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 on April 23, and its replacement, statalist.org is already up and running.


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

st: Problem with standard errors using eststo


From   Sutirtha Bagchi <sutirthabagchi@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   st: Problem with standard errors using eststo
Date   Wed, 6 Feb 2013 09:33:30 -0500

I am having a problem with standard errors using eststo.

My problem is that if I estimate a regression without using eststo and
then estimate exactly the same regression with eststo, my standard
errors change. This is a problem which had been raised about a year
ago in this post:
http://www.stata.com/statalist/archive/2012-01/msg00085.html and
although the final verdict on this from Nick Cox was that "-eststo- is
not
in any sense changing standard errors and that your main problem lies
elsewhere", I am not so sure that the advice holds in my case. In any
event, I am attaching the two lines of code that I run and the output
that I get from those. Please note how the standard errors (and the
associated t-stats) change when I use eststo in the second line of
code.

Line of code #1: xi: xtreg fundratio compttve7, fe cluster(county)

Line of code #2: eststo: xi: xtreg fundratio compttve7, fe cluster(county)

Output from line of code #1:
Fixed-effects (within) regression               Number of obs        =      3385
Group variable: municipali~m                    Number of groups   =      1262

R-sq:  within  = 0.0266                         Obs per group: min =         1
          between = 0.0000
avg =       2.7
          overall = 0.0042
   max =         3

                                                         F(1,64)
      =     12.52
corr(u_i, Xb)  = -0.1392                        Prob > F           =    0.0008

                                (Std. Err. adjusted for 65 clusters in county)
------------------------------------------------------------------------------
                |               Robust
   fundratio |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
compttve7 |  -336.4875   95.10119    -3.54   0.001    -526.4739    -146.501
      _cons |   115.1093   10.24336    11.24   0.000     94.64585    135.5728
-------------+----------------------------------------------------------------
  sigma_u |  116.71559
  sigma_e |  91.487051
          rho |  .61941958   (fraction of variance due to u_i)
------------------------------------------------------------------------------

Output from line of code #2:

Fixed-effects (within) regression                Number of obs       =      3385
Group variable: municipali~m                    Number of groups   =      1262

R-sq:  within  = 0.0266                         Obs per group: min =         1
          between = 0.0000
avg =       2.7
          overall = 0.0042
  max =         3

                                                         F(1,64)
     =      7.85
corr(u_i, Xb)  = -0.1392                        Prob > F           =    0.0067

                                (Std. Err. adjusted for 65 clusters in county)
------------------------------------------------------------------------------
               |               Robust
  fundratio |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
compttve7 |  -336.4875   120.0782    -2.80   0.007    -576.3713   -96.60364
      _cons |   115.1093   12.93364     8.90   0.000      89.2714    140.9472
-------------+----------------------------------------------------------------
     sigma_u |  116.71559
     sigma_e |  91.487051
             rho |  .61941958   (fraction of variance due to u_i)
------------------------------------------------------------------------------
(est146 stored)

I should also mention that this problem comes up only when I am using
the cluster option. That is, when I run:
xi: xtreg fundratio compttve7, fe and eststo: xi: xtreg fundratio
compttve7, fe, I get exactly the same results (i.e. same standard
errors and same t-stats). Also, I have updated all my packages using
adoupdate and given the status message: "(no packages require
updating)"

I look forward to hearing the listserv's solutions to the problem.
Thank you,
Sutirtha Bagchi
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


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