[求助]garch模型的选择标准

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

在用eviews拟和garch模型中，怎么就是合理的，对变量的显著性要求，拟和度的选择标准如何？请高手指点！

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏0 回帖

关键词：GARCH模型 ARCH模型 GARCH 选择标准 ARCH 模型 如何

Maximum likelihood and BIC or AIC(there are many many others, but not for Eviews users)

then how to determine the (p,q) of GARCH if different (p,q)s can satify the MLE, AIC and BIC?

我觉得用AIC或SIC的方法是不妥的．

To ojsapp: there are some tests for higher order ARCH effect. But there is problems with testing against higher order GARCH effect(am I correct?). For this you can read Tim Bollerslev's original paper. If you want to consider asymmetric effect (the so called news impact), you can read Engle's paper (For example this is included in Engle's "ARCH models, Selected readings") which formalize three tests for asymmetry. .....

[此贴子已经被作者于2006-4-21 18:22:55编辑过]

Thank you, 掌门人. I will read the papers you recommend.

Also, do you know how to use EVIEW to estimate Garch parameters when assuming error is t-distributed? Suppose we know (p,q) already.

Also, can we use GMM to estimate Garch parameters? Normally we use MLE. What is the difference? Or can we use no-linear least square? Can we get the same results using these three methods?

Thanks in advance.

GMM is not usually chosen for general GARCH model. The essence for GMM is that modeller does not want to assume any distribution, just moment conditions. In general GMM is not appropriate for Forecasting (I am not very sure about this, but I will give an example later on). Similar to OLS, nonlinear least square also does not see the ditribution assumption as a must (of course when error term follows a normal one in OLS, common F and t tests become exact and convinient for statistical inference). In GARCH literature, MLE (QMLE) is the usual choice for parameter estimation. As we know that MLE requires assumption for error term. and this may be the exact reason why people model financial returns time series (high kurtosis, volatility clustering,...). In univariate GARCH models, fat-tailed distribution has been a routine choice except for some more theoretical works where relevant for the model with normal distribution has been readily available. In Eviews it is simple to choose t or GED(for v5?) as long as you are able to use Eviews. For Stochastic volatility model (SV) one can use GMM technique for parameter estimation but for volatility forecating, the power of GMM deserts us. So QMLE with Kalman fitler is the first choice for estimation and voaltility forecating. There are other methods for SV,e.g. Bayesian method and simulated maximumlikelihood method.   

[此贴子已经被作者于2006-4-22 20:13:41编辑过]

To : 掌门人

You are really expert in econometrics!

I have another question:

I want to determine the (p,q) in GARCH as follows:

Let p fixed (for exmaple : let p=1), then change q (1 to 9); then chage p nut still fixed (for exmaple : let p=2), then change q (1 to 9); .....................................

I want to determine (p,q) by comparing Lagrange Multiplier statistics. Do you think this approach approriately? Or I should use AIC and BIC to select the correct ARMA orders. Or some other criterions will works?

Thanks in advance.

rejusted of square R,AIC, SIC and so on.