Interesting topic! Seem I think quite hard to find some statistical evidence for your topic. Am I correct?

Yes, as I have trouble figuring out which analysis to use I also wouldn't know how to find the evidence. Would hypothesis testing with for example a T-test help in this situation?

Yes, T-tests can help you.

The market model is the most common analysis used for an event study. This methodology looks at the actual returns of a baseline reference market and tracks the correlation of a company's stock with the baseline.

Yes I already used the market model and I have the CAR for all 4 event windows, however how do I test whether it is statistically significant. I have learned that I should probably regress it against a constant. Do you have a suggestion to what this constant should be?

I do not see how you can tell based on short time frame. GE failed miserably, but the failure wasn't identified until a decade after turning a well-run company into a financial leveraged zombie company that failed due to incompetent business decisions made a decade earlier.

WorldCom same.

Enron Same

Musk buying Twitter is a way to take control of speech away from far left extremists and give it back to a population - will it get as much money? Not from money launderers and far left extremist groups.

How do you measure failed?

There are aspects such as merging cultures, systems, and accounting methods. There are political power battles of executives and they could care less about shareholders ( look at Twitter Board - they could care less until their names came public - thank you Elon Musk for exposing the nature of Twitter's board...)

Also, economic dynamics change. Currently, the desperate Central Banks, especially the federal Reserve, lost in debt and fake, fiat currency backed by nothing and trusted by nobody, is quickly failing. Their desperate measures are too little too late, and they are being replaced by the Quantum Financial System.

So, I would suggest lengthening your examination time period, looking at company performance relative to benchmarks pre and post event, and I would suggest that some variables may appear to be correlated but may be random events, and not directly correlated.

At the point of determining the level of significance, a pair-wise t-test or Analysis of variance (ANOVA) would solve that problem, after establishing break-points, to determine pre- and post- events.

Hi Jip..

An event Study is basically a study that tries to test whether an event either micro (the company level) or macro (macro level) is anticipated by market players making them decide to buy, hold, or sell the securities. A working paper by Kothari and Warner (2006) clearly elaborates the methodology of doing an event study. you might log in to: https://www.bu.edu/econ/files/2011/01/KothariWarner2.pdf (or in B. Espen Eckbo (ed.), Handbook of Corporate Finance: Empirical Corporate Finance, Volume A (Handbooks in Finance Series, Elsevier/North-Holland) chapter 6)

In terms of stock market reactions to an event, we usually test for the existence of abnormal returns (pre-event, event date, and post-event) and whether there are differences in abnormal returns between before and after the event.

in the case of testing the existence of abnormal return, a parametric test is employed when the data is normally distributed (meeting normal distribution) or a non-parametric test for non-normal distribution.

You may also read:

John J. Binder. 1985. On the Use of the Multivariate Regression Model in Event Studies.Journal of Accounting Research. Vol. 23, No. 1 pp. 370-383 to get some issues on the use of regression in the event study research.

Hope it helps.

Hi Jip..

It depends on your purpose of study. If you want to test the market response, you need only a test of the existence of Abnormal returns and the differences (maybe 5 days before the event date, the event date, and 5 days after the event date, depending on the your assumption of market reaction) of abnormal return. You can use the t test for means and t test for pair samples (when the data met the normal distribution assumption) or Wilcoxon one sample and pair samples when the assumption of normality is not satisfied.

Use T-test.

@Kenneth you measure the abnormal returns at the date the takeover was canceled, some takeovers never succeed as in they do not happen, not as in they do not work after completion, that's a different, but also an interesting topic.

Thank you for all your help, I have eventually gone with a one-sample T-test to check whether the means were significantly different from zero and used further hypothesis testing for my moderating variables. I do have a problem with measuring endogeneity, is there a "best practice" model for measuring endogeneity in event studies?