Hello,
I have a question regarding running the xtabond2 command, with alternating years of data. I am using an unbalanced database of firms with three years of data (2011, 2013, 2015). I would like to estimate a system GMM model for labor productivity, and as GMM style instruments, I want to use the first and second lags of the variables (the first lag in this case is information from 2 years ago, and the second lag, whenever relevant, is from four years ago). xtabond2 gives the following error message, if I try to estimate the model with the years coded as above:
So, I created a new time variable,
and then ran the command for xtabond2 above. The results are as follows.
My question is regarding the computation of the Hansen chi-squared statistic. The chi-squared statistic is computed assuming I have three consecutive periods of data, however this is not the case (this is also why I am using the first and second lags as instruments, because in reality, they represent lags of 2 years and 4 years respectively)). Is there some way to estimate the model using non-consecutive years of data, or to recompute this Hansen statistic assuming that the lags are actually for the second and fourth periods, instead of the first and second (I might be wrong, but my hunch is that the null hypothesis is being rejected so strongly, because of the presence of the gmm style instruments in the first lag form)?
I have a question regarding running the xtabond2 command, with alternating years of data. I am using an unbalanced database of firms with three years of data (2011, 2013, 2015). I would like to estimate a system GMM model for labor productivity, and as GMM style instruments, I want to use the first and second lags of the variables (the first lag in this case is information from 2 years ago, and the second lag, whenever relevant, is from four years ago). xtabond2 gives the following error message, if I try to estimate the model with the years coded as above:
Code:
xtset q1 Year
panel variable: q1 (unbalanced)
time variable: Year, 2011 to 2015, but with gaps
delta: 1 unit
Code:
xtabond2 lnlabprod_v2 lncaplab_v2 lnrawlab_v2 lnnplab_v2 pc_aq pc_fire pc_light q122 q124 q117a6 q101c q111b q115a q34a q36a q29a q105 i.q17a i.Year if (q13a == 1 & q11a == 0), gmm (lnlabprod_v2 lncaplab_v2 lnrawlab_v2 lnnplab_v2 pc_aq pc_fire pc_light q122 q124 q117a6 q101c q111b q115a q34a q36a q29a q105 , lag(1 2)) iv(i.Year, eq(level)) iv(i.q17a, eq(both)) robust twostep
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.
Difference-in-Sargan/Hansen statistics may be negative.
estimates post: matrix has missing values
stata(): 3598 Stata returned error
xtabond2_mata(): - function returned error
<istmt>: - function returned error
So, I created a new time variable,
Code:
gen time = 1 if Year == 2011 replace time= 2 if Year == 2013 replace time = 3 if Year == 2015
Code:
. xtabond2 lnlabprod_v2 lncaplab_v2 lnrawlab_v2 lnnplab_v2 pc_aq pc_fire q122 q124 q117a6 q101c q111b q115a q34a q36a q29a q105 i.q1
> 7a i.Year if (q13a == 1 & q11a == 0), gmm (lnlabprod_v2 lncaplab_v2 lnrawlab_v2 lnnplab_v2 pc_aq pc_fire q122 q124 q117a6 q101c q
> 111b q115a q34a q36a q29a q105, lag(1 2)) iv(i.Year, eq(level)) iv(i.q17a, eq(both)) robust twostep
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 optimal weighting matrix for two-step estimation.
Difference-in-Sargan/Hansen statistics may be negative.
Dynamic panel-data estimation, two-step system GMM
------------------------------------------------------------------------------
Group variable: q1 Number of obs = 5679
Time variable : time Number of groups = 2983
Number of instruments = 102 Obs per group: min = 1
Wald chi2(39) = 22624.99 avg = 1.90
Prob > chi2 = 0.000 max = 3
-----------------------------------------------------------------------------------------------------------------------------------
| Corrected
lnlabprod_v2 | Coef. Std. Err. z P>|z| [95% Conf. Interval]
------------------------------------------------------------------+----------------------------------------------------------------
lncaplab_v2 | .0354353 .0430597 0.82 0.411 -.0489602 .1198307
lnrawlab_v2 | .1227379 .0259091 4.74 0.000 .071957 .1735189
lnnplab_v2 | .3460553 .0384345 9.00 0.000 .2707251 .4213856
pc_aq | 4.34e-06 1.92e-06 2.26 0.024 5.81e-07 8.09e-06
pc_fire | -1.08e-06 7.95e-07 -1.35 0.176 -2.63e-06 4.83e-07
q122 | .0329375 .0496159 0.66 0.507 -.0643078 .1301829
q124 | .1886265 .0834946 2.26 0.024 .0249801 .3522729
q117a6 | .0005963 .0004886 1.22 0.222 -.0003612 .0015539
q101c | .0015383 .0011335 1.36 0.175 -.0006834 .0037599
q111b | .0792683 .3197831 0.25 0.804 -.547495 .7060316
q115a | .1115387 .1038907 1.07 0.283 -.0920833 .3151608
q34a | .0210799 .0097299 2.17 0.030 .0020097 .0401501
q36a | .1160642 .1216182 0.95 0.340 -.122303 .3544315
q29a | -.0012474 .0009021 -1.38 0.167 -.0030154 .0005207
q105 | -.002237 .1018361 -0.02 0.982 -.2018321 .1973581
|
q17a |
Agriculture | 0 (empty)
Food and beverages | .3506242 .4971583 0.71 0.481 -.6237882 1.325037
Tobacco | .4395318 .5532646 0.79 0.427 -.6448469 1.523911
Textiles | .2753416 .4979738 0.55 0.580 -.7006691 1.251352
Apparel | .0278234 .4942391 0.06 0.955 -.9408675 .9965142
Leather | .2896563 .5070813 0.57 0.568 -.7042048 1.283517
Wood | .1764778 .5001476 0.35 0.724 -.8037935 1.156749
Paper | .4107523 .4950751 0.83 0.407 -.559577 1.381082
Publishing and printing | .3756701 .4917285 0.76 0.445 -.5881 1.33944
Refined petroleum etc. | .2307311 .5362108 0.43 0.667 -.8202228 1.281685
Chemical products etc. | .3164512 .5179295 0.61 0.541 -.698672 1.331574
Rubber | .3203078 .4937529 0.65 0.517 -.64743 1.288046
Non-metallic mineral products | .0294455 .5063753 0.06 0.954 -.9630319 1.021923
Basic metals | .6232988 .5141709 1.21 0.225 -.3844577 1.631055
Fabricated metal products | .3364873 .4934035 0.68 0.495 -.6305657 1.30354
Electronic machinery, computers, radio, tv, etc. | .5475978 .4950758 1.11 0.269 -.422733 1.517929
Motor vehicles etc. | -.0088229 .5261101 -0.02 0.987 -1.03998 1.022334
Other transport equipment | .4125957 .5896561 0.70 0.484 -.7431091 1.5683
Furniture, jewellery, music equipment, watches, toys and medic.. | .1689693 .4968482 0.34 0.734 -.8048353 1.142774
Recycling etc. | -.1834269 .5786 -0.32 0.751 -1.317462 .9506083
Services | 0 (omitted)
|
Year |
2011 | 0 (empty)
2013 | -4.525192 .0364661 -124.09 0.000 -4.596664 -4.453719
2015 | -4.416423 .0372227 -118.65 0.000 -4.489378 -4.343468
|
_cons | 15.94037 .906944 17.58 0.000 14.1628 17.71795
-----------------------------------------------------------------------------------------------------------------------------------
Instruments for first differences equation
Standard
D.(0b.q17a 1.q17a 2.q17a 3.q17a 4.q17a 5.q17a 6.q17a 7.q17a 8.q17a 9.q17a
10.q17a 11.q17a 12.q17a 13.q17a 14.q17a 15.q17a 16.q17a 17.q17a 18.q17a
19.q17a 20.q17a)
GMM-type (missing=0, separate instruments for each period unless collapsed)
L(1/2).(lnlabprod_v2 lncaplab_v2 lnrawlab_v2 lnnplab_v2 pc_aq pc_fire q122
q124 q117a6 q101c q111b q115a q34a q36a q29a q105)
Instruments for levels equation
Standard
0b.q17a 1.q17a 2.q17a 3.q17a 4.q17a 5.q17a 6.q17a 7.q17a 8.q17a 9.q17a
10.q17a 11.q17a 12.q17a 13.q17a 14.q17a 15.q17a 16.q17a 17.q17a 18.q17a
19.q17a 20.q17a
2011b.Year 2013.Year 2015.Year
_cons
GMM-type (missing=0, separate instruments for each period unless collapsed)
D.(lnlabprod_v2 lncaplab_v2 lnrawlab_v2 lnnplab_v2 pc_aq pc_fire q122 q124
q117a6 q101c q111b q115a q34a q36a q29a q105)
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z = -5.68 Pr > z = 0.000
Arellano-Bond test for AR(2) in first differences: z = . Pr > z = .
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(62) =1162.57 Prob > chi2 = 0.000
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(62) = 183.34 Prob > chi2 = 0.000
(Robust, but weakened by many instruments.)
Difference-in-Hansen tests of exogeneity of instrument subsets:
GMM instruments for levels
Hansen test excluding group: chi2(30) = 79.08 Prob > chi2 = 0.000
Difference (null H = exogenous): chi2(32) = 104.26 Prob > chi2 = 0.000
iv(2011b.Year 2013.Year 2015.Year, eq(level))
Hansen test excluding group: chi2(60) = 169.46 Prob > chi2 = 0.000
Difference (null H = exogenous): chi2(2) = 13.89 Prob > chi2 = 0.001
iv(0b.q17a 1.q17a 2.q17a 3.q17a 4.q17a 5.q17a 6.q17a 7.q17a 8.q17a 9.q17a 10.q17a 11.q17a 12.q17a 13.q17a 14.q17a 15.q17a 16.q17a
> 17.q17a 18.q17a 19.q17a 20.q17a)
Hansen test excluding group: chi2(43) = 160.39 Prob > chi2 = 0.000
Difference (null H = exogenous): chi2(19) = 22.96 Prob > chi2 = 0.239
