How to create a matrix with the weighted average of a variable?

[Henri Bussink]

Hi all,

For my thesis I want to explore the impact of a weighted average of the corporate tax rates abroad on the corporate tax base in country i. So my dependent variable is a proxy of the tax base of country i and my explanatory variables are the domestic tax rate, a weighted average of the tax rates abroad, and some controls. Since the tax rate variables are endogenous and I include the lagged dependent variable I employ system GMM. My data is an unbalanced panel of 45 countries covering most of the period 1988-2016.

Now I want to weigh the tax rates abroad in several ways. One way is to weigh the tax rate in each country j by its GDP as a share of total GDP in all countries other than i. More formally, the weight will look something like the following:

Anyone who has an idea how to construct such a weight with my data in Stata/IC 15? Which commands to use? Do I have to construct a weighting matrix?

[Jorrit Gosens]

webuse grunfeld, clear
ren company country
ren invest GDP
ren mvalue taxrate
drop kstock time

bysort year (country): egen globalgdp=total(GDP)
gen globalgdp_exdomestic=globalgdp-GDP
gen weight=GDP/globalgdp_exdomestic

Edit: I think this is what you want. I dont understand entirely the mechanics of your weights here.