Estimation of a System of Simultaneous equations- 3SLS

[Kath]

Dear all,
I am conducting a research which involves estimate the parameters of a system of simultaneous equations using Three-stage least squares. So, I select object/new object/system and I specified my two equations. The problem is that I have country dummies and I don’t know I to include them in the specification.

My specification is:
risk=c(11)+c(12)*indep+c(13)*size+c(14)*unit+ c(15)*mem+c(16)*cap+ c(17)*cro+c(18)*comt
indep=c(21)+c(22)*risk+c(23)*size+c(24)*unit+c(25)*ten+c(26)*age+c(27)*mbr
inst size unit mem cap cro comt ten age mbr

How can I specify the country dummies? Previously I have created “COUNTID” as the identifier of countries. So, I have:
COUNTID=AUSTRIA
COUNTID=BELGIUM


If I include the countries dummies as the presented above an error message appears in the scren.

risk=c(11)+c(12)*indep+c(13)*size+c(14)*unit+c(15)*mem+c(16)*cap+c(17)*cro+c(18)*comt+c(19)*COUNTID="AUSTRIA"
indep=c(21)+c(22)*risk+c(23)*size+ c(24)*unit+ c(25)*ten+c(26)*age+ c(27)*mbr+C(28)*COUNTID="AUSTRIA"
inst size unit mem cap cro comt ten age mbr COUNTID="AUSTRIA"


Could you please help me?
Thanks in advance.

[EViews Gareth]

Put parens around the dummy expression:

Code: Select all

+c(19)*(COUNTID="AUSTRIA")

[Kath]

Many thanks for your precious help.

[marvin_mah]

Hi,
I have a 3SLS that is giving me a "Near singular matrix" error when I include 2014 in my time series (1981-2014). However when I leave 2014 out (1981-2013) the system solves.
Additionally, I need initial parameters for my coefficients to avoid "Near singular matrix" for 1981-2013.
Can you please point me in the right direction?

Thanks

System: SYS01
Estimation Method: Three-Stage Least Squares
Date: 01/11/16 Time: 12:00
Sample: 1982 2013
Included observations: 33
Total system (unbalanced) observations 383
Estimation settings: tol=0.00010, derivs=analytic (linear)
Initial Values: C(1)=-145.438, C(2)=1.06951, C(3)=-0.15769, C(5)=26.1106,
C(7)=0.00304, C(8)=0.00229, C(9)=114.680, C(10)=-120.897,
C(13)=5.34915, C(14)=0.66254, C(15)=1.14918, C(16)=0.04861,
C(18)=-19388.9, C(19)=0.29376, C(20)=30.3530, C(21)=62.2162,
C(31)=0.25209, C(22)=-592.014, C(23)=0.83477, C(24)=0.02067,
C(25)=0.89788, C(26)=206.077, C(27)=-70.3275, C(28)=0.00386,
C(29)=1.00791, C(30)=-17846.5, C(32)=-14347.9, C(33)=8.21254,
C(47)=0.23358, C(34)=-23130.0, C(35)=308.873, C(36)=0.57219,
C(37)=-0.03172, C(48)=-26.0476, C(38)=-12764.9, C(39)=0.80323,
C(40)=0.83909, C(49)=-0.49879, C(51)=8937.48, C(52)=0.96347,
C(53)=0.62532, C(54)=-0.83593, C(55)=0.45000, C(56)=0.41755,
C(57)=0.82569, C(58)=1148.56, C(17)=0.68927, C(41)=0.51760,
C(50)=1.03044
Iterate coefficients after one-step weighting matrix
Convergence achieved after: 1 weight matrix, 22 total coef iterations

Coefficient Std. Error t-Statistic Prob.

