Event Studies

1
Event Studies

• Reference Campbell, Lo and MacKinlay, The
  Econometrics of Financial Markets, Ch. 4.
• Basic Idea
• Quantify how markets react, typically over a very
  short horizon, to announcements made by firms.
• Some examples
• Stock splits
• Earnings announcements
• Seasoned equity offerings.

2
Interpretations

• If markets are efficient with respect to the
  information contained in the announcement, then
  prices should react quickly.
• New evidence challenges this view (underreaction,
  overreaction, etc.).

3
Seasoned Equity Offerings

• SEO event returns
• Stock price drops on announcement (about 2 on
  average)
• Prior to announcement, price runs up dramatically
  (40-70 in prior year)
• After announcement, stock return is relatively
  low (about 30 below comparable firms after 5
  years).

4
Event Study Recipe

• Event definition
• Selection criteria
• Normal and abnormal returns
• Estimation procedure
• Testing procedure
• Empirical results
• Interpretation and conclusions

5
Event Study

• Event Definition
• Define event of interest (e.g. earnings
  announcement)
• Define event window (typically in days day of
  announcement and day after)
• Selection criteria
• What firms are included?
• Over what time horizon?
• Possible seleciton biases?
• Normal and abnormal returns

6
Event Study

• Estimation Procedure
• Define estimation window (e.g. 120 days prior to
  event)
• Define conditioning information (e.g. market
  return)
• Testing procedure
• Calculate abnormal returns
• Determine statistical significance

7
Event Study

• Empirical results
• Cross-sectional dependence
• Robustness
• Interpretation and conclusions
• How does information release affect security
  prices?
• Can alternative hypotheses be rejected?

8
Measuring Normal Performance

• Economic vs. Statistical Models
• A statistical model for normal returns simply
  incorporates potentially relevant conditioning
  information (e.g. market returns).
• An economic model typically places restrictions
  on the statistical models (e.g. the CAPM).
• Event studies usually make use of statistical
  models. Examples
• Market model
• Factor models
• Matched characteristics (e.g. by size)

9
Constant-Mean-Return Model

• In daily data, apply to nominal returns.
• In monthly data, apply to excess or real returns.

10
Market Model

• Common choices for market
• CRSP Value Weighted Index,
• CRSP Equal Weighted Index,
• SP 500.

11
Factor Model

• Common Factors
• Fama-French factors.
• Anything else that is return relevant.

12
Matching Models

• Common matches
• Size and/or book-to-market match,
• industry match,
• momentum match,
• Size and/or book-to-market index match.

13
Economic Models

• CAPM and APT
• These models are not often used.
• Think of the relevant models as serving to
  condition on contemporaneous macro information
  affecting all returns.
• For a typical short horizon event study, the
  choice of conditioning model has little effect on
  results. It's important, however, to check
  different models and verify that results are
  robust.

14
Using the Market Model

• Using data from the estimation window, perform
  the following regression
• Aggregating across time

15
Using the Market Model

• Aggregating across individuals

16
Examples

• From CLM Ch. 4
• Canadian Stock Splits