cross section sur-weights?DW stat

[fionarh]

hi there..
I'm using eviews 6
could you tell me what the diffrence between cross-section weights and cross-section sur? I also don't understand about inconclusive Durbin-Watson stat..

>> I'm doing my thesis with panel data (N= 4, T=15, parameter=6)
after being tested the model is fixed effect.
when I put cross section weights on GLS weights and coef covariance method: ordinary, its result:

Dependent Variable: LNVM
Method: Panel EGLS (Cross-section weights)
Date: 03/27/12 Time: 21:04
Sample: 1996 2010
Periods included: 15
Cross-sections included: 4
Total panel (balanced) observations: 60
Linear estimation after one-step weighting matrix

Variable Coefficient Std. Error t-Statistic Prob.

LNY 1.205451 0.172478 6.989002 0.0000
LNER 1.003585 0.211068 4.754802 0.0000
LNTI 0.606960 0.068607 8.846889 0.0000
DCR -0.619337 0.131172 -4.721547 0.0000
DCAFTA 7.197720 0.680220 10.58146 0.0000
C -37.47860 6.219131 -6.026340 0.0000

Effects Specification

Cross-section fixed (dummy variables)

Weighted Statistics

R-squared 0.922621 Mean dependent var 18.65101
Adjusted R-squared 0.910483 S.D. dependent var 11.96971
S.E. of regression 0.339130 Sum squared resid 5.865473
F-statistic 76.01201 Durbin-Watson stat 1.101326
Prob(F-statistic) 0.000000

Unweighted Statistics

R-squared 0.902665 Mean dependent var 12.53870
Sum squared resid 6.734975 Durbin-Watson stat 0.945752

because the weighted sum square resid(SSR) is < unweighted SSR it means there's heteroscedasticity (is it true?). so that I change the GLS weights into cross section SUR and the result:

Dependent Variable: LNVM
Method: Panel EGLS (Cross-section SUR)
Date: 03/27/12 Time: 21:09
Sample: 1996 2010
Periods included: 15
Cross-sections included: 4
Total panel (balanced) observations: 60
Linear estimation after one-step weighting matrix

Variable Coefficient Std. Error t-Statistic Prob.

LNY 1.135411 0.174917 6.491146 0.0000
LNER 0.789081 0.184399 4.279212 0.0001
LNTI 0.601056 0.056445 10.64855 0.0000
DCR -0.419443 0.127046 -3.301509 0.0018
DCAFTA 7.246325 0.571141 12.68744 0.0000
C -33.75415 6.183727 -5.458545 0.0000

Effects Specification

Cross-section fixed (dummy variables)

Weighted Statistics

R-squared 0.914448 Mean dependent var 21.55627
Adjusted R-squared 0.901028 S.D. dependent var 60.10012
S.E. of regression 0.951442 Sum squared resid 46.16734
F-statistic 68.14119 Durbin-Watson stat 1.300408
Prob(F-statistic) 0.000000

Unweighted Statistics

R-squared 0.895877 Mean dependent var 12.53870
Sum squared resid 7.204674 Durbin-Watson stat 0.945225

there's no more heteroscedacity.

>> when I see the DW stat (1.300) based the table in my book, it means inconclusive. there's no further explanation.what does inconlusive mean? Is it really matter to my estimation?

Thanks.
NB: I hope you can understand my question, because I'm from south East asia that rarely speaks english