Panel Data for Mergers & Acquisitions

[Mustabsar Awais]

I have run event study of more than 2000 M&A deals and collected CARs. Now I would like to regress the CARs with social media data. However, I have firm level data of social media of 6 years. On the other hand there are instances where single has more than one CAR due to several M&A deals. How can I develop Panel data? How should I arrange the data? Please guide. Thanks

[Attaullah Shah]

In this case, your panel identifier is not firm, rather event. Create a separate identifier for each event in both the data sets and then merge the two data sets using event id as the key variable.

[Mustabsar Awais]

Thanks for your response. But what if in case a single company has multiple deals in same year? How can we deal with this?

[Carlo Lazzaro]

Mustabsar:
you can consider collapsing your dataset and then run panel data analysis on the collapsed dataset, like in the following toy-example, where one firm has multiple deals during a 2-year period:

set obs 4
g firm=1
g deal=_n
g year=1 in 1/2
replace year=2 if year==.
collapse (sum) deal, by(firm year)

[Marcos Almeida]

I wonder whether a - mixed - model wouldn't apply to your needs.

[Attaullah Shah]

Thanks for your response. But what if in case a single company has multiple deals in same year? How can we deal with this?

This is exactly what I pointed out in post #2. If a company has multiple events, each event will have a separate ID. So, let us assume that ABOTT company has three events in year 2007, then each event will carry a unique identifier such as 1, 2, 3, and hence there will be no duplicates. Then assume that OGDC company has two events in the same year, we shall continue with the sequential numbering of events, so the two events of OGDC will get numerical identifiers of 4 and 5.

I do not think that sums or averages of the events will capture the dynamics of each event. Usually, in event methodology, we investigate each event separately and see its dynamics. Averaging or summing will result in losing important information.

[Mustabsar Awais]

Thank you very much for your help. It is sorted now.