What estimation methods can endogenicity, heteroskedasticity, autocorrelation and fixed effects be solved without using GMM?

[Koffi Yves YA]

Hello netizens of Stata forum.
I have a panel of 13 countries and 8 years. I tried with the GMM but the results are not satisfactory because the number of instruments is always higher than the number of groups even if I drastically reduce the number of regressors.
What are the appropriate estimation methods to simultaneously control endogeneity, heteroscedasticity, autocorrelation and fixed effects in Stata?
Endogeneity, heteroscedasticity, autocorrelation and fixed effects are the problems in my data.

Sebastian Kripfganz, Phil Bromiley, Joao Santos Silva, Carlo Lazzaro , Clyde Schechter , Nick Cain, ...

Thank you

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a helpful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

The most likely candidates are xtivreg, xtivreg2, and the extended regression procedures (with dummies for the panel effects).

[Koffi Yves YA]

Hello,
I performed these two regressions that follow with xtivreg and xtivreg2. I notice that the first command that is available in the Stata menu does not take into account instrumental variables even without checking the box that asks you to ignore them.
Question 1: Are instrumental variables taken into account with the xtivreg command?
Question 2: What are the validity tests I can still perform on these two estimated models?
Question 3: Are the very significant nature of the constant term and its volume a handicap for xtivreg models?
Question 4: In view of these results, are there other suitable methods that I could implement?
Thanks

xtivreg IDHIx logTFM TIBCP IDEx CIFSPx logAPD CPES VOIX SP (logPIBH= logCONSO), fe vce(robust) small
Fixed-effects (within) IV regression            Number of obs     =        104
Group variable: COUNTRY1                        Number of groups  =         13
R-sq:                                           Obs per group:
     within  = 0.7010                                         min =          8
     between = 0.5640                                         avg =        8.0
     overall = 0.5729                                         max =          8
                                                F(  22,     82)   =    311.09
corr(u_i, Xb)  = -0.6243                        Prob > F          =     0.0000
                              (Std. Err. adjusted for 13 clusters in COUNTRY1)
------------------------------------------------------------------------------
             |               Robust
       IDHIx |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
     logPIBH |   7.888746   3.513366     2.25   0.027     .8995425    14.87795
      logTFM |   .6971156   .4159521     1.68   0.098    -.1303457    1.524577
       TIBCP |    .094436   .0245572     3.85   0.000      .045584     .143288
        IDEx |  -.0625537   .0086585    -7.22   0.000    -.0797782   -.0453291
      CIFSPx |   .3263396   .0844697     3.86   0.000     .1583024    .4943768
      logAPD |   1.128704   .8568108     1.32   0.191    -.5757658    2.833173
        CPES |  -3.607277   .8204238    -4.40   0.000    -5.239361   -1.975193
        VOIX |   4.313346   1.975139     2.18   0.032     .3841652    8.242526
          SP |  -1.192385   1.052773    -1.13   0.261    -3.286687    .9019166
       _cons |  -66.38546   20.41832    -3.25   0.002     -107.004   -25.76691
-------------+----------------------------------------------------------------
     sigma_u |  4.1087018
     sigma_e |  1.7627786
         rho |  .84454368   (fraction of variance due to u_i)
------------------------------------------------------------------------------
Instrumented:   logPIBH

Instruments:    logTFM TIBCP IDEx CIFSPx logAPD CPES VOIX SP logCONSO




. xtivreg2 IDHIx logTFM TIBCP ATMx IDEx CIFSPx logAPD CPES VOIX SP (logPIBH= logCONSO), fe small
FIXED EFFECTS ESTIMATION
------------------------
Number of groups =        13                    Obs per group: min =         8
                                                               avg =       8.0
                                                               max =         8
IV (2SLS) estimation
--------------------
Estimates efficient for homoskedasticity only
Statistics consistent for homoskedasticity only
                                                      Number of obs =      104
                                                      F( 10,    81) =    28.41
                                                      Prob > F      =   0.0000
Total (centered) SS     =  852.2103746                Centered R2   =   0.7747
Total (uncentered) SS   =  852.2103746                Uncentered R2 =   0.7747
Residual SS             =  192.0250974                Root MSE      =     1.54
------------------------------------------------------------------------------
       IDHIx |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
     logPIBH |   4.182864   2.122697     1.97   0.052    -.0406373    8.406366
      logTFM |   .3157388   .4966917     0.64   0.527    -.6725219       1.304
       TIBCP |   .0545084   .0225163     2.42   0.018      .009708    .0993088
        ATMx |   .0677325   .0159252     4.25   0.000     .0360462    .0994187
        IDEx |  -.0474127   .0151067    -3.14   0.002    -.0774704   -.0173551
      CIFSPx |   .1853724   .0711998     2.60   0.011     .0437071    .3270377
      logAPD |   .7865625   .6499773     1.21   0.230    -.5066884    2.079813
        CPES |  -1.728317   1.228947    -1.41   0.163    -4.173536    .7169013
        VOIX |   3.000796   1.212118     2.48   0.015     .5890613     5.41253
          SP |  -.4210053   .7849878    -0.54   0.593    -1.982885    1.140874
------------------------------------------------------------------------------
Underidentification test (Anderson canon. corr. LM statistic):          59.488
                                                   Chi-sq(1) P-val =    0.0000
------------------------------------------------------------------------------
Weak identification test (Cragg-Donald Wald F statistic):              152.914
Stock-Yogo weak ID test critical values: 10% maximal IV size             16.38
                                         15% maximal IV size              8.96
                                         20% maximal IV size              6.66
                                         25% maximal IV size              5.53
Source: Stock-Yogo (2005).  Reproduced by permission.
------------------------------------------------------------------------------
Sargan statistic (overidentification test of all instruments):           0.000
                                                 (equation exactly identified)
------------------------------------------------------------------------------
Instrumented:         logPIBH
Included instruments: logTFM TIBCP ATMx IDEx CIFSPx logAPD CPES VOIX SP
Excluded instruments: logCONSO
------------------------------------------------------------------------------