Endogeneity test using ivreg2

Hamza Almustafa

Join Date: Feb 2016
Posts: 72

Endogeneity test using ivreg2

22 Nov 2016, 05:50

Dear helpful stata users,

I am trying to perform endogeneity test on my research variables using iverg2 command. The results showing high p-value, thus, can not reject the null hypothesis, which is unusual in similar research!!

I am not sure about the variables-lists specifications, since its showing below my exogenous variables as the included instruments rather than "excluded".

Thus, I would appreciate if you could help me make sure that the below command varilist specified correctly!

Thanks,
Hamza

Code:

. . noi: ivreg2 lnQ LNFAGE i.YEAR i.INDCLASS (l.lnQ LNBSIZE BOIND DUALITY FAM_OWN INST1 OWNTOTAL LNMARKT leverage = ld.lnQ ld.LNBSIZE ld.BOIND ld.DUALITY ld.FAM_OWN ld.INST1 ld.OWNTOTAL ld.LNMARKT ld.leverage ) if touseqJO&touseqUAE , endog( l.lnQ LNBSIZE BOIND DUALITY FAM_OWN INST1 OWNTOTAL LNMARKT leverage)
Warning - collinearities detected
Vars dropped:       2014.YEAR

IV (2SLS) estimation
--------------------

Estimates efficient for homoskedasticity only
Statistics consistent for homoskedasticity only

                                                      Number of obs =      763
                                                      F( 31,   731) =    17.32
                                                      Prob > F      =   0.0000
Total (centered) SS     =  90.85385763                Centered R2   =   0.6962
Total (uncentered) SS   =  92.87221415                Uncentered R2 =   0.7028
Residual SS             =  27.60204893                Root MSE      =    .1902

------------------------------------------------------------------------------
         lnQ |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         lnQ |
         L1. |   .9213701    .237351     3.88   0.000     .4561708    1.386569
             |
     LNBSIZE |   .1088334   .4609364     0.24   0.813    -.7945854    1.012252
       BOIND |  -.3843212   .4413782    -0.87   0.384    -1.249407    .4807642
     DUALITY |  -.0008265   .3205917    -0.00   0.998    -.6291746    .6275215
     FAM_OWN |  -.3130595   1.616354    -0.19   0.846    -3.481055    2.854936
       INST1 |   -.406788    1.28204    -0.32   0.751     -2.91954    2.105964
    OWNTOTAL |   .4410835   1.248548     0.35   0.724    -2.006025    2.888192
     LNMARKT |   .0215612   .0421565     0.51   0.609    -.0610641    .1041864
    leverage |  -.3419764   .8898935    -0.38   0.701    -2.086136    1.402183
      LNFAGE |  -.0146987   .0496335    -0.30   0.767    -.1119787    .0825812
             |
        YEAR |
       2008  |          0  (empty)
       2010  |  -.0741879   .0424993    -1.75   0.081    -.1574849    .0091091
       2011  |  -.0733074   .0308454    -2.38   0.017    -.1337633   -.0128514
       2012  |  -.0542644   .0233213    -2.33   0.020    -.0999733   -.0085555
       2013  |  -.0087028   .0221902    -0.39   0.695    -.0521948    .0347893
       2014  |          0  (omitted)
             |
    INDCLASS |
         33  |  -.0861202   .1270836    -0.68   0.498    -.3351996    .1629591
         44  |   .0073667   .1866398     0.04   0.969    -.3584406     .373174
         55  |  -.0056867    .246152    -0.02   0.982    -.4881357    .4767623
         66  |    .001329   .1046121     0.01   0.990     -.203707     .206365
         77  |   .0915878   .1087546     0.84   0.400    -.1215673    .3047429
         88  |  -.0123205   .0851857    -0.14   0.885    -.1792813    .1546404
         99  |    .016835   .2102778     0.08   0.936     -.395302     .428972
        110  |  -.0268373   .2157919    -0.12   0.901    -.4497817    .3961071
        111  |   .0031519   .1042046     0.03   0.976    -.2010853    .2073891
        222  |  -.0788749    .129087    -0.61   0.541    -.3318808     .174131
        333  |  -.0221477   .0994895    -0.22   0.824    -.2171435    .1728482
        444  |  -.2919729   .2596686    -1.12   0.261     -.800914    .2169681
        555  |  -.0535039   .0501137    -1.07   0.286     -.151725    .0447172
        666  |  -.0767877   .2159302    -0.36   0.722     -.500003    .3464276
        777  |  -.0163787   .0826444    -0.20   0.843    -.1783589    .1456014
        888  |   .0256633    .124964     0.21   0.837    -.2192616    .2705883
        999  |   .0518433    .250908     0.21   0.836    -.4399274     .543614
             |
       _cons |  -.1930042   1.479967    -0.13   0.896    -3.093686    2.707677
------------------------------------------------------------------------------
Underidentification test (Anderson canon. corr. LM statistic):           0.639
                                                   Chi-sq(1) P-val =    0.4241
------------------------------------------------------------------------------
Weak identification test (Cragg-Donald Wald F statistic):                0.068
Stock-Yogo weak ID test critical values:                       <not available>
------------------------------------------------------------------------------
Sargan statistic (overidentification test of all instruments):           0.000
                                                 (equation exactly identified)
-endog- option:
Endogeneity test of endogenous regressors:                               4.136
                                                   Chi-sq(9) P-val =    0.9023
Regressors tested:    L.lnQ LNBSIZE BOIND DUALITY FAM_OWN INST1 OWNTOTAL
                      LNMARKT leverage
------------------------------------------------------------------------------
Instrumented:         L.lnQ LNBSIZE BOIND DUALITY FAM_OWN INST1 OWNTOTAL
                      LNMARKT leverage
Included instruments: LNFAGE 2010.YEAR 2011.YEAR 2012.YEAR 2013.YEAR
                      33.INDCLASS 44.INDCLASS 55.INDCLASS 66.INDCLASS
                      77.INDCLASS 88.INDCLASS 99.INDCLASS 110.INDCLASS
                      111.INDCLASS 222.INDCLASS 333.INDCLASS 444.INDCLASS
                      555.INDCLASS 666.INDCLASS 777.INDCLASS 888.INDCLASS
                      999.INDCLASS
Excluded instruments: LD.lnQ LD.LNBSIZE LD.BOIND LD.DUALITY LD.FAM_OWN LD.INST1
                      LD.OWNTOTAL LD.LNMARKT LD.leverage
Dropped collinear:    2014.YEAR
------------------------------------------------------------------------------