C(1) -111.9673 14.67599 -7.629283 0.0000
C(2) 0.844722 0.114205 7.396564 0.0000
C(3) 0.086967 0.128667 0.675906 0.4996
C(5) 24.06437 3.414602 7.047490 0.0000
C(7) 0.003233 0.000485 6.660192 0.0000
C(8) 0.002216 0.000455 4.867910 0.0000
C(9) 179.5820 23.19049 7.743775 0.0000
C(10) -180.6735 20.65139 -8.748733 0.0000
C(17) 0.888834 0.125205 7.099003 0.0000
C(13) 5698.793 1475.092 3.863347 0.0001
C(14) 0.763377 0.051191 14.91231 0.0000
C(15) 0.570818 0.123853 4.608825 0.0000
C(16) 0.060176 0.039776 1.512887 0.1313
C(18) -19805.85 2787.268 -7.105829 0.0000
C(19) 0.300375 0.070185 4.279735 0.0000
C(20) 28.22466 3.885806 7.263526 0.0000
C(21) -10.73913 74.58624 -0.143983 0.8856
C(31) 0.318643 0.088031 3.619664 0.0003
C(22) 600.9806 575.2091 1.044804 0.2969
C(23) -0.478335 0.485196 -0.985860 0.3249
C(24) 0.032940 0.005855 5.625944 0.0000
C(25) 1.001904 0.045315 22.10959 0.0000
C(26) 26.60316 302.7639 0.087868 0.9300
C(27) -21.17465 23.17807 -0.913564 0.3616
C(28) 0.015259 0.005414 2.818411 0.0051
C(29) 0.915598 0.051574 17.75324 0.0000
C(30) -20293.05 4654.089 -4.360262 0.0000
C(32) -8438.556 4090.261 -2.063085 0.0399
C(33) 6.884565 0.381318 18.05466 0.0000
C(47) 0.320010 0.057675 5.548485 0.0000
C(34) -33963.61 4300.148 -7.898243 0.0000
C(35) 260.0861 41.65370 6.244009 0.0000
C(36) 0.657952 0.099012 6.645139 0.0000
C(37) 0.037524 0.122285 0.306859 0.7591
C(48) 652.6425 256.7812 2.541629 0.0115
C(38) -20601.30 3427.809 -6.010048 0.0000
C(39) 0.598268 0.056054 10.67302 0.0000
C(40) 1.061391 0.068384 15.52115 0.0000
C(41) 0.668223 0.125768 5.313142 0.0000
C(49) 1.000000 9.59E-17 1.04E+16 0.0000
C(50) 1.008439 0.027412 36.78788 0.0000
C(51) 12288.64 3561.414 3.450494 0.0006
C(52) 0.652736 0.138859 4.700719 0.0000
C(53) 0.912479 0.184423 4.947752 0.0000
C(54) -0.610051 0.332865 -1.832730 0.0677
C(55) 0.483797 0.152868 3.164798 0.0017
C(56) 0.276819 0.175742 1.575146 0.1162
C(57) 0.835920 0.045352 18.43175 0.0000
C(58) 1499.889 986.4165 1.520543 0.1293

Determinant residual covariance 1.60E+29

[EViews Gareth]

Dummy variable trap?

[marvin_mah]

I'm not sure if that is the issue. I tried to remove the dum and it still didn't solve.


Here's my system:
LF = c(1) + c(2)*Pop1864 + c(3)*LF(-1)

emp-emp(-1) = c(5) + c(7)*(Inv(-1)-Inv(-2)) + c(8)*(Inv-Inv(-1))

netmigration = c(9) + c(10)*((1-emp/LF)/canunemp) + [ar(1)=c(17)]

ABDispInc = c(13) + c(14)*Persex + c(15)*(govcons+govinv) + c(16)*Inv(-1)

Persex = c(18) + c(19)*ABDispInc + c(20)*Emp + c(21)*i + c(31)*Persex(-1)

GovCons = c(22) + c(23)*ABPop + c(24)*ogrev + c(25)*GovCons(-1)

GovInv = c(26) + c(27)*(i-ABinflation) + c(28)*ogrev + c(29)*GovInv(-1)

OX = c(30) + c(32)*Fx + c(33)*USGDP + c(47)*Inv

Inv = c(34) + c(35)*rpwtican + c(36)*X + c(37)*persex + c(48)*(i-ABinflation)

M = c(38) + c(39)*Inv + c(40)*persex + [ar(1) = c(41)]

PersInc = c(49)*ABDispInc + [ar(1) = c(50)]

rgdp = c(51) + c(52)*consump + c(53)*govcons + c(54)*govinv + c(55)*inv + c(56)*m1 + c(57)*x + c(58)*dum

Inst ABPop Pop1864 Inv(-1) (i-ABinflation) ogrev rpwtican Fx USGDP govcons(-1) govinv(-1) dum persex(-1)

[EViews Gareth]

Might be that the system is really really close to singular, and adding that extra observation pushes it to being numerically indistinguishable from singular.

[marvin_mah]

What's would be the best way of determining that?...Sorry I'm a newbie at Eviews.