Dataset Construction and Event Study Methodology Questions

[Guest]

Hello Stata Community,

i face 2 problems regarding some research I carry out at the moment for which I am not educated enough to validate my suggested solution.

1. Lets assume I want to analyze for example the stock price of n=10 firms over a certain time period and I now want to control for the influence of a market index in a cross section regression. How do I construct the Dataset since I have 10 daily values for each of the 10 firms but only 1 value for each day for the stock index. Is the solution to copy the stock value 10 times to create a panel structure or would this be completely wrong ?

2. Lets say I want to analyze CARs of different firms in a cross sectional regression and lets say the cummulation period is t=3 days. If I now want to include controls these controls also need to be somehow prepraed in order to fit to the CAR values. So my suggestion would be to take the 3 day average of these controls and inculde them then as controls. Is this correct ?

Thank you in advance! I tried to find an answer to these questions in related books but since those questions are maybe too trivial I could not find any help there. Sorry in advance if a smililar question was asked before in another thread I overlooked.

[George Ford]

my thoughts. 1. yes, in either wide or long format, depending on which is best for the model you aim to estimate. With long, you can get a single set of AR for the entire group. Can do same wide, but with cnsreg. 2. i don't think so. the coefficients on event dummies already account for the controls. (the dummy coefficients are the abnormal returns).

[Guest]

Hello Mr. Ford thank you very much for your answer and your thoughts. I think I have to specify 2. a bit more. Maybe I do not fully get your point but Iam not planning to use dummies for all events in a one step regression. I plan to first estimate the ARs with a market model and additionally compute CARs which I want to analyze in a cross sectional regression as it is done in various studies.