Page 1 of 2
elasticity at the mean
Posted: Mon Mar 21, 2011 9:47 pm
by bparksb
Eviews 7
View Coefficient Diagnostics Scaled Coefficients
reports Elasticities at the mean - the help file states
The elasticity at the means are the point estimates of the coefficients scaled by the mean of the dependent variable divided by the mean of the regressor.
Yes, that is what is calculated but that is WRONG. Elasticity is calculated as slope*X/Y -
e.g., change in Y/Y divided by change in X/X
or view
http://www.stata.com/support/faqs/stat/mfx_eyex.html
The standardized coefficients are correct.
Bob Parks
Re: elasticity at the mean
Posted: Tue Mar 22, 2011 8:11 am
by EViews Gareth
EViews calculates the elasticity at means as Coefficient_on_X * Mean(X)/Mean(Y), which is what you state is correct. The documentation might be a little unclear, but it says that it is calculated as the point estimates scaled by the mean of Y divided by the mean of X, where "scaled" is being used for "divided by".
Re: elasticity at the mean
Posted: Wed Mar 23, 2011 6:16 am
by bparksb
Sorry, I constructed an example and alas too late at night because today I see that you are right.
Bob
Re: elasticity at the mean
Posted: Mon Jul 22, 2013 5:28 pm
by zeca
Hi there,
I need to obtain the coeffients elasticity but the option Scaled Coefficients doesn't exist in View Coefficient Diagnostics.
I have the eviews8
Could you give me some help on this?
thanks!
Re: elasticity at the mean
Posted: Mon Jul 22, 2013 6:46 pm
by EViews Gareth
What sort of equation do you have?
Re: elasticity at the mean
Posted: Tue Jul 23, 2013 3:47 am
by zeca
What sort of equation do you have?
Panel Least Squares
Re: elasticity at the mean
Posted: Tue Jul 23, 2013 6:22 am
by EViews Gareth
It isn't offered for panel equations.
Re: elasticity at the mean
Posted: Tue Jul 23, 2013 7:51 am
by zeca
It isn't offered for panel equations.
ok, and there is any other way to calculate it for panel data equations in eviews?
thanks
Re: elasticity at the mean
Posted: Tue Jul 23, 2013 7:59 am
by EViews Gareth
I'm sure it can be calculated manually, if you can figure out what the calculation is.
Re: elasticity at the mean
Posted: Tue Jul 23, 2013 8:13 am
by zeca
I'm sure it can be calculated manually, if you can figure out what the calculation is.
Are you thinking in the following calculation?
If equation Y=a+bX1 + cX2
Elasticity X1 = Coefficient_on_X1 * Mean(X1)/Mean(Y)
Elasticity X2 = Coefficient_on_X2 * Mean(X2)/Mean(Y), ...
Re: elasticity at the mean
Posted: Tue Jul 23, 2013 8:17 am
by EViews Gareth
I suppose the question is, what is the mean in a panel?
Re: elasticity at the mean
Posted: Tue Jul 23, 2013 9:00 am
by zeca
I suppose the question is, what is the mean in a panel?
as you probably gathered so far, my econometric knowledge is not so great, could you be more specific please?
Many thnks, we've been real helpful to me!
Re: elasticity at the mean
Posted: Tue Jul 23, 2013 9:36 am
by EViews Gareth
I don't know what the calculation of elasticity at means is in a panel setting. If you can figure that out, you can probably calculate it easily enough in EViews.
Re: elasticity at the mean
Posted: Tue Jul 23, 2013 11:57 am
by zeca
I don't know what the calculation of elasticity at means is in a panel setting. If you can figure that out, you can probably calculate it easily enough in EViews.
me neither :( I tried searching for it in the web but I couldn't find anything. The papers I read do not explain the calculation, just present the results....
If someone else have any idea, please help...
Re: elasticity at the mean
Posted: Sun Sep 22, 2013 4:47 pm
by bparksb
The elasticity at the mean is exactly
If equation Y=a+bX1 + cX2 + crossection effects + time effects
Elasticity X1 = Coefficient_on_X1 * Mean(X1)/Mean(Y)
Elasticity X2 = Coefficient_on_X2 * Mean(X2)/Mean(Y), ...
So long as the coefficients do not change across time or cross section, the elasticity at each point of the data would be the series
Elasticity X1 = Coefficient_on_X1 *X1/Y
If the coefficients differ over cross sections, then it is the coefficient for each cross section times the ratio of means for that cross section.
Alas, I don't have a reference and I am too lazy to write it out long hand.