面板平稳性检验可以休矣?

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

且不说各检验的一堆假设，要么原假设为
所有截面都不平稳（ Levin–Lin–Chu (2002), Harris–Tzavalis (1999), Breitung (2000; Breitung and Das 2005), Im–Pesaran–Shin (2003), and Fisher-type (Choi 2001) tests ），
要么所有截面都平稳（Hadri (2000) Lagrange multiplier (LM) test ）
而做微观研究往往有一大堆截面个体，这种检验让人倍受折磨
更重要的是我们处理的一般为非平衡面板，阉割成平衡的不甘心，有时也没办法。而对于非平衡面板的平稳性，Eviews厚着脸给出了LL*****PPF等结果，但是STATA的大姥姥是怎么说的呢，请看下面：

Date: Wed, 11 Aug 2004 15:41:45 -0400
From: Kit Baum <baum@bc.edu>
Subject: st: Re: Unit root tests for unbalanced data panels in Stata

Dear Jean
This question comes up from time to time on Statalist. I do not believe
that any of those routines will work with unbalanced panels. Some of
the underlying analytics for some of the tests may be compatible with
unbalanced panels (or at least timeseries without gaps) but none of the
routines {hadrilm, ipshin, levinlin, madfuller, nharvey} are
implemented to allow for unbalanced panels.
Thanks
Kit Baum

On Aug 11, 2004, at 3:06 PM, Salvati, Jean wrote:

> Dear Mister Baum,
>
> I work at the IMF, in the Econometric Support group. I saw in Sata's
> online help that you have written several ado files to perform unit
> root tests on panel data. Do these programs work with unbalanced
> panels? If not, do you know any program that supports unbalanced
> panels?
>
> Thanks a lot.
>
> Best regards,
>
> Jean Salvati
>  Econometric Support
>  (202) 623-7804
>  IS 12-1328

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1 回帖

关键词：平稳性检验 平稳性 Econometric unbalanced underlying 平衡

本来这种检验的局限性就很大啊。你看null的设置就知道了。都是一竿子打翻一船人的啊。再说就算unbalanced不好办，也不能说“面板平稳性检验可以休矣”。另外，没有好的检验正是未来研究的空间啊~（话说这个怎么是04年的啊。。。）

另外，这个供参考：

1. st: Unit root tests for unbalanced data panels in Stata
   
2. From           "Gindo Tampubolon" <gindo.tampubolon@man.ac.uk>
   
3. To           statalist@hsphsun2.harvard.edu
   
4. Subject           st: Unit root tests for unbalanced data panels in Stata
   
5. Date           Thu, 12 Aug 2004 09:25:31 +0100
   
6. 
   
7. Hi all,
   
8. 
   
9. since this questions come up quite often in Stata list, may I suggest Maddala&Wu
   
10. 1999, esp p 636ff? It has the virtue of simplicity and one can probably do it by hand.
    
11. 
    
12. Maddala & Wu 1999. A comparative study of unit root tests with panel data and a
    
13. new simple test. Ox. Bull. Ec. Stats. Special Issue.
    
14. 
    
15. Gindo
    
16. 
    
17. 
    
18. >>>>>>>>>>>>>>>>>>>>>>>>
    
19. Date: Wed, 11 Aug 2004 15:41:45 -0400
    
20. From: Kit Baum <baum@bc.edu>
    
21. Subject: st: Re: Unit root tests for unbalanced data panels in Stata
    
22. 
    
23. Dear Jean
    
24. This question comes up from time to time on Statalist. I do not believe
    
25. that any of those routines will work with unbalanced panels. Some of
    
26. the underlying analytics for some of the tests may be compatible with
    
27. unbalanced panels (or at least timeseries without gaps) but none of the
    
28. routines {hadrilm, ipshin, levinlin, madfuller, nharvey} are
    
29. implemented to allow for unbalanced panels.
    
30. Thanks
    
31. Kit Baum
    
32. 
    
33. On Aug 11, 2004, at 3:06 PM, Salvati, Jean wrote:
    
34. 
    
35. > Dear Mister Baum,
    
36. >
    
37. > I work at the IMF, in the Econometric Support group. I saw in Sata's
    
38. > online help that you have written several ado files to perform unit
    
39. > root tests on panel data. Do these programs work with unbalanced
    
40. > panels? If not, do you know any program that supports unbalanced
    
41. > panels?
    
42. >
    
43. > Thanks a lot.
    
44. >
    
45. > Best regards,
    
46. >
    
47. > Jean Salvati
    
48. >  Econometric Support
    
49. >  (202) 623-7804
    
50. >  IS 12-1328
    
51. 
    
52. Dr. Gindo Tampubolon
    
53. gindo.tampubolon@man.ac.uk
    
54. International Data Service -
    
55. Economic and Social Data Services (Funded by ESRC & JISC)
    
56. - and -
    
57. Department of Sociology
    
58. University of Manchester

复制代码

stata也有可以处理非平衡面板数据的，可以看下帮助
不过大多数都是平衡面板

The assorted tests make different asymptotic assumptions regarding the number of panels in your dataset and the number of time periods in each panel. xtunitroot has all your bases covered, including tests appropriate for datasets with a large number of panels and few time periods, datasets with few panels but many time periods, and datasets with many panels and many time periods. The majority of the tests assume that you have a balanced panel dataset, but the Im–Pesaran–Shin and Fisher-type tests allow for unbalanced panels.

5# ywh19860616

两个F检验确实可用，但是IPS的使用在在有time gap时就用不了了，比如我想使用2000-2010的上市公司的季度数据，但是2002年前只有半年度的数据，这样IPS就要罢工了。
再比如说要检验残差的平稳性，如果使用了动态面板，那自然残差就有Gap了，于是IPS又清闲了
最可恨的是，好不容易把Sargan弄伏贴了，残差平稳性出问题了

哎，愚钝，看不懂

我一般不用 觉得单位根这东西很扯淡

我一般不用 觉得单位根这东西很扯淡

楼上高见,但是要发文章也只能硬着头皮死磕了