To zhaosweden

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

To zhaosweden:

How to test the serial correlation of residuals? That is : how to test that the standardized residuals are uncorrelated? Do we use Q statistics? If so, how to determine the conventional level?

How to test GARCH effects? That is : HOW TO TEST that squared residuals are uncorrelated? Do we use t-test for individual parameters and F-test for the null hypo that the cofficient on all laged values are jointly equal to zero? It is enough or we also need Q statistics? If so, how to determine the conventional level?

Also, how to test the leverage effects? Do we firstly use t-test for individual parameters, then use F-test for the null hypo that the cofficient on all laged values are jointly equal to zero, then use Engle-Ng sigh bias test?

I hope your answer. Mnay thanks in advance.

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：Sweden AOS WED Conventional Standardized zhaosweden

How to test the serial correlation of residuals?: Q; Ljung-box test.

How to test GARCH effects?:(1) You can also use the same test statistic but with the squared returns, say. But note that this is a portmateau test, which means that it doesnot have a specific H1(alternative hypothesis). So it may not have ideal power. original form is chi-square, better to use F-form to expect a improved small sample property. (the Lm test of N*R^2 form, or F form)

How to test GARCH effects? (2) Engle's ARCH effect test works for ARCH and GARCH models. Wooldridge (regression-based) has a robust version of this test (robust against to heterscedasticity?). there is also some other technoques (some Austrian guy): robust against the distribution based on probability integral/inverse transform.

test the leverage effects?: Engle and Ng 1993 J of Finance, yes.

I am sorry for that I can not go into the detail since it take me a lot of time and this is BBS where everything is provided for free.

What I also want to say is that BBS is is not a good place to learn econometrics, it is always the textbooks or journal articles.

Furthermore I can not understand "conventional level", significance level? confidence level?

Good luck

强悍！英语都这么好啊！

Thanks zhao. You can say "go to ...textbook and .....journal paper to find your answer".

You can say "go to ...textbook and .....journal paper to find your answer".

======================

As a matter of fact, this is correct. My supervisor always tell : you can see ### 1999 paper to have the answer. Nothing wrong with this. When I told him that there is something wrong my GAUSS code, he says: check it again.

======================

I think it is perfect right to have such answers. The point is he/she tells you which paper which textbook you should read at the first place.