Tags: None

Phil Bromiley

Join Date: Apr 2014

Posts: 4348
#2

22 Nov 2016, 09:52

I couldn't read your full code - it was cut off. You appear to be testing for many variables being endogenous. You might see what happens if you only test for endogeneity of the variables you think might be endogenous.
Comment
Hamza Almustafa

Join Date: Feb 2016

Posts: 72
#3

23 Nov 2016, 03:23

Hi Prof Phil, thanks for replying to my post.

The above code is the full one i have used to test endogeneity of variables! see it below please.

Code:

noi: ivreg2 lnQ LNFAGE i.YEAR i.INDCLASS (l.lnQ LNBSIZE BOIND DUALITY FAM_OWN INST1 OWNTOTAL LNMARKT leverage = ld.lnQ ld.LNBSIZE ld.BOIND ld.DUALITY ld.FAM_OWN ld.INST1 ld.OWNTOTAL ld.LNMARKT ld.leverage ) if touseqJO&touseqUAE , endog( l.lnQ LNBSIZE BOIND DUALITY FAM_OWN INST1 OWNTOTAL LNMARKT leverage)

However, i have tried the code with each single variable and still the test yielding high p-value!! a something seems to be strange in this field of research! I have no idea what would be the reason behind this.

Even i have tried it with alternate performance variables and still can not reject the null hypothesis !

What to do in this case, is it just to take the face value of these results and assume these variables are exogenous!??

Thanks once again,

Hamza

Last edited by Hamza Almustafa; 23 Nov 2016, 03:31.
Comment

Hamza Almustafa

Join Date: Feb 2016
Posts: 72

23 Nov 2016, 03:46

Just would like to add that I have followed (Baum, Schaffer, & Stillman, 2007), and perform the test using orthog option rather than endog, and again the C statistic results were negative!! see below please:

Code:

noi: ivreg2 lnQ LNFAGE i.YEAR i.INDCLASS l.lnQ LNBSIZE BOIND DUALITY FAM_OWN INST1 OWNTOTAL LNMARKT leverage (= ld.lnQ ld.LNBSIZE ld.BOIND ld.DUALITY ld.FAM_OWN ld.INST1 ld.OWNTOTAL ld.LNMARKT ld.leverage ) if touseqJO&touseqUAE , orthog( l.lnQ LNBSIZE BOIND DUALITY FAM_OWN INST1 OWNTOTAL LNMARKT leverage)
Warning - collinearities detected
Vars dropped:       2014.YEAR

OLS estimation
--------------

Estimates efficient for homoskedasticity only
Statistics consistent for homoskedasticity only

                                                      Number of obs =      763
                                                      F( 31,   731) =    87.05
                                                      Prob > F      =   0.0000
Total (centered) SS     =  90.85385763                Centered R2   =   0.7869
Total (uncentered) SS   =  92.87221415                Uncentered R2 =   0.7915
Residual SS             =  19.36443564                Root MSE      =    .1593

