|
|
楼主

楼主 |
发表于 2009-3-31 21:04:57
|
只看该作者
请教检验结果p值是Biased的问题
运行[code:2rd6qsbm]proc model data=n1;
parms ma1 ma2 mu arch0 arch1 arch2 garch1 garch2;
et=mu+ma1*zlag1(resid.et)+ma2*zlag2(resid.et);
h.et=arch0+arch1*xlag1(resid.et**2,mse.et)+arch2*xlag2(resid.et**2,mse.et)+garch1*xlag1(h.et,mse.et)+garch2*xlag2(h.et,mse.et);
fit et/ method=marquardt fiml out=forcast outall normal;
run;
quit;[/code:2rd6qsbm]
结果输出:[img] Nonlinear FIML Parameter Estimates
Approx Approx
Parameter Estimate Std Err t Value Pr > |t|
ma1 0.228927 0.0513 <------ Biased
ma2 0.6842 0.0438 <------ Biased
mu 0.968339 0.3448 <------ Biased
arch0 4022.376 0 <------ Biased
arch1 0.210796 0.0754 <------ Biased
arch2 0.084113 0.0717 <------ Biased
garch1 -0.4796 0.1028 <------ Biased
garch2 0.275604 0.0802 <------ Biased
NOTE: The model was singular. Some estimates are marked 'Biased'.
[/img]请问这是什么原因呢? |
|