Good evening,
I am using StataIC16, Panel Data and I am running a FE-IV Regression. I adapted the model and from my perspective it seems reasonable but I am still not sure whether it is trust worthy.
My regressors are relevant and exogenous, R^2 is quite high and the regressor of interest is significant at the ten percent level. Furthermore, I am correcting for heteroskedasticity and serial correlation. The result varies strongly from the FE regression but this doesn't seem unreasonable, since the average spending variable is endogenous and could be negatively correlated with the error term.
Am I missing something here? Are their any other possibilities to check the validity of the model?
Thank you very much for your help!
[xtivreg2 math4 (aexpp = lfound l96 l97 l98) ia1 ia2 ia3 y96 y97 y98 $control, fe cluster(distid)]
Warning - singleton groups detected. 7 observation(s) not used.
FIXED EFFECTS ESTIMATION
------------------------
Number of groups = 543 Obs per group: min = 2
avg = 4.0
max = 4
IV (2SLS) estimation
--------------------
Estimates efficient for homoskedasticity only
Statistics robust to heteroskedasticity and clustering on distid
Number of clusters (distid) = 543 Number of obs = 2152
F( 11, 542) = 86.58
Prob > F = 0.0000
Total (centered) SS = 201356.9083 Centered R2 = 0.3610
Total (uncentered) SS = 201356.9083 Uncentered R2 = 0.3610
Residual SS = 128668.9395 Root MSE = 8.942
------------------------------------------------------------------------------
| Robust
math4 | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
aexpp | 79.07982 44.23961 1.79 0.074 -7.628217 165.7878
ia1 | 11.57426 5.810394 1.99 0.046 .1860937 22.96242
ia2 | 18.51577 7.438171 2.49 0.013 3.937223 33.09432
ia3 | 11.1585 9.108134 1.23 0.221 -6.693115 29.01011
y96 | -103.4514 51.78953 -2.00 0.046 -204.957 -1.945742
y97 | -167.2465 67.15044 -2.49 0.013 -298.8589 -35.63407
y98 | -90.99324 82.25853 -1.11 0.269 -252.217 70.23051
lunch | .4020226 .4997337 0.80 0.421 -.5774375 1.381483
lunchsq | -.001686 .0053255 -0.32 0.752 -.0121238 .0087519
lenrol | 114.7754 69.24006 1.66 0.097 -20.93266 250.4834
lenrolsq | -6.272133 4.191827 -1.50 0.135 -14.48796 1.943698
------------------------------------------------------------------------------
Underidentification test (Kleibergen-Paap rk LM statistic): 27.608
Chi-sq(4) P-val = 0.0000
------------------------------------------------------------------------------
Weak identification test (Cragg-Donald Wald F statistic): 55.024
(Kleibergen-Paap rk Wald F statistic): 13.185
Stock-Yogo weak ID test critical values: 5% maximal IV relative bias 16.85
10% maximal IV relative bias 10.27
20% maximal IV relative bias 6.71
30% maximal IV relative bias 5.34
10% maximal IV size 24.58
15% maximal IV size 13.96
20% maximal IV size 10.26
25% maximal IV size 8.31
Source: Stock-Yogo (2005). Reproduced by permission.
NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors.
------------------------------------------------------------------------------
Hansen J statistic (overidentification test of all instruments): 3.946
Chi-sq(3) P-val = 0.2673
------------------------------------------------------------------------------
Instrumented: aexpp
Included instruments: ia1 ia2 ia3 y96 y97 y98 lunch lunchsq lenrol lenrolsq
Excluded instruments: lfound l96 l97 l98
------------------------------------------------------------------------------
I am using StataIC16, Panel Data and I am running a FE-IV Regression. I adapted the model and from my perspective it seems reasonable but I am still not sure whether it is trust worthy.
My regressors are relevant and exogenous, R^2 is quite high and the regressor of interest is significant at the ten percent level. Furthermore, I am correcting for heteroskedasticity and serial correlation. The result varies strongly from the FE regression but this doesn't seem unreasonable, since the average spending variable is endogenous and could be negatively correlated with the error term.
Am I missing something here? Are their any other possibilities to check the validity of the model?
Thank you very much for your help!
[xtivreg2 math4 (aexpp = lfound l96 l97 l98) ia1 ia2 ia3 y96 y97 y98 $control, fe cluster(distid)]
Warning - singleton groups detected. 7 observation(s) not used.
FIXED EFFECTS ESTIMATION
------------------------
Number of groups = 543 Obs per group: min = 2
avg = 4.0
max = 4
IV (2SLS) estimation
--------------------
Estimates efficient for homoskedasticity only
Statistics robust to heteroskedasticity and clustering on distid
Number of clusters (distid) = 543 Number of obs = 2152
F( 11, 542) = 86.58
Prob > F = 0.0000
Total (centered) SS = 201356.9083 Centered R2 = 0.3610
Total (uncentered) SS = 201356.9083 Uncentered R2 = 0.3610
Residual SS = 128668.9395 Root MSE = 8.942
------------------------------------------------------------------------------
| Robust
math4 | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
aexpp | 79.07982 44.23961 1.79 0.074 -7.628217 165.7878
ia1 | 11.57426 5.810394 1.99 0.046 .1860937 22.96242
ia2 | 18.51577 7.438171 2.49 0.013 3.937223 33.09432
ia3 | 11.1585 9.108134 1.23 0.221 -6.693115 29.01011
y96 | -103.4514 51.78953 -2.00 0.046 -204.957 -1.945742
y97 | -167.2465 67.15044 -2.49 0.013 -298.8589 -35.63407
y98 | -90.99324 82.25853 -1.11 0.269 -252.217 70.23051
lunch | .4020226 .4997337 0.80 0.421 -.5774375 1.381483
lunchsq | -.001686 .0053255 -0.32 0.752 -.0121238 .0087519
lenrol | 114.7754 69.24006 1.66 0.097 -20.93266 250.4834
lenrolsq | -6.272133 4.191827 -1.50 0.135 -14.48796 1.943698
------------------------------------------------------------------------------
Underidentification test (Kleibergen-Paap rk LM statistic): 27.608
Chi-sq(4) P-val = 0.0000
------------------------------------------------------------------------------
Weak identification test (Cragg-Donald Wald F statistic): 55.024
(Kleibergen-Paap rk Wald F statistic): 13.185
Stock-Yogo weak ID test critical values: 5% maximal IV relative bias 16.85
10% maximal IV relative bias 10.27
20% maximal IV relative bias 6.71
30% maximal IV relative bias 5.34
10% maximal IV size 24.58
15% maximal IV size 13.96
20% maximal IV size 10.26
25% maximal IV size 8.31
Source: Stock-Yogo (2005). Reproduced by permission.
NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors.
------------------------------------------------------------------------------
Hansen J statistic (overidentification test of all instruments): 3.946
Chi-sq(3) P-val = 0.2673
------------------------------------------------------------------------------
Instrumented: aexpp
Included instruments: ia1 ia2 ia3 y96 y97 y98 lunch lunchsq lenrol lenrolsq
Excluded instruments: lfound l96 l97 l98
------------------------------------------------------------------------------
