----- Original Message -----
From: "Botelho, Delane" <[email protected]>
> I ran "xtlogit" (random effect) and "mfx compute, eyex" to get the
> elasticities. Then I calculated the same elasticities using "predict var,
> pu0" and plugging in the formula of elasticities of logit (p. 111
Ben-Akiva
> & Lerman). Fortunately I got the same results, and I understood what STATA
> did.
Is this possible? When I try "mfx compute, eyex" after xtlogit, re I get "
eyex not available"
Example, from [R] xtlogit page 380
. use http://www.stata-press.com/data/r7/union.dta
(NLS Women 14-24 in 1968)
. qui xtlogit union age grade not south southXt, i(id) re
. mfx compute, eyex
y=__0004OV< 0, eyex not available
r(459);
I believe you would want to use mfx compute, predict(pu0) in this case.
. mfx compute, predict(pu0) nose
Marginal effects after xtlogit
y = Pr(union=1 assuming u_i=0) (predict, pu0)
= .08693276
----------------------------------------------------------------------------
---
variable | dy/dx X
---------------------------------+------------------------------------------
---
age | .0007334 30.4322
grade | .0066681 12.7615
not_smsa*| -.0195416 .283702
south*| -.0867433 .413015
southXt | .0018886 3.96874
----------------------------------------------------------------------------
---
(*) dy/dx is for discrete change of dummy variable from 0 to 1
> Then, I ran "regress" (another dataset, different variables), using
natural
> logs of quantity (depvar) and price (varlist). The coefficient of price is
> supposed to be the elasticity. But when I ran "regress" not using natural
> logs, and afterwards "mfx compute, eyex", those elasticities are not the
> same as the coefficients in the first regression.
If your model is already in double log form than the coefficients are the
marginal effects. Or if you want to use mfx compute, than it should be mfx
compute, dydx.
Alternatively, if you model was in levels than you would want to use mfx
compute , eyex.
Example
. use "C:\Stata\auto.dta", clear
(1978 Automobile Data)
. gen lnprice = ln(price)
. gen lnmpg = ln(mpg)
. reg lnprice lnmpg , nohead
----------------------------------------------------------------------------
--
lnprice | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
lnmpg | -.826847 .1484986 -5.57
0.000 -1.122873 -.5308204
_cons | 11.14146 .4507755 24.72 0.000 10.24286
12.04007
----------------------------------------------------------------------------
--
. reg price mpg , nohead
----------------------------------------------------------------------------
--
price | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
mpg | -238.8943 53.07669 -4.50
0.000 -344.7008 -133.0879
_cons | 11253.06 1170.813 9.61 0.000 8919.088
13587.03
----------------------------------------------------------------------------
--
. mfx compute, eyex nose
Elasticities after regress
y = Fitted values (predict)
= 6165.2568
----------------------------------------------------------------------------
---
variable | ey/ex X
---------------------------------+------------------------------------------
---
mpg | -.825238 21.2973
----------------------------------------------------------------------------
---
> 2. I also need to run Poisson regression and calculate elasticities. If I
> use "mfx compute, eyex" after "poisson", how (which formula) STATA will
> calculate the elasticity?
>
> Please, give at least some reference.
J. Scott Long and Jeremy Freese in "Regression Models for Categorical
Dependent Variables Using Stata" use 'mfx, compute' in their example of
calculating marginal change after poisson, pages 234 -235.
Scott Merryman
*
* 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/
