eventstudy 2 by Thomas Kaspereit - Implementation and conceptual question

[Philipp Auinger]

Dear all,

I'm working on my Master's Thesis and came across eventstudy2, an excellent event study tool.

However, as I'm fairly new to Stata, I have two questions concerning the command.

Questions:

1. When I tried to modify the data sets provided in the samples to use them further, I came across the variable "earnings_surprise". Unfortunately, I'm not sure which economic value this variable represents. Is this the return observed on an event date ("Date") that matches the "Security ID"?
2. I thought about applying the event study methodology to analyze the influence of multiple M&A announcements on the stock price of one single company. Is it possible to employ eventstudy2 in this case, or is it limited to multiple events concerning multiple companies?

I'm looking forward to any helpful advice since I haven't found an answer on Statalist.

Best regards,

Philipp

[Thomas Kaspereit]

Dear Philipp,

1) The earnings_surprise variable captures how much the company surprised the capital markets with its earnings annoucement. It follows the formula :

SUE = (earnings_realized - earnings_expected) / earnings_expected

while earnings_expected is either the prior year's earnings or the consensus analyst forecast. You find tons of literature on Google scholar if you use search for the terms "earnings surprise", "earnings response coefficients". It is not related to any security returns. Those are kept in the security returns file.

Please keep in mind that my sample files are generic and you cannot re-identify the real firms. Thus, they are not usable for real research.

2) This is certainly possible to implement with eventstudy2. Your company will then appear times the number of its M+A annoucements in the eventlist file. Eventstudy2's unit of investigation is the event, not the company. Thus, each line in the eventfile represents an event (either of the same or of different companies). The company identifier only serves to find the right security data in the security file.

Please reference eventstudy2 in your thesis, if you decide to stick with it.

Best
Thomas

[Philipp Auinger]

Dear Thomas,

Thank you for the prompt reply and the detailed clarifications.
They were really helpful!

Is it also possible to implement long-term analysis through the event-time (BHAR) and calendar-time (Carhart Four-Factor Model) approach with eventstudy2?

I understood that it is possible to calculate both.

Best regards,

Philipp

[Thomas Kaspereit]

Dear Philipp,

BHAR, yes, calendar time, no. The latter follows a fundamentally different data management logic, which is why have not implemented it in eventstudy2.

Best Thomas

[Philipp Auinger]

Dear Thomas,

Thank you for the input!

Best,

Philipp