st: Re: log transformations of dependent and independent variables

[Kit Baum <baum@bc.edu>]

This kind of specification (log Y vs log X) is a constant elasticity  
model (see mfx, eyex) in which the coefficients represent the effect  
of a percent change in X in terms of percent changes of Y. If you  
leave some variables untransformed (including dummies) then those  
coefficients are semi-elasticities (d ln Y / dX) in which you examine  
the effect of a one unit change in X in terms of percent changes of Y  
(see mfx, eydx). You may have reason to believe that a unit change in  
some (non-dummy) Xs has a proportional effect on Y. An obvious  
example is a log trend model, ln Y = a + b t; each unit of time  
changes Y by approximately b percent (rather than b units of Y, as  
would be the case without logs).

Kit Baum, Boston College Economics
http://ideas.repec.org/e/pba1.html


On Sep 7, 2005, at 2:33 AM, Chriswrote:


> yes, this is a common thing to do.  wage regressions are a good  
> example.=20
> economists usually take the log of the dependent variable (wages).   
> and=20
> then some regressors (such as education) typically enter the  
> equation=20
> linearly -- including some dummy variables like race or sex  
> dummies.=20
> while other regressors (such as parents' income) typically enter in  
> log=20
> form.
>
> chris
*
*   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/