test of difference in coefficients

[Helen Chang]

Hi, I have two subsample groups: profit firm group (1) and loss firm group (2).
I want to have the p-value regarding the test of difference in coefficients (1) - (2), so that I can generate the table below.

Variable---------------Profit firms (1) ------------Loss firms (2) ---------------------p-value: test of difference in coefficients (1)-(2)

Var1
Var2
Var3
Firm cluster included
Year fixed effect included

[Jeff Wooldridge]

Best is to pool the data, choose one group as the base group, and then use a dummy for the second group interacted with Var1, Var2, and so on. Test the coefficients on the interactions.

[Phil Bromiley]

The alternative is to run them separately and then use suest or you could use sureg. These may give slightly different results than the interaction approach - the interaction approach forces the same parameters between groups on any variables you don't explicitly interact with the group variable, and also does one error variance estimate instead of two.