For a regression model: Y = B0 + B1.X + B2.X2 + U
, B1
and B2
is the marginal effect on Y
wrt X
and X2
.
Now, for a log - log regression model, log(𝑌) = 𝐵0+ 𝐵1.log(𝑋) + 𝐵2.log(𝑋2) + 𝑈
, B1
and B2
is the elasticity Y wrt X
and X2
.
In my case, I'd like to estimate elasticity Y
wrt X
, when X2
is not constant. i.e cross elasticity effect.
In my case, Y
is units sold and X
is price and X2
is available substitutes. The standard interpretation doesn't apply here because there is cross-elasticity effect of X2
, number of available substitutes on Y
.
How would I interpret elasticities and cross-elasticities in this case?
ps. I realize price is endogenous and need to add an instrument to deal with problem.