Including both main variables and interaction terms when estimating ad valorem equivalents

[Page Alarcon]

Hello, I am calculating country-specific ad valorem equivalents for non-tariff measures (trade policies such as quotas that are not tariffs but affect trade, etc), and the specification used by the literature is the count of these measures and an interaction term consisting of this measures count and the world share of a country in that product. None of the literature includes the world share as a main variable in the regression, but I know that main variables should always generally be included to avoid misspecification, unless economic theory says otherwise [sic]. When I include world shares, it is not significant in my regression and the coefficients do not change by too much.

What could be the reason why the literature does not include the world shares as a main variable?

[Carlo Lazzaro]

Page:
I cannot answer to your question, but I would sponsor your approach.
The lack of statistical significance is a possible state of the world, not (in general) a model shortcomings.
If, as per PAQ, you posted what you typed and what Stata gave you back, more positive replies could possibly come alive.