Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: how to access e(b) and e(V)


From   Maarten buis <maartenbuis@yahoo.co.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: how to access e(b) and e(V)
Date   Tue, 20 Oct 2009 04:36:59 -0700 (PDT)

--- On Tue, 20/10/09, Philomela@gmx.net <Philomela@gmx.net> wrote:
> Actually i would like to come up with a linear demand
> function like that:
> Consumption= constant+ beta_1*Price+beta_2*Income+ ...
> and a loglog demand function.
> I used the spreadheet programm and manually calculated
> Consumption for various prices and got the two demand
> functions from that, but unfortunatly, due to the log
> transformation i have the following problem:
> exp(E(ln(Consumption))) is not equal to E(Consumption).
> I thought more precise estimates would help-but as it
> turned out now- they don't.
> So, can I  estimate the consumption for different
> prices directly in Stata to get the demand functions?
> Then, I should be able to deal with the retransformation
> bias  as suggested in
> Cameron&Trivedi(2009):Microeconometrics using Stata.

You have two options:
1) Use -glm- with a log link function, and after that
you can just use -predict- to get the predicted values.
So you would type something like:

glm consumption price income ... , link(log)
predict consumptionhat, mu

2) Alternatively the method proposed by Cameron & 
Trivendi is implemented in Stata by Kit Baum as
the -levpredict- package. You can install that
by typing in Stata: -ssc install levpredict-.

IMHO the -glm- solution is more elegant as it tackles
the problem at its root: In this case -glm- models 
log(E(consumption)) while -regress- models 
E(log(consumption)). Because of this the exponentiating
the left hand side and right hand side after a -glm-
model (which happens when you use -predict-, -adjust-,
or specify the -eform- option) will lead to results
in the original unit of consumption, where this is not
true after -regress-. The solution offered by Cameron
& Trivendi seems to me a bit of a post hoc solution.

-- Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------

Send instant messages to your online friends http://uk.messenger.yahoo.com 

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   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