Serial correlation

[moirapasseron]

Hello, I estimated the following multiple linear regression in Eviews:


Dependent Variable: LNPIBPC
Method: Least Squares
Date: 08/17/10 Time: 15:39
Sample: 1970 2000
Included observations: 31

Variable Coefficient Std. Error t-Statistic Prob.

C 6.666410 0.559503 11.91489 0.0000
Z 0.541974 0.056812 9.539794 0.0000
CC -0.944654 0.353005 -2.676033 0.0125
W 0.246999 0.044910 5.499807 0.0000

R-squared 0.839519 Mean dependent var 11.85432
Adjusted R-squared 0.821687 S.D. dependent var 0.308671
S.E. of regression 0.130343 Akaike info criterion -1.117385
Sum squared resid 0.458709 Schwarz criterion -0.932354
Log likelihood 21.31946 Hannan-Quinn criter. -1.057069
F-statistic 47.08124 Durbin-Watson stat 1.111689
Prob(F-statistic) 0.000000

And I'm doubting about the Durbin Watson statistic, I think is kind of low but to make a better analysis i've looked the RESIDUALS GRAPH (attached below) but I'm not sure i'm interpretating it well.
The "actual" and "fitted" curves are the ones of the regression right? And about the Residual curve, can i say it shows a serial correlation or not?

Thanks,

Moira

[startz]

With that Durbin-Watson you definitely have serial correlation.

[moirapasseron]

With that Durbin-Watson you definitely have serial correlation.

Is there any Eviews tool to fix the correlation? something about AR(1)?
Thanks for your time

[startz]

Yes, basically add ar(1) to the list of variables.