Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: RE: RE: Random effect model same as OLS in dynamic model


From   Nick Cox <njcoxstata@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: RE: RE: Random effect model same as OLS in dynamic model
Date   Thu, 29 Mar 2012 14:09:34 +0100

This can only be answered by those familiar both with LIMDEP and Stata
and with the models in question (not me).

They would ideally need to have access to

1. One or more datasets for which apparently discrepant results have
been obtained.

2. The exact command syntax used in each case so it can ascertained
whether like is being compared with like.

3. The exact displayed results from both programs.

I don't think any serious discussion is likely otherwise.

Nick

On Thu, Mar 29, 2012 at 1:07 PM, Chi-hong Tsai <C.Tsai@econ.usyd.edu.au> wrote:

> I have seen these previous discussions and have a good idea why it happens in Stata.
> But what still bothers me is the different outcomes from Stata and Limdep.
> In Limdep, I can get a reasonable RE estimation result with Sigma_u > 0.
> Is it because the algorithm behind these two packages are different?
> If so, may I use the RE estimation results from Limdep and compare it with other estimators in Stata?

[snip]

chihongt

>> I am running OLS,FE,and RE models on my panel data set.
>> A strange thing happens when I use RE in the dynamic model
>> (with one lagged variable).
>> That is, the sigma_u appears to be zero which means rho=0,
>> and thus the estimates and s.e. are exactly the same as OLS.
>> This only happens when I include the lagged variable in the
>> dynamic model but not in the static model.
>>
>> And when I use Limdep on the same model with the same data
>> set, RE gives different results from OLS.
>>
>> I have had a search on Statalist, for example,
>> http://statalist.1588530.n2.nabble.com/Same-results-OLS-and-Ra
>> ndom-Effects-td5199084.html
>>
>> but it didn't give a suggestion to deal with this problem.
>>
>> Below is my RE regression results.
>>
>> So my two questions are:
>> (1) Why sigma_u=0 in the RE dynamic model?  (pls don't tell
>> me I should just use FE or GMM which I have already done. I
>> want to do an exploratory
>> analysis)
>>
>> (2) Why Stata and Limdep give different results for the same
>> model and data set ( even in OLS the estimates are slightly
>> different)?
>>
>> Anyone has any idea?
>>
>> Patrick
>>
>>
>> Random-effects GLS regression                   Number of obs
>>      =
>> 236
>> Group variable: group20                         Number of
>> groups   =
>> 20
>>
>> R-sq:  within  = 0.1554                         Obs per
>> group: min =
>> 11
>>        between = 0.9752
>>  avg =
>> 11.8
>>        overall = 0.8220
>>  max =
>> 12
>>
>>                                                 Wald chi2(7)       =
>> 1052.77
>> corr(u_i, X)   = 0 (assumed)                    Prob > chi2
>>      =
>> 0.0000
>>
>> --------------------------------------------------------------
>> ----------------
>>       pttrip |      Coef.   Std. Err.      z    P>|z|     [95% Conf.
>> Interval]
>> -------------+------------------------------------------------
>> ----------
>> -------------+------
>>       pttrip |
>>          L1. |   .5171986   .0571405     9.05   0.000     .4052053
>> .629192
>>              |
>>        price |  -.0651085   .0299721    -2.17   0.030    -.1238527
>> -.0063643
>>      pincome |  -1.364764   .7551779    -1.81   0.071    -2.844885
>> .1153577
>>         age1 |  -3.508893   .9180143    -3.82   0.000    -5.308168
>> -1.709618
>>  pdensity_cd |   .0117259   .0053559     2.19   0.029     .0012285
>> .0222233
>>  edensity_tz |   .0012031   .0003378     3.56   0.000      .000541
>> .0018652
>>   walkshare2 |   .0495796    .035056     1.41   0.157    -.0191288
>> .118288
>>        _cons |   .3881918   .0934927     4.15   0.000     .2049494
>> .5714342
>> -------------+------------------------------------------------
>> ----------
>> -------------+------
>>      sigma_u |          0
>>      sigma_e |  .10070681
>>          rho |          0   (fraction of variance due to u_i)

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index