Hello guys, I want to ask something about one step difference gmm with xtabond2 command in state, where my N=34 and T=7
I'm doing one step difference gmm with xtabond2 command, with and without time dummies
without time dummies I found my ar, sargan, and hansen statistic test is invalid but my model and all the independent variables are statistically significant, Here is my code and the result without time dummies
.
xtabond2 lnIPM l1.lnIPM lnPLN lnSL lnJALAN lnSLS lnSAML, gmm(l1.lnIPM, collapse) iv(lnPLN lnSL lnJALAN lnSL
> S lnSAML) noleveleq nodiffsargan robust
Favoring speed over space. To switch, type or click on mata: mata set matafavor space, perm.
Warning: Two-step estimated covariance matrix of moments is singular.
Using a generalized inverse to calculate robust weighting matrix for Hansen test.
Dynamic panel-data estimation, one-step difference GMM
------------------------------------------------------------------------------
Group variable: provinsi Number of obs = 165
Time variable : tahun Number of groups = 33
Number of instruments = 10 Obs per group: min = 5
Wald chi2(0) = . avg = 5.00
Prob > chi2 = . max = 5
------------------------------------------------------------------------------
| Robust
lnIPM | Coefficient std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
lnIPM |
L1. | .5464483 .109845 4.97 0.000 .3311559 .7617406
|
lnPLN | .0556826 .0350971 1.59 0.113 -.0131064 .1244716
lnSL | -.0169182 .0074623 -2.27 0.023 -.031544 -.0022924
lnJALAN | .002181 .0138822 0.16 0.875 -.0250276 .0293896
lnSLS | -.0043936 .0032954 -1.33 0.182 -.0108524 .0020652
lnSAML | .0299093 .0069274 4.32 0.000 .0163317 .0434868
------------------------------------------------------------------------------
Instruments for first differences equation
Standard
D.(lnPLN lnSL lnJALAN lnSLS lnSAML)
GMM-type (missing=0, separate instruments for each period unless collapsed)
L(1/6).L.lnIPM collapsed
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z = -2.43 Pr > z = 0.015
Arellano-Bond test for AR(2) in first differences: z = -3.21 Pr > z = 0.001
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(4) = 75.75 Prob > chi2 = 0.000
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(4) = 24.74 Prob > chi2 = 0.000
(Robust, but weakened by many instruments.)
with time dummies, I found my ar, sargan, and hansen statistic test is valid, but my coefficient for the independent variable drastically changed and become insignificant, here is my code and result with time dummies
xtabond2 lnIPM l1.lnIPM lnPLN lnSL lnJALAN lnSLS lnSAML tahun_*, gmm(l1.lnIPM, collapse) iv(lnPLN lnSL lnJA
> LAN lnSLS lnSAML tahun_*) noleveleq nodiffsargan robust
Favoring speed over space. To switch, type or click on mata: mata set matafavor space, perm.
tahun_1 dropped due to collinearity
tahun_3 dropped due to collinearity
Warning: Two-step estimated covariance matrix of moments is singular.
Using a generalized inverse to calculate robust weighting matrix for Hansen test.
Dynamic panel-data estimation, one-step difference GMM
------------------------------------------------------------------------------
Group variable: provinsi Number of obs = 165
Time variable : tahun Number of groups = 33
Number of instruments = 15 Obs per group: min = 5
Wald chi2(0) = . avg = 5.00
Prob > chi2 = . max = 5
------------------------------------------------------------------------------
| Robust
lnIPM | Coefficient std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
lnIPM |
L1. | .9731091 .2590362 3.76 0.000 .4654076 1.480811
|
lnPLN | -.0121004 .0340991 -0.35 0.723 -.0789334 .0547326
lnSL | -.0082242 .0033951 -2.42 0.015 -.0148786 -.0015699
lnJALAN | -.0112446 .0083678 -1.34 0.179 -.0276453 .005156
lnSLS | -.002733 .0030194 -0.91 0.365 -.0086509 .0031848
lnSAML | .012342 .0128841 0.96 0.338 -.0129105 .0375945
tahun_2 | -.0008708 .002072 -0.42 0.674 -.0049318 .0031903
tahun_4 | -.0003193 .0034074 -0.09 0.925 -.0069977 .006359
tahun_5 | -.0114039 .0063957 -1.78 0.075 -.0239393 .0011315
tahun_6 | -.0048179 .0047947 -1.00 0.315 -.0142154 .0045795
tahun_7 | .0002146 .0056747 0.04 0.970 -.0109075 .0113368
------------------------------------------------------------------------------
Instruments for first differences equation
Standard
D.(lnPLN lnSL lnJALAN lnSLS lnSAML tahun_1 tahun_2 tahun_3 tahun_4 tahun_5
tahun_6 tahun_7)
GMM-type (missing=0, separate instruments for each period unless collapsed)
L(1/6).L.lnIPM collapsed
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z = -1.36 Pr > z = 0.173
Arellano-Bond test for AR(2) in first differences: z = -1.65 Pr > z = 0.099
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(4) = 2.04 Prob > chi2 = 0.728
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(4) = 6.43 Prob > chi2 = 0.169
(Robust, but weakened by many instruments.)
***Note : tahun_ is my time dummies***
.
here is what I want to ask:
1. Is it normal when my coefficient and significant level changed due to time dummies introduction to the model?
2. if I stick with my first code (without time dummies) what kind of special treatment i need to do to deal with invalid ar, sargan, and hansen statistic test?
Thank You!
