我也想知道，谁会阿……

用proc model,很强大的，％ma ％ar，用h.来设定garch模型，如

10. QGARCH
/* Estimate Quadratic GARCH (QGARCH) Model */
proc model data = index ;
parms arch0 .1 arch1 .2 garch1 .75 phi .2;
/* mean model */
i1 = intercept ;
/* variance model */
h.i1 = arch0 + arch1*xlag(resid.i1**2,mse.i1) + garch1*xlag(h.i1,mse.i1) +
phi*xlag(-resid.i1,mse.i1);
/* fit the model */
fit i1 / method = marquardt fiml ;
run ;