Dear Statalist community,
I run -ivreg2h- command from SSC with the fe option to implement the estimator proposed by Lewbel (2012, JBES) "Using heteroskedasticity to identify and estimate mismeasured and endogenous regressor models" for a dynamic panel of macroeconomic data with time and country fixed effects. I follow the guide by Baum and Lewbel (2019, The Stata Journal) "Advice on using heteroskedasticity-based identification" when implementing the command.
Baum and Lewbel (2019) suggest to run -ivhettest- after -ivreg2h- in order to test for heteroskedasticity in the residuals of the first stage regression, conditional on the constructed instruments. However, when using -ivhettest-, I get the following error "last estimates not found, r(301)".
I am using Stata/SE 16.0 on Windows 10.
I run the following code where d_L1Y is the endogenous variable to be instrumented. I also include a time fixed effect (i.quarter) which I then partial out. Moreover, z() reports the (assumed exogenous) variables used to build the instruments. These variables are used in the first stage regression. The outcome variable of the second stage regression is fa_Y. I use the xi: option in order to account for time dummies.
I get the following results:
Then, I run - ivhettest- and get the following error:
I use the following versions of -ivreg2h- and of - ivhettest- which should be also the last updates (at the best of my knowledge):
I hope the information I provided are useful to help me to figure out what I am doing wrong.
Thank you very much in advance.
Giacomo
I run -ivreg2h- command from SSC with the fe option to implement the estimator proposed by Lewbel (2012, JBES) "Using heteroskedasticity to identify and estimate mismeasured and endogenous regressor models" for a dynamic panel of macroeconomic data with time and country fixed effects. I follow the guide by Baum and Lewbel (2019, The Stata Journal) "Advice on using heteroskedasticity-based identification" when implementing the command.
Baum and Lewbel (2019) suggest to run -ivhettest- after -ivreg2h- in order to test for heteroskedasticity in the residuals of the first stage regression, conditional on the constructed instruments. However, when using -ivhettest-, I get the following error "last estimates not found, r(301)".
I am using Stata/SE 16.0 on Windows 10.
I run the following code where d_L1Y is the endogenous variable to be instrumented. I also include a time fixed effect (i.quarter) which I then partial out. Moreover, z() reports the (assumed exogenous) variables used to build the instruments. These variables are used in the first stage regression. The outcome variable of the second stage regression is fa_Y. I use the xi: option in order to account for time dummies.
Code:
xi: ivreg2h fa_Y L1d_L1Y L2d_L1Y L1fa_Y L2fa_Y dlrealgdp dstock_prices dgov_bond_yields ///
i.quarter (d_L1Y =) if noIRE==1, fe robust bw(2) gmm2s ///
z(L1d_L1Y L2d_L1Y L1fa_Y L2fa_Y dlrealgdp dstock_prices dgov_bond_yields) ///
partial(i.quarter)
Code:
IV with Generated Instruments only
Fixed Effects by(CountryCode), 16 groups
Instruments created from Z:
L1d_L1Y L2d_L1Y L1fa_Y L2fa_Y dlrealgdp dstock_prices dgov_bond_yields
Warning - collinearities detected
Vars dropped: _Iquarter_41 _Iquarter_235
Warning - collinearities detected
Vars dropped: _Iquarter_41 _Iquarter_235
2-Step GMM estimation
---------------------
Estimates efficient for arbitrary heteroskedasticity and autocorrelation
Statistics robust to heteroskedasticity and autocorrelation
kernel=Bartlett; bandwidth=2
time variable (t): quarter
group variable (i): CountryCode
Number of obs = 1389
F( 8, 1172) = 18.03
Prob > F = 0.0000
Total (centered) SS = 50718.29679 Centered R2 = 0.5192
Total (uncentered) SS = 50718.29679 Uncentered R2 = 0.5192
Residual SS = 24387.76391 Root MSE = 4.215
----------------------------------------------------------------------------------
| Robust
fa_Y | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-----------------+----------------------------------------------------------------
d_L1Y | 1.098956 .1145213 9.60 0.000 .8744988 1.323414
L1d_L1Y | -.1575488 .0408916 -3.85 0.000 -.2376948 -.0774028
L2d_L1Y | -.1898572 .055578 -3.42 0.001 -.2987881 -.0809263
L1fa_Y | .0513925 .0330912 1.55 0.120 -.013465 .11625
L2fa_Y | .1006611 .0362383 2.78 0.005 .0296354 .1716868
dlrealgdp | -.3768432 .2409836 -1.56 0.118 -.8491624 .095476
dstock_prices | .0123677 .0375827 0.33 0.742 -.061293 .0860284
dgov_bond_yields | -.5828282 .4648038 -1.25 0.210 -1.493827 .3281705
----------------------------------------------------------------------------------
Underidentification test (Kleibergen-Paap rk LM statistic): 5.054
Chi-sq(7) P-val = 0.6534
------------------------------------------------------------------------------
Weak identification test (Cragg-Donald Wald F statistic): 11.374
(Kleibergen-Paap rk Wald F statistic): 1.657
Stock-Yogo weak ID test critical values: 5% maximal IV relative bias 19.86
10% maximal IV relative bias 11.29
20% maximal IV relative bias 6.73
30% maximal IV relative bias 5.07
10% maximal IV size 31.50
15% maximal IV size 17.38
20% maximal IV size 12.48
25% maximal IV size 9.93
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): 2.331
Chi-sq(6) P-val = 0.8869
------------------------------------------------------------------------------
Instrumented: d_L1Y
Included instruments: L1d_L1Y L2d_L1Y L1fa_Y L2fa_Y dlrealgdp dstock_prices
dgov_bond_yields
Excluded instruments: d_L1Y_L1d_L1Y_g d_L1Y_L2d_L1Y_g d_L1Y_L1fa_Y_g
d_L1Y_L2fa_Y_g d_L1Y_dlrealgdp_g d_L1Y_dstock_prices_g
d_L1Y_dgov_bond_yields_g
Partialled-out: _Iquarter_42 _Iquarter_43 -- _Iquarter_234
nb: total SS, model F and R2s are after partialling-out;
any small-sample adjustments include partialled-out
variables in regressor count K
Dropped collinear: _Iquarter_41 -------------- _Iquarter_235
------------------------------------------------------------------------------
Warning: variables have been centered
-----------------------------
Variable | GenInst
-------------+---------------
d_L1Y | 1.099
| .115
L1d_L1Y | -.1575
| .0409
L2d_L1Y | -.1899
| .0556
L1fa_Y | .05139
| .0331
L2fa_Y | .1007
| .0362
dlrealgdp | -.3768
| .241
dstock_pri~s | .01237
| .0376
dgov_bond_~s | -.5828
| .465
-------------+---------------
N | 1389
rmse | 4.21
j | 2.33
jdf | 6
jp | .887
-----------------------------
legend: b/se
Warning - collinearities detected
Vars dropped: _Iquarter_41 _Iquarter_235
Code:
. ivhettest last estimates not found r(301);
Code:
. which ivreg2h C:\Users\Giacomo.Rella\ado\plus\i\ivreg2h.ado *! ivreg2h 1.1.03 07feb2019 cfb/mes *! cloned from *! xtivreg2 1.0.13 28Aug2011 *! author mes . which ivhettest C:\Users\Giacomo.Rella\ado\plus\i\ivhettest.ado *! ivhettest 1.1.9 15Aug2013 *! author mes
Thank you very much in advance.
Giacomo
Comment