I'm doing one step difference gmm with xtabond2 command, with and without time dummies
without time dummies I found my ar, sargan, and hansen statistic test is invalid but my model and all the independent variables are statistically significant, Here is my code and the result without time dummies
Code:
xtabond2 lnIPM l1.lnIPM lnPLN lnSL lnJALAN lnSLS lnSAML, gmm(l1.lnIPM, collapse) iv(lnPLN lnSL lnJALAN lnSLS lnSAML) noleveleq nodiffsargan robust
xtabond2 lnIPM l1.lnIPM lnPLN lnSL lnJALAN lnSLS lnSAML, gmm(l1.lnIPM, collapse) iv(lnPLN lnSL lnJALAN lnSL
> S lnSAML) noleveleq nodiffsargan robust
Favoring speed over space. To switch, type or click on mata: mata set matafavor space, perm.
Warning: Two-step estimated covariance matrix of moments is singular.
Using a generalized inverse to calculate robust weighting matrix for Hansen test.
Dynamic panel-data estimation, one-step difference GMM
------------------------------------------------------------------------------
Group variable: provinsi Number of obs = 165
Time variable : tahun Number of groups = 33
Number of instruments = 10 Obs per group: min = 5
Wald chi2(0) = . avg = 5.00
Prob > chi2 = . max = 5
------------------------------------------------------------------------------
| Robust
lnIPM | Coefficient std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
lnIPM |
L1. | .5464483 .109845 4.97 0.000 .3311559 .7617406
|
lnPLN | .0556826 .0350971 1.59 0.113 -.0131064 .1244716
lnSL | -.0169182 .0074623 -2.27 0.023 -.031544 -.0022924
lnJALAN | .002181 .0138822 0.16 0.875 -.0250276 .0293896
lnSLS | -.0043936 .0032954 -1.33 0.182 -.0108524 .0020652
lnSAML | .0299093 .0069274 4.32 0.000 .0163317 .0434868
------------------------------------------------------------------------------
Instruments for first differences equation
Standard
D.(lnPLN lnSL lnJALAN lnSLS lnSAML)
GMM-type (missing=0, separate instruments for each period unless collapsed)
L(1/6).L.lnIPM collapsed
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z = -2.43 Pr > z = 0.015
Arellano-Bond test for AR(2) in first differences: z = -3.21 Pr > z = 0.001
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(4) = 75.75 Prob > chi2 = 0.000
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(4) = 24.74 Prob > chi2 = 0.000
(Robust, but weakened by many instruments.)
with time dummies, I found my ar, sargan, and hansen statistic test is valid, but my coefficient for the independent variable drastically changed and become insignificant, here is my code and result with time dummies
Code:
xtabond2 lnIPM l1.lnIPM lnPLN lnSL lnJALAN lnSLS lnSAML tahun_*, gmm(l1.lnIPM, collapse) iv(lnPLN lnSL lnJALAN lnSLS lnSAML tahun_*) noleveleq nodiffsargan robust
> LAN lnSLS lnSAML tahun_*) noleveleq nodiffsargan robust
Favoring speed over space. To switch, type or click on mata: mata set matafavor space, perm.
tahun_1 dropped due to collinearity
tahun_3 dropped due to collinearity
Warning: Two-step estimated covariance matrix of moments is singular.
Using a generalized inverse to calculate robust weighting matrix for Hansen test.
Dynamic panel-data estimation, one-step difference GMM
------------------------------------------------------------------------------
Group variable: provinsi Number of obs = 165
Time variable : tahun Number of groups = 33
Number of instruments = 15 Obs per group: min = 5
Wald chi2(0) = . avg = 5.00
Prob > chi2 = . max = 5
------------------------------------------------------------------------------
| Robust
lnIPM | Coefficient std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
lnIPM |
L1. | .9731091 .2590362 3.76 0.000 .4654076 1.480811
|
lnPLN | -.0121004 .0340991 -0.35 0.723 -.0789334 .0547326
lnSL | -.0082242 .0033951 -2.42 0.015 -.0148786 -.0015699
lnJALAN | -.0112446 .0083678 -1.34 0.179 -.0276453 .005156
lnSLS | -.002733 .0030194 -0.91 0.365 -.0086509 .0031848
lnSAML | .012342 .0128841 0.96 0.338 -.0129105 .0375945
tahun_2 | -.0008708 .002072 -0.42 0.674 -.0049318 .0031903
tahun_4 | -.0003193 .0034074 -0.09 0.925 -.0069977 .006359
tahun_5 | -.0114039 .0063957 -1.78 0.075 -.0239393 .0011315
tahun_6 | -.0048179 .0047947 -1.00 0.315 -.0142154 .0045795
tahun_7 | .0002146 .0056747 0.04 0.970 -.0109075 .0113368
------------------------------------------------------------------------------
Instruments for first differences equation
Standard
D.(lnPLN lnSL lnJALAN lnSLS lnSAML tahun_1 tahun_2 tahun_3 tahun_4 tahun_5
tahun_6 tahun_7)
GMM-type (missing=0, separate instruments for each period unless collapsed)
L(1/6).L.lnIPM collapsed
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z = -1.36 Pr > z = 0.173
Arellano-Bond test for AR(2) in first differences: z = -1.65 Pr > z = 0.099
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(4) = 2.04 Prob > chi2 = 0.728
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(4) = 6.43 Prob > chi2 = 0.169
(Robust, but weakened by many instruments.)
***Note : tahun_ is my time dummies***
.
here is what I want to ask:
1. Is it normal when my coefficient and significant level changed due to time dummies introduction to the model?
2. if I stick with my first code (without time dummies) what kind of special treatment i need to do to deal with invalid ar, sargan, and hansen statistic test?
Thank You!