Dear all,
I read previous posts about event study, but I still cannot find my starting point for coding my event study model in Stata. I really appreciate it if someone could help me with recognizing which package of event study is most suitable for my methodology.
To study the relationship between stock portfolios and monetary policy surprises, my model will be the event study used by Haitsma et al. (2016) which is attached to the post.
the number of events (monetary policy announcements) is limited to 272 dates, and I want to regress this equation for the DAX index on a daily basis.
Thank you so much in advances, Shahrzad Tarrahi
Dear Shahrzad Tarrahi ,
You can code it manually, as I am doing right now with a paper. Using the commands for dealing with interactions embedded in Stata is enough.
The point is that instead of introducing treated*post for capturing the effect you allow for more flexibility, interacting treated with every time variable but the one one period before treatment kicks in. There are several ways of doing but the most common one is just to take as reference such a period and entering lags and leads. One would expect that the "treatment" before such a period are 0 and that after the treatment there is some effect at some point (increasing, decreasing, etc.).
Nevertheless, there are several new tools. This package can help
https://ideas.repec.org/c/boc/bocode/s458676.html
Check also the 1st working paper in the list:
http://goodman-bacon.com/workingpapers.html
I hope it helps.
Best,
José Ignacio
When we want to predict the value of a dependent variable based on the value of
an independent variable, linear regression is utilized, also known as simple linear regression or linear bivariate regression.
Dear José-Ignacio Antón ,
I tried to apply your recommended method, but I cannot because my data structure is not panel data.
the list of variables for linear regression in an event study model is:
1- "return of the DAX index" as my dependent variable;
2- "event dates or policy announcement dates" as the treatment variable which takes 1 when the event occurs;
3- "MSCI world index" and "crisis dummy" as the control variables;
4- "Conventional surprise pre-crisis", "Expected change pre-crisis", Conventional surprise crisis", "Expected change crisis" as my independent variables.
as I understand from the reference you introduced me, I should have different groups/factor variables which have category within the variable to be able to use DD method. but my variable is only the DAX index which has no category and it is a time series.
could you please guide me or correct me if I am wrong about it?
Many thanks!!!
You are right. Sorry, I did not notice you had just time-series data. I think you should go in that direction, exploring interrupted-time-series analysis and similar methods.
Maybe this resources can be of help:
Article Using time-series analysis to evaluate the impact of policy ...
https://gh.bmj.com/content/5/10/e003567
https://academic.oup.com/ije/article/46/1/348/2622842
https://www.edx.org/course/policy-analysis-using-interrupted-time-series
I am an applied microeconometrician. I think that in Finance and Macro you might find applications more tailed to the problem you have to deal with.
I hope it helps.
Best regards,
José-Ignacio
Dear José-Ignacio Antón ,
again thank you for your help and for your quick response. I think it would be the right one.
best regards,
Shahrzad
Stata statistical software has powerful features I think can help you meet your study objective
- Badges
- Science method
- Similar topics
- Methodology