epoh 发表于 2011-7-5 11:51
R package "mgarch"
matlab toolbox Ucsd_garch
都没有conditional mean equation
因为数据都已事先处理.
[parameters, .....] = full_bekk_mvgarch(data,p,q, BEKKoptions)
    data - A zero mean t by k vector of residuals from some filtration

如果你需要conditional mean equation
就用s-plus
跑出的结果,可能更适合去理解.
底下是例子

%%%%%%in matlab
      0.0895     -0.0204      0.1274      0.1987      -0.1812     0.0729   
      0.3619      0.9776      0.0534     -0.0200       0.9169

######in s-plus
bekk=mgarch(y~-1, ~bekk(1,1))

bekk$coef
                      COEF
       A(1, 1)  0.09397350
       A(2, 1) -0.01728920
       A(2, 2)  0.12951584
ARCH(1; 1, 1)  0.20703205
ARCH(1; 2, 1) -0.17501221
ARCH(1; 1, 2)  0.06207963
ARCH(1; 2, 2)  0.36299024
GARCH(1; 1, 1)  0.97544367
GARCH(1; 2, 1)  0.05209714
GARCH(1; 1, 2) -0.01631104
GARCH(1; 2, 2)  0.91769227

######in s-plus include conditional mean
bekk=mgarch(y~1, ~bekk(1,1))

Coefficients:
                        
          C(1)  0.02087
          C(2)  0.03257
       A(1, 1)  0.09759
       A(2, 1) -0.01035
       A(2, 2)  0.12927
ARCH(1; 1, 1)  0.21222
ARCH(1; 2, 1) -0.17089
ARCH(1; 1, 2)  0.05736
ARCH(1; 2, 2)  0.36167
GARCH(1; 1, 1)  0.97385
GARCH(1; 2, 1)  0.05036
GARCH(1; 1, 2) -0.01490
GARCH(1; 2, 2)  0.91893


你说的A,B,C代表什么,总该注明一下吧
A: Arenas  ?
  B: Bryant   ?
  C: Carter   ?