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

st: Simple Panel Data Model

From   "Sukhdev Kumar" <[email protected]>
To   <[email protected]>
Subject   st: Simple Panel Data Model
Date   Thu, 1 Apr 2004 12:25:14 +0100


I'm pretty new to Stata and am having to use it for my dissertation, so I
would really appreciate anyones help!
I'm doing a simple panel data model of the impact of aids on economic
growth, and using the simple model recommended in a journal I found, am

gdp growth = constant + aidsdeaths + productivity growth + population growth
+ literacy + error term.

I have balanced panel data for 78 countries for two time periods (1999 &

I've tsset my data and ran xtreg....fe and (along with the
hausman test) for the overall model and for subsections based on gdp per
capita. Could anyone help me with the following questions.

1) Are there any alternative regressions I can run for better results?
(according to the probability statistic, my results are generally really
poor, even compared to the journal which used data very similar to mine).

2) Its been recommended that I use lags of my variables as instruments in
the regression (to prevent feedback from growth to the right hand side
variables), but is this possible seeing as my data is for only 2 separated
time period, 1999 & 2001? If so, how would I do this?

3) Although I have used the Hausman test to see if fixed effects or random
effects GLS regression is preferred, can I also test for autocorrelation,
heterogeneity and heteroskedasticity?

4) How do I test for non-linear determinants of the model?

5) If I wanted to run a regression and leave a couple of the countries (i)
out, how would I write this?

6) According to the journal I am basing my methodology on (Edwards &
Al-Hmoud), they ran the regressions as simple OLS Huber/White Robust
variance procedures - is this possible, seeing as it is panel data, and if
so, how would I do this?

I'm really sorry for the number of questions, but anyones help would be
truly appreciated! I am used to working with Microfit on time series data,
so this is totally new to me! I've looked at Stata's FAQ's, the cross
sectional Stata guidebook and other sources, but can't figure out the above

Apologies fo any inconvenience.

Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (
Version: 6.0.637 / Virus Database: 408 - Release Date: 20/03/2004

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index