eventstudy2 missing cumulative abnormal returns

[Patrick Mauder]

Hi,

This might be tricky to work out without the data but I will try to describe it.

I am using stata 14.2 and need cumulative abnormal returns which I got using eventstudy2. I am trying to reconcile the number of observations (i.e. cumulative abnormal returns or CAR) I have which is 18. the observation window covers the month of June 2016 which has 30 days and had 9 non-trading days, I thought then that I need to deduct 2 days since my event window is (-1,0,1). When I look at the cumulative abnormal returns I get, there are a few things I understand and again some I don't. As expected I do not have CAR for the non-trading days, makes sense. However the odd thing is I have a CAR for the 1st of June and in total I have 18 observations of CAR. So from 30 days in June, I take 9 non-trading days, giving me 21. I don't have CAR for the 30th of June which I thought makes sense due to the event window being -1,0,1 but then oddly I have two days in the middle (10th June and 15th June) which have no CARs and I struggle to understand why.

Here is what I entered:

eventstudy2 security_id date using security_returns_file, evwlb(-1) evwub(1) eswlb(-250) returns(simple_return) model(MA) marketfile(marketfile) mar(market) idmar(marketref) risk(rf)

The dates in the security_returns_file go from 01st June 2015 to 30th June 2016.
Simple return is closing minus opening for each day for that period and the market file data contains the fields for each day

Do you have any idea why I might be missing the two dates but the 1st June is included (i.e. why i get 18 CAR)?

Thanks,

Patrick

[Clyde Schechter]

This might be tricky to work out without the data but I will try to describe it.

Quite so, an understatement even. Why make life harder for those who might help you? You're more likely to get a prompt and helpful response if you show an example of your data. (If the data are confidential and you can't disclose them, just jitter the numbers a bit before posting, or replace identifying variables with meaningless pseudo-id numbers.)

While we're on the topic of posting questions effectively, -eventstudy2- is not a part of official Stata. The FAQ asks that you state where it comes from when posting a question about it.

[Patrick Mauder]

Hi Clyde,

I thought that might be tricky, but since there are so many files involved I wasn't sure how to handle that. I could copy in sections of each file to show what they include or attach the actual files?

As far as I can tell eventstudy2 is not an official command but contributed by Thomas Kaspereit.

[Clyde Schechter]

I would post an excerpt of the data consisting of a set of observations that exhibits the problem you are having, not entire data sets. If I understand #1, it would be a month of data on one firm, the month and firm selected because the number of days of cumulative abnormal returns you get is different from what you expect.

[Patrick Mauder]

Here is the command lines and an extract from each file:

cd "$MY_PATH"
pwd
use eventfile, clear
eventstudy2 security_id date using security_returns_file, evwlb(-1) evwub(1) eswlb(-250) returns(simple_return) model(MA) marketfile(marketfile) mar(market) idmar(marketref) risk(rf)

Here are extracts from the files the command refers to:

eventfile

date security_id marketref id 1-Jun-16 AHG 1 AHG 2-Jun-16 AHG 1 AHG 3-Jun-16 AHG 1 AHG 6-Jun-16 AHG 1 AHG 7-Jun-16 AHG 1 AHG 8-Jun-16 AHG 1 AHG 9-Jun-16 AHG 1 AHG 10-Jun-16 AHG 1 AHG 14-Jun-16 AHG 1 AHG 15-Jun-16 AHG 1 AHG 16-Jun-16 AHG 1 AHG 17-Jun-16 AHG 1 AHG 20-Jun-16 AHG 1 AHG 21-Jun-16 AHG 1 AHG 22-Jun-16 AHG 1 AHG 23-Jun-16 AHG 1 AHG 24-Jun-16 AHG 1 AHG 27-Jun-16 AHG 1 AHG 28-Jun-16 AHG 1 AHG 29-Jun-16 AHG 1 AHG 30-Jun-16 AHG 1 AHG

security_returns_file (example of one company (I have 74 in total)

security_id open high low close date log_return marketref AAC 1.455 1.47 1.45 1.47 2-Jan-15 0.010256 1 ….. AHG 3.85 3.905 3.85 3.88 1-Jun-16 0.007762 1 AHG 3.86 3.89 3.8 3.85 2-Jun-16 -0.00259 1 AHG 3.89 3.91 3.84 3.86 3-Jun-16 -0.00774 1 AHG 3.83 3.85 3.8 3.82 6-Jun-16 -0.00261 1 AHG 3.83 3.86 3.76 3.78 7-Jun-16 -0.01314 1 AHG 3.75 3.79 3.71 3.77 8-Jun-16 0.005319 1 AHG 3.78 3.81 3.76 3.79 9-Jun-16 0.002642 1 AHG 3.77 3.77 3.71 3.73 10-Jun-16 -0.01067 1 AHG 3.68 3.7 3.64 3.68 14-Jun-16 0 1 AHG 3.68 3.7 3.65 3.67 15-Jun-16 -0.00272 1 AHG 3.69 3.72 3.68 3.71 16-Jun-16 0.005406 1 AHG 3.73 3.75 3.705 3.74 17-Jun-16 0.002677 1 AHG 3.75 3.8 3.75 3.79 20-Jun-16 0.01061 1 AHG 3.74 3.82 3.74 3.79 21-Jun-16 0.01328 1 AHG 3.8 3.835 3.76 3.81 22-Jun-16 0.002628 1 AHG 3.82 3.83 3.78 3.8 23-Jun-16 -0.00525 1 AHG 3.84 3.84 3.69 3.7 24-Jun-16 -0.03714 1 AHG 3.69 3.755 3.66 3.75 27-Jun-16 0.016129 1 AHG 3.7 3.71 3.64 3.71 28-Jun-16 0.002699 1 AHG 3.74 3.78 3.72 3.73 29-Jun-16 -0.00268 1 AHG 3.76 3.8 3.73 3.76 30-Jun-16 0 1

marketfile

date marketref market smb hml rf 2-Jul-90 1 0.41 -0.06 0.54 0.03 …….. 1-Jun-16 1 -0.37 0.38 0.14 0 2-Jun-16 1 -0.29 0.23 0.22 0 3-Jun-16 1 1.59 -0.38 -0.3 0 6-Jun-16 1 0.51 -0.45 0.38 0 7-Jun-16 1 1.1 -0.21 0.26 0 8-Jun-16 1 0.05 0.42 0.12 0 9-Jun-16 1 -0.37 0.55 0.43 0 10-Jun-16 1 -1.15 0.36 0.16 0 13-Jun-16 1 -0.92 0.34 -0.45 0 14-Jun-16 1 -1.37 0.42 0.65 0 15-Jun-16 1 -0.09 0.4 0.07 0 16-Jun-16 1 -0.78 0.46 -0.19 0 17-Jun-16 1 0.43 -0.1 -0.43 0 20-Jun-16 1 1.8 -1.2 -0.01 0 21-Jun-16 1 0.18 0.21 -0.31 0 22-Jun-16 1 0.4 -0.4 -0.26 0 23-Jun-16 1 0.9 -0.24 -0.19 0 24-Jun-16 1 -3.51 0.47 1.18 0 27-Jun-16 1 -0.91 -0.08 0.48 0 28-Jun-16 1 0.27 0.27 -0.22 0 29-Jun-16 1 1.56 -0.27 -0.31 0 30-Jun-16 1 1.71 -1.03 -0.04 0

I also have an events file but I am not sure at which point that was used: