I have a balanced data of N=26 and T=1. Since I have a dynamic panel data I plan to use the GMM estimator. However, I am not sure whether I should choose the difference- or system-GMM. The existing literature makes extensive use of the system-GMM, but given my sample size and when I use the system-GMM, I get the following:
I have also tried to the difference-GMM using the command -xtdpd- and I get the following:
Given the above two estimations, can somebody please advice me as to how can I decide if I should use the difference- or system-GMM?
Code:
xtabond2 patr1 l.patr1 elec_pred elec_end right left electpredright electpredleft elec
> tendright electendleft l.lngdp l.inflation l.govtexp l.popbelow14 l.pop65andabove, gmm
> (l.patr1 l.govtexp l.inflation l.lngdp, lag (1 1) collapse eq(diff)) iv(elec_pred elec
> _end right left electpredright electpredleft electendright electendleft l.popbelow14 l
> .pop65andabove, eq(diff)) gmm(l.patr1 l.govtexp l.inflation l.lngdp, lag (1 1) collaps
> e eq(level))twostep robust
Favoring speed over space. To switch, type or click on mata: mata set matafavor space, p
> erm.
Dynamic panel-data estimation, two-step system GMM
------------------------------------------------------------------------------
Group variable: country Number of obs = 516
Time variable : year Number of groups = 26
Number of instruments = 19 Obs per group: min = 19
Wald chi2(14) = 1389.44 avg = 19.85
Prob > chi2 = 0.000 max = 20
--------------------------------------------------------------------------------
| Corrected
patr1 | Coef. Std. Err. z P>|z| [95% Conf. Interval]
---------------+----------------------------------------------------------------
patr1 |
L1. | .9843186 .1015519 9.69 0.000 .7852805 1.183357
|
elec_pred | -.0005532 .0032641 -0.17 0.865 -.0069507 .0058443
elec_end | -.006229 .0091323 -0.68 0.495 -.0241279 .0116699
right | -.0003428 .0042954 -0.08 0.936 -.0087616 .008076
left | .0015003 .0043135 0.35 0.728 -.006954 .0099546
electpredright | -.0001101 .0067846 -0.02 0.987 -.0134077 .0131875
electpredleft | -.0065637 .0049074 -1.34 0.181 -.0161821 .0030547
electendright | .0050187 .0085232 0.59 0.556 -.0116865 .0217239
electendleft | .0100014 .0101061 0.99 0.322 -.0098062 .0298089
|
lngdp |
L1. | .0036574 .0041362 0.88 0.377 -.0044494 .0117643
|
inflation |
L1. | .0008939 .0007865 1.14 0.256 -.0006477 .0024354
|
govtexp |
L1. | .0131019 .0457429 0.29 0.775 -.0765526 .1027564
|
popbelow14 |
L1. | .1104197 .2028556 0.54 0.586 -.2871701 .5080095
|
pop65andabove |
L1. | -.1206514 .1884528 -0.64 0.522 -.4900121 .2487093
|
_cons | -.0980632 .1206471 -0.81 0.416 -.3345271 .1384008
--------------------------------------------------------------------------------
Instruments for first differences equation
Standard
D.(elec_pred elec_end right left electpredright electpredleft
electendright electendleft L.popbelow14 L.pop65andabove)
GMM-type (missing=0, separate instruments for each period unless collapsed)
L.(L.patr1 L.govtexp L.inflation L.lngdp) collapsed
Instruments for levels equation
Standard
_cons
GMM-type (missing=0, separate instruments for each period unless collapsed)
DL.(L.patr1 L.govtexp L.inflation L.lngdp) collapsed
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z = -2.39 Pr > z = 0.017
Arellano-Bond test for AR(2) in first differences: z = 0.70 Pr > z = 0.482
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(4) = 12.73 Prob > chi2 = 0.013
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(4) = 5.00 Prob > chi2 = 0.288
(Robust, but weakened by many instruments.)
Difference-in-Hansen tests of exogeneity of instrument subsets:
GMM instruments for levels
Hansen test excluding group: chi2(0) = 0.00 Prob > chi2 = .
Difference (null H = exogenous): chi2(4) = 5.00 Prob > chi2 = 0.288
gmm(L.patr1 L.govtexp L.inflation L.lngdp, collapse eq(diff) lag(1 1))
Hansen test excluding group: chi2(0) = 0.00 Prob > chi2 = .
Difference (null H = exogenous): chi2(4) = 5.00 Prob > chi2 = 0.288
gmm(L.patr1 L.govtexp L.inflation L.lngdp, collapse eq(level) lag(1 1))
Hansen test excluding group: chi2(0) = 0.00 Prob > chi2 = .
Difference (null H = exogenous): chi2(4) = 5.00 Prob > chi2 = 0.288
Code:
xtdpd patr1 l.patr1 elec_pred elec_end right left electpredright electpredleft electen
> dright electendleft l.govtexp l.inflation l.lngdp l.popbelow14 l.pop65andabove, div(el
> ec_pred elec_end right left electpredright electpredleft electendright electendleft l.
> govtexp l.inflation l.lngdp l.popbelow14 l.pop65andabove) dgmmiv(patr1) twostep
Dynamic panel-data estimation Number of obs = 516
Group variable: country Number of groups = 26
Time variable: year
Obs per group: min = 19
avg = 19.84615
max = 20
Number of instruments = 204 Wald chi2(14) = 4326.34
Prob > chi2 = 0.0000
Two-step results
--------------------------------------------------------------------------------
patr1 | Coef. Std. Err. z P>|z| [95% Conf. Interval]
---------------+----------------------------------------------------------------
patr1 |
L1. | -.1364141 .4727951 -0.29 0.773 -1.063075 .7902473
|
elec_pred | .0036256 .0026501 1.37 0.171 -.0015684 .0088197
elec_end | .0185794 .0145694 1.28 0.202 -.0099761 .047135
right | .0039098 .0050598 0.77 0.440 -.0060071 .0138268
left | .003864 .0034252 1.13 0.259 -.0028494 .0105773
electpredright | -.0044389 .0027864 -1.59 0.111 -.0099001 .0010222
electpredleft | -.0078717 .0038704 -2.03 0.042 -.0154576 -.0002858
electendright | -.0190151 .0144789 -1.31 0.189 -.0473932 .009363
electendleft | -.0118486 .0131506 -0.90 0.368 -.0376232 .0139261
|
govtexp |
L1. | -.613838 .3344723 -1.84 0.066 -1.269392 .0417156
|
inflation |
L1. | -.0035736 .0040095 -0.89 0.373 -.0114321 .0042848
|
lngdp |
L1. | -.1978443 .0707709 -2.80 0.005 -.3365528 -.0591359
|
popbelow14 |
L1. | -1.542465 .4905102 -3.14 0.002 -2.503847 -.5810828
|
pop65andabove |
L1. | -.2904311 .2420565 -1.20 0.230 -.7648531 .1839909
|
_cons | 5.953773 2.118015 2.81 0.005 1.802541 10.10501
--------------------------------------------------------------------------------
Warning: gmm two-step standard errors are biased; robust standard
errors are recommended.
Instruments for differenced equation
GMM-type: L(2/.).patr1
Standard: D.elec_pred D.elec_end D.right D.left D.electpredright
D.electpredleft D.electendright D.electendleft LD.govtexp
LD.inflation LD.lngdp LD.popbelow14 LD.pop65andabove
Instruments for level equation
Standard: _cons
. estat abond
Arellano-Bond test for zero autocorrelation in first-differenced errors
+-----------------------+
|Order | z Prob > z|
|------+----------------|
| 1 |-.28245 0.7776 |
| 2 | .17322 0.8625 |
+-----------------------+
H0: no autocorrelation
. estat sargan
Sargan test of overidentifying restrictions
H0: overidentifying restrictions are valid
chi2(189) = 5.92085
Prob > chi2 = 1.0000
