Problem with panel data

[Mike Lewis]

I am trying to study the impact of M&A in a firm's ROE one year after the deal has happened and trying to understand if this impact differs with 4 aspects:

• if it is a cross-border vs domestic deal
• if it is a growth or scale deal
• if it varies with the size of the target
• if it varies with the acquirers acquisition experience

I am having trouble with running the regressions in STATA as I fail to understand how to mark the acquisition in time. Right now I have a dummy with acquisition marked as 1 in years where the firm acquires and 0 otherwise but then when I want to mark domestic deals I have another dummy with 1 in years where the firm acquires domestically, when I want to mark cross.border deals I have another dummy with 1 in years where the firm acquires cross border and with 1 in both if in the same year it acquires both. The same goes for growth and scale deals. How else can I mark the acquisition in time without doing it multiple times for the same acquisition and creating multicollinearity ?

[Phil Bromiley]

You don't seem to have received an answer. You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.
Probably the easiest way is to use factor variable notation (look it up in the subject index or User Guide).
Suppose you have a dummy for a deal and another for a cross border deal. Then, the effect on the dummy for the deal is the effect of a domestic deal and the sum of the parameters on the two dummies is the effect for a cross border deal. If you're getting a colinearity problem (with dummies it will appear as no estimate), then you may have categories that have no observations.