Hi Mark,
Thank you so much for your help, and your prompt response, that's
exactly what is happening, I just checked with my sample as well. In
addition David Drukker has e-mailed me saying that if I specify the
option small I will get an accurate F test, which is what I was wanting
to get anyway.
Thank you again,
Sandra
***********************************************
Sandra Mortal
Assistant Professor of Finance
516 Cornell Hall
University of Missouri
Columbia, MO 65211
Office: (573) 884-1684
Fax: (573) 884-6296
***********************************************
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Schaffer,
Mark E
Sent: Thursday, August 03, 2006 9:47 AM
To: [email protected]
Subject: st: RE: FW: Xtivreg and Chi-Square
Sandra,
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of Mortal,
> Sandra C.
> Sent: 03 August 2006 15:26
> To: [email protected]
> Cc: Lipson, Marc
> Subject: st: FW: Xtivreg and Chi-Square
>
> Hi,
>
> I am using the command xtivreg, I have an unbalanced panel and am
> using fixed effects. My question is regarding the overall goodness of
> fit of the regressors. STATA reports a Wald Chi Square statistic of
> 434634.68.
> However, when I use the test command to test the hypothesis that all
> of the coefficients (except the firm dummies) are equal to zero I get
> a much more reasonable statistic. I get a value of 7236.41. Also,
> I've been able to replicate this number, while I am completely unable
> to replicate the number reported right after the iv regression
> command. I wonder if anyone could tell me the reason for this
> discrepancy.
I just checked this out, and I think it's a bug in -xtivreg-. If you
add the constant to the variables you check after the estimation, I'll
bet you get the same chi-sq statistic that -xtivreg- reports for the
overall significance of the regression. It's a bug because if this were
by design, the degrees of freedom in your example would be 12, not 11.
BTW, you're using fixed effects, so -xtivreg2- could be a useful
alternative.
Cheers,
Mark
> I have
> included the respective outputs below.
>
> Thank you in advance for your help,
> Sandra
>
>
>
> . xtivreg leveragemvf1 (turnover=turnoveri) drating mb eta depa irdd
> randda logsizebd ret pgy2d cpspread , fe i(permno);
>
> Fixed-effects (within) IV regression Number of obs =
> 45945
> Group variable: permno Number of groups =
> 8026
>
> R-sq: within = 0.1595 Obs per group: min =
> 1
> between = 0.1716 avg =
> 5.7
> overall = 0.1762 max =
> 19
>
> Wald chi2(11) =
> 434634.68
> corr(u_i, Xb) = -0.1251 Prob > chi2 =
> 0.0000
>
> --------------------------------------------------------------
> ----------
> ------
> leveragemvf1 | Coef. Std. Err. z P>|z| [95% Conf.
> Interval]
> -------------+------------------------------------------------
> ----------
> ------
> turnover | -2.320024 .2632847 -8.81 0.000 -2.836052
> -1.803995
> drating | 2.152626 .2693379 7.99 0.000 1.624733
> 2.680519
> mb | -2.011589 .1087904 -18.49 0.000 -2.224814
> -1.798363
> eta | -3.368408 .1067333 -31.56 0.000 -3.577602
> -3.159215
> depa | .6071639 .1246147 4.87 0.000 .3629235
> .8514043
> irdd | -.8173588 .4365988 -1.87 0.061 -1.673077
> .0383591
> randda | -.8690489 .149891 -5.80 0.000 -1.16283
> -.575268
> logsizebd | 10.25134 .2971439 34.50 0.000 9.668953
> 10.83374
> ret | -.3286456 .0161068 -20.40 0.000 -.3602144
> -.2970768
> pgy2d | -4.800349 .2518534 -19.06 0.000 -5.293973
> -4.306726
> cpspread | 7.88824 .2836412 27.81 0.000 7.332313
> 8.444166
> _cons | 32.36606 .278241 116.32 0.000 31.82072
> 32.91141
> -------------+------------------------------------------------
> ----------
> ------
> sigma_u | 20.522888
> sigma_e | 11.67359
> rho | .75554783 (fraction of variance due to u_i)
> --------------------------------------------------------------
> ----------
> ------
> F test that all u_i=0: F(8025,37908) = 10.23
> Prob > F =
> 0.0000
> --------------------------------------------------------------
> ----------
> ------
>
> . test turnover drating mb eta depa irdd randda logsizebd ret pgy2d
> cpspread;
>
> ( 1) turnover = 0
> ( 2) drating = 0
> ( 3) mb = 0
> ( 4) eta = 0
> ( 5) depa = 0
> ( 6) irdd = 0
> ( 7) randda = 0
> ( 8) logsizebd = 0
> ( 9) ret = 0
> (10) pgy2d = 0
> (11) cpspread = 0
>
> chi2( 11) = 7236.41
> Prob > chi2 = 0.0000
>
>
>
>
>
>
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
