請教關(guān)于工具變量中異方差已經(jīng)面板數(shù)據(jù)聯(lián)立方程

是 否 +2 論壇幣

k人 參與回答

經(jīng)管之家送您一份

應(yīng)屆畢業(yè)生專屬福利!

求職就業(yè)群

趙安豆老師微信：zhaoandou666

經(jīng)管之家聯(lián)合CDA

送您一個全額獎學(xué)金名額~ !

立即領(lǐng)取

感謝您參與論壇問題回答

經(jīng)管之家送您兩個論壇幣！

+2 論壇幣

請教大家?guī)讉�問題：

1、對于這樣一個面板數(shù)據(jù)的聯(lián)立方程應(yīng)該如何用3sls進行系統(tǒng)分析呢？xtivreg只能進行2sls分析。如果是截面數(shù)據(jù)可以用reg3命令，但是面板里面好像沒有xtreg3。
y1=x1+x2+y2
y2=x1+x3+y2
2、對于面板的聯(lián)立方程回歸出來的結(jié)果，如何判斷是否有異方差和自相關(guān)？如果存在，該如何解決？如果是單方程的，可以改用xtgls，p(h)解決異方差問題。但是對于聯(lián)立方程組怎么辦？也沒有xtivgls這個命令。
3、如果存在工具變量和被工具變量的交乘項該怎么處理？
y1=x1+x2+y2*x3
y2=x1+x3+y2

掃碼加我 拉你入群

請注明：姓名-公司-職位

以便審核進群資格，未注明則拒絕

分享0 收藏2 回帖

關(guān)鍵詞：工具變量 面板數(shù)據(jù) 聯(lián)立方程 異方差 XTIVREG 工具 方差 方程 面板 變量

Stata doesn't have a built-in routine for panel data simultaneous
equations. Maybe someone out there has written their own program for
this, but when I did -findit panel simultaneous- nothing came up.

If you can transform or recast your model into a non-panel framework
(e.g., by transforming into deviations from means or using explicit
dummies for observational units) you could use -reg3- to do 3-stage
least squares.

The alternative is to estimate the system equation-by-equation using
-xtreg- or -xtabond- or -xtivreg- or whatever. It won't be efficient
(in the way that a system estimator like -reg3- is) but your results
will be consistent, and will have the usual advantage over system
estimation in that if one equation is misspecified, it won't spill
over and contaminate the estimation results for the other equation.

Hope this helps.

What do you mean by not efficient?

xtsur