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

Re: st: Model specification


From   Daniel Feenberg <feenberg@nber.org>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Model specification
Date   Tue, 11 May 2004 07:58:07 -0400 (EDT)

On Tue, 11 May 2004, joe J. wrote:

> Dear all,
> 
> I have a labour productivity equation which is in log-log form, i.e.,
> lnQ-lnL=LnK+....
> I want to add a dummy variable and a time trend (categorical variables) to

You will have trouble taking the log of zero, but even assuming you change
the dummy to be 1/2 instead of 0/1 it is unhelpfull to log the value. This
would be equivalent to have a dummy variable of log(1) or log(2), which
will cause the estimated coefficient on that variable to be different, but
won't change any t-statistic or coefficient on any other variable.

The time trend without logs estimates a constant percentage change in the
underlying dependent variable per time period. If you logged time, then
you would be estimating a relationship in which instead of a constant
growth rate, there was a growth rate constant in the log of time, which
seems pretty odd. After all, it would depend on when "time started".

> the basic model. My question is should these two variables be also used in
> log-form or not?
> 
> Thanks for any pointers.
> Joe
> 
> _________________________________________________________________
> Seized by wanderlust? Have the best vacation ever.
> http://www.msn.co.in/Travel/ Team up with MSN Travel!
> 
> *
> *   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/
> 


*
*   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