有谁懂面板数据联立方程

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

有谁懂面板数据联立方程？如何用软件来做，我问过很多人，没有一上人会。求高人指点

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3 回帖

关键词：联立方程 面板数据 求高人指点 高人指点 如何用 数据 方程 面板

Seemingly Unrelated Regression?  Not so sure....

在易丹辉的《数据分析与eviews应用》有详细介绍。

那仅仅是面板数据和联立方程的作法。

我说的是用面板数据构建的联立方程的作法。

难道国内真没有人会做吗？

我见过外国人做的结果，但不知道是怎么做出来的。

请大家问问各自的导师，看看能不能做，我这里先行谢过了。

真没有高人吗

At first make clear your question.

Your model has several equations and the data set is panel data. You want to know how to estimate this model.

This is SEM (Simultaneous-Equations Model). You can use 2sls or 3sls to estimate it whatever the data set is panel or not.

But the problem is if the data set is panel, the heteroscedasticity problems are always exist. To increase efficiency, you can use 2SIS-GLS or 3SLS-GlS to estimate this model, here the structure of error terms of GLS estimates follows the property of panel data-- ei`*ei/T. i is the number of groups and T is the number of periods. (chapter 13 Greene)

If you want to calculate the fixed effect in this model, you only need to add dummy variables for each group.(remember fixed effect model is consistant but not efficient if Heteroscedasticity exists. actually it is an OLS with dummy variables)

I do not know which software can do this directly, But SAS/IML can do it by programming.

You can refer the greene's book-- econometric analysis( fifth version chapter 15).

If you have any comments or opinions, please let me know.

感谢楼上的朋友，你理解了我的意思。但我还有一个疑问，一是如果要求随机效应怎么办，如何判断是固定效应还是随机效应？如果要用EVIEWS来做的话，是在面板对象里作还是在系统对象里作呢？

谢谢！

请高手指教

原理和做FEM 是一样的， 区别只在于 当你要做REM， 你要构造适当的OMIGA 矩阵

1. See Greene Page 406 and equation (15-31) for estimating model. (English fifth version )

2. See (15-28) and (14-3) which is how to construct Omega matrix for original 3SLS.

3. See (13-20) and 13-21 for How to construct Omega matrix of REM.

Substitute 3 into 2, and using 1 to estimate your model.

SAS/IML can do it, but you have to write code.

you can still use Hauman test to compare FEM and REM. When you compute the difference of variance and covariance matrix, you should use the Omega matrix you construct as 3.

I don't know how to use Eview.