How to use dummy variables in a time-series regression?

[NTUSenna]

Hi, I'm new to eviews, though I do have some basic econometrics experience.

I'm looking to create a regression to measure the determinants of GDP growth given 5 variables: Investment, Savings, Government Expenditure, Exports and Interest Rates.
I am using annual data from 1960 to 2010 and would like to explore a pre-1997 and post-1997 era for analysing how the variables' influence changed in the two periods. But to improve the reliability of the results rather than split the data set into two regressions (1960-1997 and 1998-2010), I thought I could use dummy variables instead and allow me to keep a longer range of data.

Could I use dummy variables for each variable to control for the two periods i.e 1 = <1997 and 0 > 1997, and if so how?

Many thanks.

[EViews Gareth]

http://forums.eviews.com/viewtopic.php?f=4&t=3840

[NTUSenna]

Thanks for that, I managed to get two dummy variables working for use in regressions to test <1997 (dummy) and >1997 (dummy2) effects. The results I got was that the coefficients etc where just the opposite values for the two dummy variables.

Variable.......................................Coefficient Std. Error t-Statistic Prob.
INVESTMENT*DUMMY.......................-0.311363 0.235947 -1.319632 0.2045
INVESTMENT*DUMMY2..................... 0.311363 0.235947 1.319632 0.2045

This was a result of doing two estimations "gdpgrowth c investment (investment*dummy)" and "gdpgrowth c investment (investment*dummy2)"

Is this supposed to happen because it seems a bit simple results?

Thanks

[startz]

Could you show the commands you used?

[NTUSenna]

So for the dummy variables I created them with:

Code: Select all

dummy = @recode (@year <1997, 1, 0)

and

Code: Select all

dummy2 = @recode (@year <1997, 0, 1)

Then I ran two estimations to test the original variable and each dummy separately:

Code: Select all

gdpgrowth c savings (savings*dummy) investment (investment*dummy) govexp (govexp*dummy) exports (exports*dummy)

and

Code: Select all

gdpgrowth c savings (savings*dummy2) investment (investment*dummy2) govexp (govexp*dummy2) exports (exports*dummy2)

[startz]

You're getting the right result. The first equation says the coefficient is 3 higher in the first period. The second says it's 3 lower in the second. Both mean the same thing.

[NTUSenna]

Apologies for the late reply, but thank you for the assistance. I think I understand it now!