# Re: st: xtivreg, fe , multiple reg with interaction effects

 From Baktigul Turdubekova To statalist@hsphsun2.harvard.edu Subject Re: st: xtivreg, fe , multiple reg with interaction effects Date Fri, 5 Nov 2004 07:31:17 -0800 (PST)

```> 1. Sorry, my question wasn't spelled out correctly.
> The models I need to implement are the following:
>
> 1) Y = const + Z*betta + time_effects*theta + u
> 2) Y = const + Z*betta + time_effects*theta +
> time_effects*Fixed effects*delta + u

>>Are your time effects actually a variable like year?
>>At any rate, I
>>don't think that this thing is easily estimable,
>>because you don't
>>have fixed effects to generate your interaction
>>terms. You can try
>>this out within the random effect framework with
->>gllamm-.

Stas,

Yes, time effects are years, but taken through decades
as four periods.

Regarding the interaction term, I suspect this thing
should be easily done in STATA but not sure which way
is correct. The term Time_effect*Fixed_Effect
introduces a structural break into the fixed effect
for the time period considered. Specifically, what I
want to do is the following:

Say, time1=1, time=1960s, 0 else

FE_UK=1, country=United Kingdom, 0 else

Then, time1*FE_UK=1, time=1960s in United Kingdom

So, in the regression we have terms:

....+ betta0*FE_UK + betta1*Time1*FE_UK

where betta0 is the estimate for the strength of FE in
all periods except for 1960s and betta1 is the
estimate for FE strength in 1960s.

I'm thinking of introducing dummies to each of the 140
countries and 4 time points and just multiply them
out, and run the regression. Can I do it? If no, why?

Baktigul

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```