Lec 4 - Intro to short-term study (2/2) Flashcards
How do we estimate expected returns E(r)?
Market Model
alpha - risk
beta - measures sensitivity of stock return to market return / measures a stock exposure to systematic risk (only systematic)
What are the two periods that need to be specified?
Test period (TP) - window where you measure abnormal returns
Estimation period (EP) - period where you estimate the parameters of your model
What happens in the Estimation period?
Estimating alpha (hat) & beta (hat) using a regression (in the estimation period)
What happens in the Test Period (TP)?
Calculate the expected return and abnoraml return in test period
How do we measure abnormal returns?
Abnormal return (ARit) measures abnormal return on a day in the test period
What is Cumulative Abnormal Return?
CAR measures the abnormal return over several days (from t1 to t2) in the test period
What is the average abnormal return (AARt)?
Average abnormal return on an event day
What is Cumulative Average Abnormal Return?
Measures in the period i.e -5 to 0 or 5 and you do the average from the period
How do we test the return statistical significance?
What is an error term?
Predicition error in the estimation period
By definition, its expected value is 0
By definition, the standard deviation or variance of the error term is no greater than the abnormal return
What trade-off is associated with choosing the test period?
Between the power of test and the comprehensiveness of the abnormal return
- shorter TP offers a higher power of test if all else remains equal
- shorter TP may leave out some price reaction, especially when the market isn’t completely efficient
- TP within 1 year is short term
(commonly used test periods are in the screenshot)
When should the estimation period be?
Normally before the event - data may not be available (IPOs)
Parameters may change due to the event
An EP after the event is allowed
How do you choose a market index?
Theoretically the market index contains all financial assets (CAPM)
- We use FTSE etc
Should the market index be equal weighted or value weighted?
If there are many stocks it’s not an issue as no stock dominates the index
Equal weight if there is concern that a few stocks dominate the index
What data frequency is used in short-term event study?
Daily returns
Higher power of test
What is the data frequency used for long-term event study?
Monthly
Easier to work with
Avoids things such as the bid-ask bounce
What is the bid ask bounce?
When the price of stock or asset goes back and forth between the bid and ask price
How do we identify day 0 in event study?
Announcement days - the ‘information revelation most intensive’
— Cross reference the press reports to ensure accuracy
— Include more days in the Test Period (-2, +2)
What are some other models used for short-term event study?
Mean-adjusted return
Market-adjusted mean
What is the mean-adjusted return? What are the strengths and weaknesses?
Strengths:
- Convenient
- Easy to implement
Weaknesses:
- Crude measure of abnormal return
- Risks not considered
- E(r) constant over time in time seires, and differs in cross section
What is the market-adjusted return?
Strengths:
- Easy to implement
- Don’t need separate Estimation Period (estimation is performed by observing the market return)
Weaknesses:
- Doesn’t consider risks fully
- Assumes alpha = 0 and beta = 1
- E(r) remains constant in cross section, differs in time series
What is CAPM?
- If CAPM is right, alpha should be Insignificantly different from 0
- But we still use non-zero value of alpha to calculate E(r) because alpha and beta are jointly significant
What are the strengths and weaknesses of CAPM?
Strengths:
- Relatively easy to implement
- Considers the market risk
Weaknesses:
- Other risks not considered
What is Fama-French 3 factor model?
If FF3F is right model, alpha should be insignificantly different from 0
But we still use the non-zero value of alpha to calculate E(r) -> because the other coefficients are jointly significant
What are the strengths and weaknesses of FF3F?
Strength:
- Consider more risk factors
- Need to calculate factor returns - aren’t always available
