Hi there, brilliant video, I had one question regarding my study. I collected tweets regarding 5 firms to see if they have an impact on stock price, and would like to calculate abnormal returns for all the firms over the 2 year period I have. I want to make all my data consistent and was wondering how do i calculate the Alpha and Beta for the 5 firms i.e. I have an excel sheet and have 85 event days (which are not in chronological order) for the 5 firms, and would like to use the market-adjusted model. So how do I determine the estimation period to calculate the values for each firm?
So I work with 250 sheets, all with different event dates. Is it possible to make it calculate the alpha for each, with a flexible calculation? Something on the lines of AverageIF - Just with Intercept and Slope ofc.
Hello sir, thank you so much for your explanation video. I would ask you about paired sample t test to test the different before and after CAR. Is the CAAR Calculation for this test done by averaging the days in event window (average before and average after) or the company-wide average in each event window? And if I use the calculation CAAR for average the company wide it was hard to test the normality data for window -1,1 because the rows is too short. Hope you can explain about my question. Thank you very much 🙏🏻
Hi Mark, thanks a lot for this helpful video! I am currently working on my bachelor thesis for an event study on orphan drug approvals. I want to use the market model (I guess it's the same like the SIM just another word). In the literature it is recommended to use the OLS regression to get alpha and beta. My problem is that when making a scatter plot, it seems like my data (share and market proxy) do not have a linear relation. Does that mean I can not use this model? In your video, you just calculated the the Intercept and slope and used it as alpha and beta. But you did not make an OLS regression in advance right? I would be really happy if you could find the time to help me out at this point :( Greetings from Germany and all the best
Hi. For my research I am conducting an event study for a sample of Pharmaceutical firms partaking in M&A in the U.S. Specifically, I want to include whether the event being a horzontal or vertical acquisition plays a part in the returns. However, I wanted to focus on a longer horizon, say, 1-year, 2-year and 3-years out from the event. Is this method still applicable? If not, how should I go about this? Thank you so much
Thank you for your sharing this information ! I have 2 questions, could you please kindly help me with that? 1/ For alpha, beta, can I use the regression model to calculate the fitted value in the model, and then run the predicted value for getting the abnormal return? 2/ For the data to calculate alpha and beta, why we need to use the return for a year, and 10 days before the announcement date? In case I would like to calculate CAR for event period 41 days, how trading year or days I need for calculating the alpha and beta? Hope to hear from you soon! Thank you so much!
Regarding your questions: 1. The reason you couldn't use a fitted model from the regression is that you need the expected return from during the event period. That is, you need to use your estimated parameters (i.e., alpha and beta) on that future stock return. 2. The use of one year of data is arbitrary. You could use slightly more or slightly less. Just avoid using such a long time period that the data is stale, or such a short period that it might not be representative. You would use it up until 10 days before the beginning of your event period in that case (i.e., from day 51 in your case). Hope that helps.
Some possibilities include: 1. Use different event windows (i.e., 3, 5, 11 day etc) 2. Use different benchmarking methods (i.e., single index, three factor, four factor models) 3. If desired, you can also use Dimson, or Scholes-Williams models in order to construct benchmark returns 4. Use different indexes. Usually this makes very little difference. But, example differences include equally weighted indexes vs value weighted indexes 5. You can also use market adjusted returns. These implicitly assume that the firm's beta is 1 and its alpha is zero
Hello, so the list of the price that we take and put to excel is the "acquirer price" and "market index price" right? and which category price do we take it from? "adjusted close"?
I'm guessing the goal is to look at whether an event caused a market wide change. The beta of the market with respect to itself is 1. Thus, a slightly different approach is necessary. You could try the following: * Compare the market index of interest to other like indexes in other countries. * Test whether the movement over that period or on that day is statistically significant (ie with respect to the indexes own time series). Here, one essentially looks at whether the movement on that day is statistically different from an ordinary random movement. This implicitly assumes that past volatility is a good predictor of future volatility so the index is a valid benchmark for itself.
Why do we calculate alpha and beta using the time series reaching 10 days prior to the event date. Whilst the difference will be minimal is there reasoning behind this decision?
You would have to do so for a couple of reasons (Regardless of whether it is multiple events spread across multiple firms or multiple events for the same firm over different days). It can be laborious. Some programs - such as Eventus or within WRDS - can automate this somewhat. It could also be programmed into various programming languages to save time. First, each firm will have its own beta (and alpha) and these can change over time. So, you would need to calculate them for each event. Second, A central part of event studies is calculating abnormal returns. This is the realized return less the expected return on that day. The expected return is derived from the market return, whcih varies day-by-day. Thus, you would get different expected returns for different events.
@@financemark Thank you so much for the reply! I'm analysing the impact of FOMC statements on equity prices - so I guess the same applies in this context? Again, thanks for taking the time to reply!
I want to as a simple question. I see that you use the estimation window from 1 year before announcement until 10 days before announcement, which is 30 November 2017. With the estimation window 11 December 2017, why 10 days is not 1 December 2017 instead? Thank you
Thank you for your reply. If I have to check for abnormal returns would it be better to take previous day since our event day is holiday and then compare it with abnormal return of next trading day, that is say Tuesday.. If my original event day is Sunday(for which friday price is taken) ?
It depends on exactly which model.you are using. Many models do not include a risk free rate. people.stern.nyu.edu/ashapiro/courses/B01.231103/FFL08.pdf In terms of using a risk free rate: if you use one constant risk free rare, then it's nonstochastic (ie it's like you're just adding a number to both the left and right of the equation). If you allow risk free rates to vary over time, it can have some impact. But, typically, it's relatively minor, especially when using daily returns. Further, give that the purpose is to look at what factors drive raw returns, it is often more desirable to use raw returns rather than include a risk free rate.
Just to add, you wouldn't run a regression of the CAPM model as CAPM forces the regression intercept to be the risk free rate which is problematic when empirically estimating beta
If you want to do abnormal returns for just an index, you really have two options: 1. You could just analyze whether the return is statistically significantlly different from zero. In this case, one would effectively need to calculate the t-stat for the return on the specific day. 2. You can compare the index against say a world index. In this case, the focal index is like the firm in this video, and the world index is the market.
In a way you can: here, you would just subtract the market return from the firm's stock return. This gives you the "Market Adjusted Return". This implictly assumes the firm's alpha is zero and the firm's beta is one.
