Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.

# Re: st: margins, eyex

 From Antoine Terracol To statalist@hsphsun2.harvard.edu Subject Re: st: margins, eyex Date Sun, 23 Sep 2012 22:06:47 +0200

That's because when you regress ln(y) on ln(x), you assume a constant elasticity, i.e. a linear relationship between ln(x) and ln(y); estimated by the coefficient of ln(x). When you regress y on x, you assume a linear relationship between x and y. The estimated relationship between ln(x) and ln(y) will not be linear in that case; and the elasticity will thus vary according to the value of x.
```

```
tl;dr: -reg y x- and -reg ln(y ln(x)- do not make the same assumptions on the relationship between y and x
```
Antoine

On 23/09/12 20:59, albert Lee wrote:
```
```Hi, I ran a simple OLS and invoked the margins, eyex(*) to calculate
the elasticities.  I try to replicate the elasticity estimates by
running the sample regression with logged dependent variable, and
independent variables.  I was surprised that the coefficient estimates
from the logged regression are not identical with the margins
estimates.  Initially, I thought the discrepancies are due to atmeans
and asobserved options with margins.  But I tried both options and
still cannot replicate elasticity estimates by the logged regression.
Could someone explain?

Thanks,

Albert.

```
```

--
Ce message a ete verifie par MailScanner
pour des virus ou des polluriels et rien de
suspect n'a ete trouve.

*
*   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/
```