[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

# Re: st: Re: Rsquared within using id & time dummies. xtreg returns collineality error

 From "Clive Nicholas" To statalist@hsphsun2.harvard.edu Subject Re: st: Re: Rsquared within using id & time dummies. xtreg returns collineality error Date Fri, 25 Nov 2005 00:18:32 -0000 (GMT)

```Jorge Morgenstern replied back to Rodrigo Alfaro:

> I am trying tu use a daily sample, with dummy variables both for country
> (30) , for year (10) and for country*year (300, as to take into account
> possible fundamentals that differ across countries and across times).
> I think that what I get using your suggestion is only the 40 dummies, for
> years and countries separated.

> The areg command with i.year*i.country dummies results in different
> coefficients than the xtreg command with i.year and fixed effects.

Assuming you're happy with Rodrigo's solution, I'd concentrate on paring
down your 30 fixed effects into two or three meaningful, as opposed to
atheoretical, variables. First, because it's far too many. Second, as any
experienced analyst of country-level panel data will tell you (not
including myself), fixed effects that merely represent names of countries
tell you next to nothing about why they explain any variation in your
dependent variable of interest.

If I'm fitting a model of inward investment (II) flows over the last 50
years, what is it about France, say, that makes them - inventing a figure
- significantly likely to attract 30% more II than, say, the Ukraine? Is
it because they've been a democracy for longer? Is it because they have
much more stable economic regime? If these reasons fit the bill, then you
should be devising variables that capture these more general factors
rather than just chucking a large bunch of FEs into the model.

The problem, of course, is that not all these general, theoretical
variables are directly measurable (how do you measure stablility?
Calculate the standard deviation of a country's GDP over 50 years? Would
you have the data to do this, anyway?). But you'll end up with a much more
parsimonius, fitter pooled regression model in the process.

Have fun. :)

CLIVE NICHOLAS        |t: 0(044)7903 397793
Politics              |e: clive.nicholas@ncl.ac.uk
Newcastle University  |http://www.ncl.ac.uk/geps

Whereever you go and whatever you do, just remember this. No matter how
many like you, admire you, love you or adore you, the number of people
turning up to your funeral will be largely determined by local weather
conditions.

*
*   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/
```

 © Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index