Estimating only time dummies in Panel Regression

[abcxyz]

I would like to know how can I get the value of the estimators of time dummies only( and also both time and cross sectional dummies ) in Panel Regression in E -Views 6. Further how can I see the interaction effect of time dummies with each regressor ( i.e. time dummies and slopes of regressors ). I would be really obliged if I am made aware of the process as soon as possible.

[EViews Glenn]

Use C as the only regressor and specify the effects that you want.

The interaction variables are harder. You can use @expand to create a full set of dummy variables (e.g., X*@expand(@year)), but avoiding the dummy variable trap can be tricky. I'd probably not include a constant, not include the in-built fixed effects and just use @expands. This, of course, is tricky for the large dimension.

[abcxyz]

Hi, I tried but I m a bit confused.

I want to estimate the constants for each time period ( like in my study there are 3 periods ) further I want to estimate the slope of independent variables in different time periods ( like bo(X1) + b1(D1* X1) + b2(D2*X1) and ; when D1=D2=0, it is the slope of period 0; when D1=1, (bo+b1)X1 is the slope of period 1 and when D2=1, (bo+b2)X1 is the slope of period 2.

How can I estimate EACH equation for different periods to verify whether the intercepts and/or slopes are statistically different o not for the three periods.

I wd be obliged if u can help me out.

Regards.

[EViews Glenn]

You can use the @expand to create the period dummies. Suppose you have Y and X and you have annual data. Then

y @expand(@year) @expand(@year)*X

give you period dummies and period interactions. You could use the built-in period dummies and drop the first term, or normalize the interaction dummies by including the X alone and then an expand dropping a category. Something like

y c x @expand(@year, @droplast)*X

where the C is included for the built-in period dummies. Note that the normalizations become tricky if you also have cross-section effects.

[abcxyz]

Thanks a Lot.

What I have done is that I have made seperate tables for each dummy and further created tables for interaction effects also ( both intercept and slopes ) and have considered the constant as the mean of the category that is not included. I think it will work too.

Finally to interpret and compare the effect of each time period I have made different equations based on the values of the dummies.

And u r right in saying that cross sections interacton with time has to be tricky and would also consume the degrees of freedom.

Thanks a lot for ur time, I will now try to use the method u proposed.

Regards.