Re: st: Structural breaks in panel data

[Benedikt Achatz <benedikt.achatz.sta@gmail.com>]

Hello Tanja,

I am not sure that I get your question completely, but here is what I 
think you are asking for:
If you look at interaction variables, you are looking at what the effect 
of x1 is in each period; if you look at the effects of dummy variables 
itself, you look at what influence the time period has on the outcome in 
each period.
So, if x1 is the number of hours spent on education in a time period, 
then the coefficent of x1*P1 would tell you the effect one more unit of 
education has in that time period on the outcome (let's say monthly 
income). The coefficent of P1, however, would give you the effect of the 
period itself on the outcome (for instance, monthly income is likely to 
increase with time occupied, regardless of education)
If coefficents are insignifcant, they are just that, as with any other 
regression model...
Whether you want to include dummies themselves or just the interaction 
variables is, as far as I know, up to you. It doesn't hurt to write up a 
do file that does both and compare the results, though.
Hope this helps!

Am 04.07.2012 14:38, schrieb Tanja Berg:
> Dear Statalisters,
>
> I have an unbalanced panel data set with structural breaks.
> In my regression I want to show if and how the dependent variable (depvar) changes during several periods.
>
> I implemented four different Dummy variables (P2-P5) for five different periods.
>
> So my regression looks like the following:
> xtreg depvar x1  x2  P2  P3  P4  P5  x1*P2  x1*P3  x1*P4  x1*P5  x2*P2  x2*P3  x2*P4  x2*P5, fe
>
> x1 and x2 are my explanatory variables.
> If the coefficient of the interaction variable (e.g. x1*P2) is significant I have an effect here in this period. So far I understand what is going on.
>
> But what do the coefficients of the Dummy variables itself tell me? What if they are insignificant?
>
> If I only want to look at one explanatory variable, do I have to include the Dummy variables here as well or is it sufficient to look only at the interaction variables like following:
> xtreg depvar x1  x1*P2  x1*P3  x1*P4  x1*P5, fe
>
> Hope someone can help me with this little understanding problem!
>
> Best regards
> Tanja
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/