Interaction Term for Fixed effects Model with Time Invariant dummy variable

[Reuben Vaughan]

Hello all

Investigating the influence of Foreign directors on firm performance I am utilising a fixed effects model. Having found an aggregate effect for all directors across all firms around the UK I want to focus on those in London, and see if they improve firm performance at a greater level.

As expected London location is time invariant, however a dummy variable for foreign director presence (FDir) is not.

My question, how should I specify this model with an interaction term? and how should I interpret the coefficient

Model 1: Firm Performance = FDir*London + Controls
And therefore compare the coefficient to a model without the interaction term London?

or

Model 2: Firm Performance = FDir*London + FDir + Controls
And therefore if Fdir*London is positive and significant conclude that Foreign directors in london firms improve firm performance greater than elsewhere?

Appreciate your guidance,

Reuben

[Clyde Schechter]

I recommend using Model 2 because the interpretation is simpler. You answer your question just by reading the interaction term line in the regression table.