Hi all,
I would like to run a diff-in-diff in order to study the effects of acquisitions on some firms performance. I have data on efficiency (dependent variable) and other control variables for both the acquired and the acquirer before and after the acquisition.
My question would be: my treatment group would get the data of the acquirer for periods after the merger, but what about the periods before? If I use again those of the acquirer, then the efficiency gains/losses after the acquisitions would be driven mainly by whether the acquired is a good/bad firm in terms of performance, thus I would not catch the effects of merging itself on performance.
I was thinking to consider as "pre-acquisition" (for the treated group) a weighted average of all the indicators between the acquirer and the acquired (weighted by some size variable like total assets), but it sounds a bit rudimentary.
Another alternative would be to control in the regression for data of the acquired, but I am afraid about problems in terms of correlation with other control variables, etc.
Any idea?
Thank you in advance for your answers,
Matteo
I would like to run a diff-in-diff in order to study the effects of acquisitions on some firms performance. I have data on efficiency (dependent variable) and other control variables for both the acquired and the acquirer before and after the acquisition.
My question would be: my treatment group would get the data of the acquirer for periods after the merger, but what about the periods before? If I use again those of the acquirer, then the efficiency gains/losses after the acquisitions would be driven mainly by whether the acquired is a good/bad firm in terms of performance, thus I would not catch the effects of merging itself on performance.
I was thinking to consider as "pre-acquisition" (for the treated group) a weighted average of all the indicators between the acquirer and the acquired (weighted by some size variable like total assets), but it sounds a bit rudimentary.
Another alternative would be to control in the regression for data of the acquired, but I am afraid about problems in terms of correlation with other control variables, etc.
Any idea?
Thank you in advance for your answers,
Matteo
Comment