How to save coefficient of fixed effects?

[Alberto Alvarez]

Dear Stata Users,

I have an unbalanced panel data. What I need is to save the coefficient of i.gvkey (for each observation - firm(gvkey)-year) from the following regression:

 
 xtset gvkey1 fyear reg bm_w cum_ret_w b1cum_ret_w b2cum_ret_w b3cum_ret_w b4cum_ret_w b5cum_ret_w b6cum_ret_w i.fyear, r

Can you please specify the way I can proceed with? I found in the previous postings of the forum the following method:

xtset gvkey1
reg bm_w cum_ret_w b1cum_ret_w b2cum_ret_w b3cum_ret_w b4cum_ret_w b5cum_ret_w b6cum_ret_w i.fyear, r 
predict gvkey1_effect

However, I am conserned if stata treats the data as a panel while estimating the coefficient.

Best regards,
Alberto

[Carlo Lazzaro]

Alberto:
you did not receive any reply because your codes are inconsistent.
I fail to get why you -xtset- your data first and then go -regress-.

[Alberto Alvarez]

Sorry, it just should be like this:

xtset gvkey1 fyear

reg bm_w cum_ret_w b1cum_ret_w b2cum_ret_w b3cum_ret_w b4cum_ret_w b5cum_ret_w b6cum_ret_w i.fyear, r

Fist, I define Panel data structure, and then run pooled ols. (Hope, I understood your question correctly)

[Carlo Lazzaro]

Alberto:
I still fail to get why you -xtset- your data first and then go -regress-.
-regress- does not need -xtset- to work properly, as you can see from the following toy-example:

. use "http://www.stata-press.com/data/r14/nlswork.dta", clear
(National Longitudinal Survey.  Young Women 14-26 years of age in 1968)


. xtset idcode year
       panel variable:  idcode (unbalanced)
        time variable:  year, 68 to 88, but with gaps
                delta:  1 unit

. reg ln_wage tenure

      Source |       SS           df       MS      Number of obs   =    28,101
-------------+----------------------------------   F(1, 28099)     =   4473.23
       Model |  880.984271         1  880.984271   Prob > F        =    0.0000
    Residual |  5533.98035    28,099  .196945811   R-squared       =    0.1373
-------------+----------------------------------   Adj R-squared   =    0.1373
       Total |  6414.96462    28,100  .228290556   Root MSE        =    .44379

------------------------------------------------------------------------------
     ln_wage |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      tenure |   .0471994   .0007057    66.88   0.000     .0458162    .0485826
       _cons |   1.529661   .0034451   444.02   0.000     1.522909    1.536414
------------------------------------------------------------------------------

. xtset, clear

. reg ln_wage tenure

      Source |       SS           df       MS      Number of obs   =    28,101
-------------+----------------------------------   F(1, 28099)     =   4473.23
       Model |  880.984271         1  880.984271   Prob > F        =    0.0000
    Residual |  5533.98035    28,099  .196945811   R-squared       =    0.1373
-------------+----------------------------------   Adj R-squared   =    0.1373
       Total |  6414.96462    28,100  .228290556   Root MSE        =    .44379

------------------------------------------------------------------------------
     ln_wage |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      tenure |   .0471994   .0007057    66.88   0.000     .0458162    .0485826
       _cons |   1.529661   .0034451   444.02   0.000     1.522909    1.536414
------------------------------------------------------------------------------

[Alberto Alvarez]

Thank you for the response, I see.. I just tried to make "Stata" understand that I use panel data.. But skipping "xtset" can you please help me with my question regarding saving the coefficient of "gvkey1" which is a firmID identifier?

[Carlo Lazzaro]

Alberto:
see Example #2 under -matrix mkmat- entry, Stata .pdf manual.
As an aside, should you have any sound methodological reason to run an OLS instead of an -xt- prefixed command with panel data, I would recommend you to cluster your standard errors on -panelid-. Otherwise, -regress-, ignoring that you're dealing with repeated measures on the same sample, will consider observations as independent (which is not the case with panel data).