help with gravity model

[thehighyam]

Hi guys,
I need your assistance, I am a newbie for Eviews

There is a model that I must explain the estimation output.
Here is the model.
log (tradeij) = β0 + β1log(GDPi*GDPj) + β2log(DIFGDPij) + β3log(RERij) + β4RTAij + εij

trade ij = trade between country i and country j (e.g. between Turkey and Germany)
β0, β1, β2, β3, β4 = unknown parameters
GDPi = Gross domestic product for country i
GDPj = Gross domestic product for country j
DIFGDPij = GDP difference between country i and country j
log(RERij) = Real Exchange Rate
RTAij = Regional Trade Agreements between country i and country j ( RTA takes only 0 or 1 values)
εij = error term

Here is the estimation output of e-views.

Dependent Variable: LOG(TRADE)
Method: Least Squares
Date: 03/27/10 Time: 22:08
Sample: 1 494
Included observations: 494

________________________________Coefficient__||__Std. Error__||__t-Statistic__||__Prob.
C_______________________________-12.31026__||___0.819165__||__-15.02781__||__0.0000
LOG(GDP_TR*GDP_EU)______________0.964166__||___0.058991__||__16.34430___||__0.0000
LOG(GDP_DIF)_____________________0.056271__||___0.117149__||__0.480334___||__0.6312
LOG(RER)_________________________0.022153__||___0.054725__||__0.404804___||__0.6858
RTA______________________________1.838773__||___0.394237__||__4.664126___||__0.0000

R-squared 0.455116
Mean dependent var -1.265413
Adjusted R-squared 0.450658
S.D. dependent var 4.514747
S.E. of regression 3.346221
Akaike info criterion 5.263610
Sum squared resid 5475.428
Schwarz criterion 5.306146
Log likelihood -1295.112
Hannan-Quinn criter. 5.280310
F-statistic 102.1095
Durbin-Watson stat 0.785040
Prob(F-statistic) 0.000000

I don't know what should I do know? What is the next step?
The model is significant or not?
What tests should I do for the model? etc.

Thanks for your support
Kind Regards.
Engin

P.S : I uploaded the eviews workfile.

[tchaithonov]

Two of the variables in your model (LOG(GRP_DIF) and LOG_RER)) are not significant. Although your F-stat suggested that the model structure has no underlying problems, I am not sure if having both LOG(GDP_TR*GDP_EU) and LOG(GDP_DIF) help explaining anything. I am not sure if you are trying to follow a theory of any kind here, but I guess if you look at trade data, you should be looking more from a time-series standpoint instead of the cross-sectional one that you are dealing with at them moment. Guys, any thoughts?

Tchaithonov

[thehighyam]

Hi I have changed the datas. And this estimation output is significant now I think.
You can find in attached file.
I am waiting your comments.
Thanks
Engin

[estrella bernabeu]

Hi,
I have the same model like yours, but i'm opering with Tunisia not Turkey and i search for the impact of the volatility of the exchange rate on the trade so in my model i add the volatility variable and the superficy.

i have no idea about the methodology with i can process, so if someone can help us it will be very nice