Coefficient that vary over time and Quandt-Andrews Breakpoin

[katrin]

Hi,

I've got data from 15 countries over 56 quarters.
1) I've got some common coefficient and cross-section specific effects (-> for each country specific effect). Now I want to implement a variable/ coefficient that vary over time. It should be a kind of dummy, that shows time dependent effects that are independent from the underlying fundamentals (common coefficients). If I implement period random effects, the results are "wrong" -> I've got results from a stata output but I have to reestimate it and to add some quarters. How can I implement such a variable, that varys over time independently from the underlying values?

2) I want to perform a Quandt-Andrews Breakpoint Test for this sample. Is it possible to perform this test for pooled data? If no, is there any possibility to programm it, or to get the code? I'm not that good in programming...

If you need more information, please tell me.
THANKS

[katrin]

Hi again,

I've searched a little about "time-dummy". If I put "@quarter" in the box for period specific coefficient it works. I get results that aren't that bad. The problem now, if I want to control for country specific effets I get the message "near singular matrix". I've tried "@crossid" and cross section fixed effects. Am I doing this wrong? Do somebody have experience with controling for country specific effects? Pool data over 10 coutries, 47 quarters

I really would need some help. If someone just has an idea I would be very pleased!
Thanks,

Katrin