Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Fixed-effects and Random-effects Regressions in STATA


From   "Sophal Ear" <sophal@alumni.princeton.edu>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: Fixed-effects and Random-effects Regressions in STATA
Date   Mon, 10 May 2004 19:12:10 +0700

Hello,

I'm trying to run some regressions using pooled data under fixed-effects
and random-effects in STATA but am running into some difficulty while
based here in Cambodia.

I have four years' worth of data (1996, 1998, 2000, and 2002) and 167
countries. The data is now pooled creating a 668 by however many
variables I have matrix (among them ratings for quality of governance,
Aid as % of GDP, per capita GDP, human development indicators, etc.). 

Among these variables are also dummy years:

When data for 1996 is available, the dummy for 1996 says 1, 0 otherwise.
When data for 1998 is available, the dummy for 1998 says 1, 0 otherwise.
When data for 2000 is available, the dummy for 2000 says 1, 0 otherwise.
When data for 2002 is available, the dummy for 2002 says 1, 0 otherwise.

I was then instructed to run OLS (which is to say the regress command in
STATA) and include the dummy years (dropping one, otherwise it would be
perfectly collinear) to get a fixed effects model (I guess based on
year?)... and report my results on the explanatory variables'
coefficients which interest me. 

However, this does not jive with the instructions in STATA regarding
fixed and random effects models, which mention only the use of xtreg
(and falls under realm of Generalized Least Squares)... and so I ran
those as well, but at this point I am not even sure if I should be
including among my regressors the dummy years per se... 

I've also tried Panel Corrected Standard Errors (PCSEs) regression but
STATA reports too few observations (also, I'm not quite sure if this is
right, but I created a unique numeric ID for each country and set that
as the panel variable, while using year as the time variable) and got
the following dead-end:

Number of gaps in sample:  331
(note: computations for rho restarted at each gap)
insufficient observations
r(2001);

Any advice would be much appreciated. Cambodia has few if any
statisticians/econometricians around, much less STATA experts so this
listserv is a lifesaver.

Many thanks,

Sophal Ear

PhD Candidate
UC Berkeley

---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.639 / Virus Database: 408 - Release Date: 3/22/2004
 

*
*   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 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index