Empirical Methods of Finance
Exam: 11-12-2017
Resit:
Part 2: Crego
Week 8: Event Studies
What is an event study?
A test of the change in stock (or bond) prices around specific events
o Examples: corporate earnings announcements, merger announcements, macro-
economic news etc.
Why do we conduct / use event studies?
Some examples
To examine the impact of some event on the wealth of
the firm’s security holders.
To test the hypothesis that the market efficiently
incorporates new information
Efficient Market Hypothesis
Characteristics of the efficient market hypothesis (EMH)
A market in which prices always fully reflect available and relevant information is called a
efficient market
Prices will change only when (new) information arrives
o But information, by definition, cannot be predicted ahead of time
o Therefore, price changes cannot be predicted ahead of time
Efficiency refers to the two aspects of a price adjustment to information: the speed and quality
(direction and magnitude) of the adjustment. The primary role of the capital market is the allocation
of ownership of the economy’s capital stock. Therefore, asset prices should provide accurate signals
for resource allocation. There are different forms within the efficient market hypothesis:
Weak form: prices reflect all information contained in the record of past prices and
fundamental values, implying that the market is efficient and thus reflecting all market
information.
Semi-strong form: prices reflect not only past information but all other published
information. This hypothesis assumes that stocks adjust quickly to absorb new information
besides the former known market information. This implies that the semi-strong form also
incorporates the weak-form hypothesis.
Strong form: prices reflect not only just public information but any information that might be
relevant. This indicates that this form incorporates both former forms and thus no investors
would be able to profit above the average investor.
,Design an Event Study
1. Identify the event of interest and the timing of the event
2. Specifying a “benchmark” model for normal stock return behaviour
Timing
For an event study we make a difference between calendar time
and event time. Suppose we conduct an event study for the
examination of ‘Empirical Methods in Finance’. We conduct several
preliminary tests before the real exam and on 11-12-2017 the real
exam takes place (t=0).
Calendar time: 11-12-2017
Event time: t=0
Typically, an event study uses many (N) observations on similar events. These can be quite far apart in
calendar time, but for conducting the study one uses event time.
In the event study we analyse the event period vs. the estimation period.
1. Measuring the event
At first event studies measure and test abnormal returns around an event.
Definition: Abnormal return = return – normal return.
Normal return: returns we expect in normal circumstances without an event.
2. Define the benchmark model (Normal Returns)
a. Own average return (mean adjusted) MARM
i. Benchmark model NR:
b. Market return (market adjusted) MAM
Benchmark model NR:
c. Market model MMM
i. Compute the market model returns:
ii. Benchmark model NR:
d. CAPM
i. Compute the CAPM return:
ii. Benchmark model NR:
iii. CAPM = MMM if
Stata Script inside loop NR in Estimation + Event .
, Conduct the event study
1. Collect return and benchmark return data
2. Calculate returns and average these over all (N) events
3. Define Average Abnormal Returns
Testing for Significance
Null Hypothesis: no abnormal price effects
H0: E(ARi,t) = 0
If the abnormal returns are independent, identically (iid) and
normally distributed, the standardized average abnormal
return has a standard normal distribution.
T-test
Typically we don’t know the variance of the returns so we need to estimate the
standard deviation. This is mostly done by using a cross sectional estimator:
Cross sectional estimator since it uses the abnormal returns of all
events
i = 1,…,N
If the abnormal returns are independent, identically (iid) and normally
distributed, this statistic has a student-t distribution with (N-1) degrees of
freedom.
The requirement that returns have a normal distribution is very strong, and typically rejected for daily
data. Fortunately we do not need to assume normality since we often can apply the Central Limit
Theorem.
N > 30 is typically needed for the approximation to be reasonably good
Notice that the assumption of independence rules out e.g. clustering of events on the same
calendar day.
Event studies can also use the cumulative abnormal return
,
Exam: 11-12-2017
Resit:
Part 2: Crego
Week 8: Event Studies
What is an event study?
A test of the change in stock (or bond) prices around specific events
o Examples: corporate earnings announcements, merger announcements, macro-
economic news etc.
Why do we conduct / use event studies?
Some examples
To examine the impact of some event on the wealth of
the firm’s security holders.
To test the hypothesis that the market efficiently
incorporates new information
Efficient Market Hypothesis
Characteristics of the efficient market hypothesis (EMH)
A market in which prices always fully reflect available and relevant information is called a
efficient market
Prices will change only when (new) information arrives
o But information, by definition, cannot be predicted ahead of time
o Therefore, price changes cannot be predicted ahead of time
Efficiency refers to the two aspects of a price adjustment to information: the speed and quality
(direction and magnitude) of the adjustment. The primary role of the capital market is the allocation
of ownership of the economy’s capital stock. Therefore, asset prices should provide accurate signals
for resource allocation. There are different forms within the efficient market hypothesis:
Weak form: prices reflect all information contained in the record of past prices and
fundamental values, implying that the market is efficient and thus reflecting all market
information.
Semi-strong form: prices reflect not only past information but all other published
information. This hypothesis assumes that stocks adjust quickly to absorb new information
besides the former known market information. This implies that the semi-strong form also
incorporates the weak-form hypothesis.
Strong form: prices reflect not only just public information but any information that might be
relevant. This indicates that this form incorporates both former forms and thus no investors
would be able to profit above the average investor.
,Design an Event Study
1. Identify the event of interest and the timing of the event
2. Specifying a “benchmark” model for normal stock return behaviour
Timing
For an event study we make a difference between calendar time
and event time. Suppose we conduct an event study for the
examination of ‘Empirical Methods in Finance’. We conduct several
preliminary tests before the real exam and on 11-12-2017 the real
exam takes place (t=0).
Calendar time: 11-12-2017
Event time: t=0
Typically, an event study uses many (N) observations on similar events. These can be quite far apart in
calendar time, but for conducting the study one uses event time.
In the event study we analyse the event period vs. the estimation period.
1. Measuring the event
At first event studies measure and test abnormal returns around an event.
Definition: Abnormal return = return – normal return.
Normal return: returns we expect in normal circumstances without an event.
2. Define the benchmark model (Normal Returns)
a. Own average return (mean adjusted) MARM
i. Benchmark model NR:
b. Market return (market adjusted) MAM
Benchmark model NR:
c. Market model MMM
i. Compute the market model returns:
ii. Benchmark model NR:
d. CAPM
i. Compute the CAPM return:
ii. Benchmark model NR:
iii. CAPM = MMM if
Stata Script inside loop NR in Estimation + Event .
, Conduct the event study
1. Collect return and benchmark return data
2. Calculate returns and average these over all (N) events
3. Define Average Abnormal Returns
Testing for Significance
Null Hypothesis: no abnormal price effects
H0: E(ARi,t) = 0
If the abnormal returns are independent, identically (iid) and
normally distributed, the standardized average abnormal
return has a standard normal distribution.
T-test
Typically we don’t know the variance of the returns so we need to estimate the
standard deviation. This is mostly done by using a cross sectional estimator:
Cross sectional estimator since it uses the abnormal returns of all
events
i = 1,…,N
If the abnormal returns are independent, identically (iid) and normally
distributed, this statistic has a student-t distribution with (N-1) degrees of
freedom.
The requirement that returns have a normal distribution is very strong, and typically rejected for daily
data. Fortunately we do not need to assume normality since we often can apply the Central Limit
Theorem.
N > 30 is typically needed for the approximation to be reasonably good
Notice that the assumption of independence rules out e.g. clustering of events on the same
calendar day.
Event studies can also use the cumulative abnormal return
,
