Lec 6 - More on Event Study

Question 1

What is the The Thin Trading Problem?

Answer 1

Occurs when assets (shares) aren’t continuously traded
- No trading at all on a specific day
- Not traded every second within a day

Question 2

What is the thin trading problem also known as?

Answer 2

Non-synchronous trading problem

Question 3

What happens to the parameters in presence of the thin trading problem?

Answer 3

They are biased
- For infrequently traded sahres, beta is biased downward
(for thinly traded shares, alpha is biased upward) c

Question 4

How can we detect thin trading?

Answer 4

• Succession of zero daily returns (price or return index doesn’t change for a number of days)
• Low trading volume

Question 5

How can we mitigate the impact of the thin trading problem? What is the problem with doing this?

Answer 5

• Use data of lower frequency
• leads to lower power of test

Question 6

What is the Scholes and Williams method (1977)?

Answer 6

Extension of standard market model designed to address biases arising from non-synchronous trading in the stock and market index data

Question 7

What does the Dimson method (1979) do?

Answer 7

Reviews problems caused by infrequent trading, and presents a method for measuring beta when share price data suffers from this

Question 8

How is the Dimson method carried out?

Answer 8

Stock returns are regressed on the contemporaneous market return, the lead market returns and lagged market returns, in a single regression:
Alpha is given by the regression

Question 9

What does the Scholes Williams method require?

Answer 9

Stock returns to be available for the day before (lead) and day after (lag) the market return-date

Question 10

What does the Dimson method allow?

Answer 10

More than one leads and lags of the market returns - it’s more useful when thin trading is very severe - stocks returns don’t have to be available successively around the market-return date

Question 11

When is it best to use the Scholes Williams method?

Answer 11

When thin trading is not very severe

Question 12

Which is a more accurate showing of returns for investors, BHAR or CAR?

Answer 12

BHAR is consistent with wealth change, CAR is not

Question 13

What is the event parameter approach?

Answer 13

Designed by Malatesta, based on asset pricing models (market model to begin)
Delta takes value of 1 if the time is in the test period, and 0 if it isn’t
Gamma (y looking one) measures average daily abnormal return

Question 14

What is a problem relating to the event parameter approach?

Answer 14

If an event fundamentally changes the entire profile of a company and the risk exposure of the company, the Beta can change (from before to after the event) - if the Beta changes, it needs to be incorporated in the regression model by introducing a dummy variable (which takes value of 1 for post event period and 0 otherwise)

Question 15

What has the time series of returns have to do with event parameter approach?

Answer 15

Time series of returns can include returns both before and after event - assumes Beta doesn’t change
Time series can have gaps if there are rumours of other events occuring

Question 16

How can you calculate the cumulative abnormal returns over several days?

Answer 16

Multiply by the number of days

Question 17

What is a discrete return?

Answer 17

Measures the holding period return without compounding (period = a day, week, month, quarter, or a year)
Some databases give the return index instead of return itself - this requires a calculation

Question 18

What is a continuous return?

Answer 18

Measures return over the holding period assuming continuous compounding - assumes dividends and interest are recieved every single second and this is reinvested in the portfolio continuously

Question 19

Which type (continuous or discrete) of data is more suitable for BHAR?

Answer 19

Discrete (because continuous introduces negative bias, Barber and Lyon 1977)

Question 20

What are we trying to understand when performing an event study? How is it carried out?

Answer 20

When performing an event study - we want to understand the wealth impact of an event by relying on a model
This model can be an asset pricing model or a characteristic based matching firm/reference portfolio

Question 21

What is the joint hypothesis problem?

Answer 21

When a significant abnormal return can be the cause of either:
- Real wealth impact of the corporat event
- Mis-specified model (so abnormal return is due to un-captured risk factors)

Question 22

When conducting an event study, what are the two things a researcher is hypothesising about?

Answer 22

• The wealth impact of an event
• Correctness of the model used
  (more of a concern for long-term study because if the model is not correct the prediction error compounds in the long run - less of an issue in short-term event study)
  (because there is no perfect model, the joint hypothesis problem is always present)

Question 23

How does the joint hypothesis probem exist in the test of market efficiency?

Answer 23

• Tests of market efficiency is a joint test of the model used in the test and the market efficiency
• Any test of efficiency must assume an equilibrium model that defines normal security returns (if efficiency is rejected, this is because the market is truly inefficient or because an incorrect equilibrium has been assumed - this joint hypothesis problem means that market efficiency as such can never be rejected)

Question 24

What do we want to test and what do we assume about market efficiency?

Answer 24

We want to test the hypothesised effect of a corporate event on asset prices
The test isn’t about market efficiency - we assume semi-strong form market efficiency

Question 25

True or false? In a semi-strong form efficient market, share price always moves in response to public announcement?

Answer 25

In theory yes, as share prices adjust to public information rapidly in an unbiased fashion (but always is wrong as analysts predictions could be correct initially in which case the share price won’t move)

Question 26

Why don’t company news announcements in Mexico have a price impact?

Answer 26

Bhattacharya, Daouk, Jorgenson and Kehr (2000) paper studying Mexico in 1970s
Insider trading is crux of issue - they were using private information and impounded the earnings information before the earnings were announced - so on the day of earnings announcement there was no market response
Today insider trading is forbidden and share prices respond to earnings announcements

Question 27

What can be dervied from the Mexico study?

Answer 27

That an efficient market doesn’t necessarily mean it is a fair on
(insiders make market more efficient but it isn’t fair for regular shareholders)

Question 28

Is a significant long-term performance after an event inconsistent wtih EMH in a clinical study?

Answer 28

If a market is efficient, the market should react to the news instantly accurately
All information should be incorporated in stock prices instantly - so in long run we shouldn’t observe any abnormal returns
However it may not be due to uncertainties post the acquisition
On the completion date, we don’t know how the acquistion will playout in the companies performance, even though we know 2 companies agreed to combine (we don’t know if this is the right decision for shareholders)