------------------------------------------------------------------------------
         lnQ |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      LNFAGE |   -.000802   .0091969    -0.09   0.931    -.0188276    .0172235
             |
        YEAR |
       2008  |          0  (empty)
       2010  |  -.0540785   .0186274    -2.90   0.004    -.0905875   -.0175694
       2011  |   -.064946   .0183886    -3.53   0.000     -.100987   -.0289051
       2012  |  -.0536106   .0182749    -2.93   0.003    -.0894288   -.0177924
       2013  |  -.0059328   .0182332    -0.33   0.745    -.0416693    .0298036
       2014  |          0  (omitted)
             |
    INDCLASS |
         33  |  -.0065348   .0237465    -0.28   0.783    -.0530772    .0400075
         44  |   .0173185   .0312877     0.55   0.580    -.0440042    .0786412
         55  |   .0598683   .0342113     1.75   0.080    -.0071845    .1269211
         66  |   .0579794   .0350609     1.65   0.098    -.0107387    .1266975
         77  |   .0012335   .0460458     0.03   0.979    -.0890147    .0914816
         88  |   .0175786   .0338148     0.52   0.603    -.0486972    .0838545
         99  |   .1453737   .0568289     2.56   0.011      .033991    .2567563
        110  |  -.0233579   .0406281    -0.57   0.565    -.1029876    .0562717
        111  |   .0060887   .0410687     0.15   0.882    -.0744046    .0865819
        222  |   .0140976   .0386923     0.36   0.716    -.0617379    .0899331
        333  |   .0099352   .0275385     0.36   0.718    -.0440392    .0639096
        444  |  -.1314222   .0747455    -1.76   0.079    -.2779206    .0150762
        555  |  -.0153823   .0247984    -0.62   0.535    -.0639862    .0332217
        666  |   .0792212   .0393177     2.01   0.044     .0021598    .1562825
        777  |   .0178356   .0268325     0.66   0.506    -.0347551    .0704264
        888  |   .0307313   .0437948     0.70   0.483    -.0551049    .1165676
        999  |   .0628773   .0748803     0.84   0.401    -.0838854      .20964
             |
         lnQ |
         L1. |   .8476117   .0196157    43.21   0.000     .8091657    .8860577
             |
     LNBSIZE |  -.0464634   .0249633    -1.86   0.063    -.0953907    .0024638
       BOIND |   .0361205   .0389038     0.93   0.353    -.0401296    .1123705
     DUALITY |  -.0092604   .0187834    -0.49   0.622    -.0460751    .0275544
     FAM_OWN |   .1270498   .0646209     1.97   0.049     .0003952    .2537044
       INST1 |   .0573098    .065276     0.88   0.380    -.0706288    .1852484
    OWNTOTAL |  -.0558769   .0648895    -0.86   0.389    -.1830579    .0713042
     LNMARKT |   .0175455   .0044694     3.93   0.000     .0087856    .0263054
    leverage |   .0274086   .0339572     0.81   0.420    -.0391462    .0939635
       _cons |  -.2272358   .0901794    -2.52   0.012    -.4039842   -.0504875
------------------------------------------------------------------------------
Sargan statistic (Lagrange multiplier test of excluded instruments):     4.136
                                                   Chi-sq(9) P-val =    0.9023
-orthog- option:
Sargan statistic (eqn. excluding suspect orthogonality conditions):      0.000
                                                   Chi-sq(0) P-val =         .
C statistic (exogeneity/orthogonality of suspect instruments):           4.136
                                                   Chi-sq(9) P-val =    0.9023
Instruments tested:   L.lnQ LNBSIZE BOIND DUALITY FAM_OWN INST1 OWNTOTAL
                      LNMARKT leverage
------------------------------------------------------------------------------
Included instruments: LNFAGE 2010.YEAR 2011.YEAR 2012.YEAR 2013.YEAR
                      33.INDCLASS 44.INDCLASS 55.INDCLASS 66.INDCLASS
                      77.INDCLASS 88.INDCLASS 99.INDCLASS 110.INDCLASS
                      111.INDCLASS 222.INDCLASS 333.INDCLASS 444.INDCLASS
                      555.INDCLASS 666.INDCLASS 777.INDCLASS 888.INDCLASS
                      999.INDCLASS L.lnQ LNBSIZE BOIND DUALITY FAM_OWN INST1
                      OWNTOTAL LNMARKT leverage
Excluded instruments: LD.lnQ LD.LNBSIZE LD.BOIND LD.DUALITY LD.FAM_OWN LD.INST1
                      LD.OWNTOTAL LD.LNMARKT LD.leverage
Dropped collinear:    2014.YEAR
------------------------------------------------------------------------------

Announcement

Endogeneity test using ivreg2

Comment

Comment

Comment