First-stage results after aidsills?

[Hanna Lindstrom]

Dear all,
I am running the command aidsills created by Lecocq and Robin (2015). I am using instrumental variables to handle endogenous prices and expenditures. This produces output in the form (and order) of (tables are cut off for brevity):

INSTRUMENTAL REGRESSION(S)
note: seas4 omitted because of collinearity

      Source |       SS           df       MS      Number of obs   =     1,743
-------------+----------------------------------   F(16, 1726)     =   2707.05
       Model |  13.2394544        16  .827465903   Prob > F        =    0.0000
    Residual |  .527587086     1,726   .00030567   R-squared       =    0.9617
-------------+----------------------------------   Adj R-squared   =    0.9613
       Total |  13.7670415     1,742  .007903009   Root MSE        =    .01748

----------------------------------------------------------------------------------
       lnpriceCB |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------+----------------------------------------------------------------
    lniv_priceCB |   .8386814   .0133072    63.02   0.000     .8125997    .8647631
   lniv_priceCPL |   .0099427   .0134555     0.74   0.460    -.0164297     .036315
    lniv_priceOB |   .0022226    .008748     0.25   0.799    -.0149232    .0193684

and after each price instrument regression (and expenditure instrument regression) comes:

AIDS - PROPER ESTIMATION WITH FIXED ALPHA_0 = 0
HOMOGENEITY AND SYMMETRY CONSTRAINED ESTIMATES

------------------------------------------------------------------------------
Equation          Obs  Parms        RMSE    "R-sq"    F( 13,  1729)   Prob > F
------------------------------------------------------------------------------
CBshare          1743     13    .0768973    0.5842          202.56      0.0000
CPLshare         1743     13     .046586    0.2778           55.45      0.0000
OBshare          1743     13    .0275746    0.7750          496.45      0.0000
OPLshare         1743     13    .0185236    0.7103          353.46      0.0000
------------------------------------------------------------------------------

-----------------------------------------------------------------------------------
                  |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
CBshare           |
  gamma_lnpriceCB |  -.8374146   .0515355   -16.25   0.000    -.9384224   -.7364068
 gamma_lnpriceCPL |   .4536907   .0724279     6.26   0.000     .3117346    .5956468
  gamma_lnpriceOB |   .2294044    .049631     4.62   0.000     .1321295    .3266794

Now, I have two questions related to this:
1) If I want to report the estimates from the Instrumental regressions (i.e., first stage), is there a more user-friendly way of doing that than copying output tables to Excel and working with them i Excel? As I understand it, the only estimates stored are those of the actual demand system, i.e., the last estimation. Basically: Do Stata store the results from the instrumental regressions so that I can retrieve and use esttab to produce Excel/Latex tables?

2) According to the Stata article by Lecocq and Robin (2015), one can test the exogeneity of the variables instrumented for by testing the significance of the coefficient of the residuals of the instrumental regression: Example from their article:

. test rho_vexpfd         ( 1)  [w1]rho_vexpfd = 0        ( 2)  [w2]rho_vexpfd = 0        ( 3)  [w3]rho_vexpfd = 0        ( 4)  [w4]rho_vexpfd = 0                 Constraint 2 dropped                     chi2(  3) =   11.05                Prob > chi2 =    0.0114

However, are there any other ways of testing the relevance/validity of instruments, e.g., tests of overidentification, Hansen test, J-test, etc after running aidsills?

References:
Sebastien Lecocq & Jean-Marc Robin, 2015. "Estimating almost-ideal demand systems with endogenous regressors," Stata Journal, StataCorp LP, vol. 15(2), pages 554-573, June.

Best regards,
Hanna Lindström

[Henni Burk]

Dear Hanna,
I would like to ask if you can provide the full command you used? I'm working on a similar model.
Due to the COVID-19 situation I cannot work on the original data set, but a data structure set which contains fake values and a lot of missing values due to data protection reasons, and I get a lot of error messages. I just would like to make sure that I did not make a mistake in the command itself. I think the errors are mainly due to the missing values though.
I'd appreciate your help.

Best regards,
Henni Burk

[Clara Daru]

Did anyone ever find a way to do this? I'm looking for the same