxfunc=15000;

model var1 var2; garch q=1 p=1 form=bekk;

run;


For the datasets that SAS manages to do the optimization, usually I get strange results, see underneath. Anybody has tips how to improve this analysis? I attached the data (+-3000obs).

                 GARCH Model Parameter Estimates

                                                             Standard

                         Parameter  Estimate        Error       t Value     Pr > |t|

               GCHC1_1            0.00045         0.00001      34.82      0.0001

               GCHC1_2            0.00005         0.00002       2.30      0.0217

               GCHC2_2            0.00002         0.00000

               ACH1_1_1           0.22444         0.03146       7.14      0.0001

               ACH1_2_1          -0.00391         0.00913      -0.43      0.6682

               ACH1_1_2           0.02691         0.03625       0.74      0.4579

               ACH1_2_2           0.50595         0.02347      21.56      0.0001

               GCH1_1_1          -0.00195         0.00000

               GCH1_2_1          -0.00186         0.00841      -0.22      0.8253

               GCH1_1_2           0.00146         0.00000

               GCH1_2_2           0.89927         0.00761     118.11      0.0001


Any help would be much appreciated.

Sincerely,

Daan

Attachment
Ask a Question
Discussion stats
  • 0 replies
  • 790 views
  • 0 likes
  • 1 in conversation