Longrun Performance after Corporate Financing Events

1
Long-run Performanceafter Corporate Financing
Events
2
(No Transcript)
3
(No Transcript)
4
Caveats

• Brav and Gompers (1997)
• VC-backed IPOs do not under-perform similar size
  and B/M firms that do not issue equity
• They also do not under-perform using Fama-French
  and using reference portfolios
• Gompers and Lerner (2001)
• IPOs from 1935 to 1972 may or may not have
  underperformed.

5
Another Caveat

• 1526 Firms is not a whole lot of data

6

7
No underperformance using CARs
Underperformance Using BHARs
8
How to measure abnormal performance?

• BHAR or CAR?
• Doesnt matter for short-run studies
• Whats the right benchmark?
• Reference portfolio?
• Control firm?
• Asset pricing model?

9
Barber and Lyon

• Experiment to find the size/power of LR tests
• Use alternative measures and benchmarks
• Pick CRSP firms at random
• Draw 1000 random samples of 200 firm event-months
  (out of a possible 1.1 million firm event-months)

10
CARs and BHARs
11
CARs and BHARs

• The standard example
• All stocks have 0 expected return
• A stock doubles, then loses 50.
• BHAR 0 - 0 0
• CAR 100 - 50 50
• Barber and Lyon conceptually we favor the use
  of BHARs
• Measurement bias CARs are a biased measure of
  BHARs

12
Direction of bias?
13

14
Skewness bias

• Stock returns are highly skewed a few big
  winners
• Particularly a problem for IPOs
• Uncertainty about how many big winners
• Positive BHAR may be indicative of (by chance)
  more winners than expected
• Presence of big winners inflates sample standard
  deviation
• Conclusion Sample mean and standard deviation
  tend to be positively correlated.
• DIRECTION OF BIAS? Not clear to me (THIS WHY WE
  SIMULATE THINGS) simulations show that its
  usually a negative bias.

15
Other Biases

• New Listing Bias
• In any reference portfolio, new firms are
  constantly being added.
• If these firms have different return patterns
  than seasoned firms, it will bias the results
• E.g., should you invest in firm A or B? Suppose
  firm C is later added, and is an underperformer.
  Then A looks to have positive performance
  relative to the market.
• Rebalancing bias
• Sample firms BHAR is calculated w/o rebalancing,
  reference portfolio is.

16
Summarizing

• With CAR
• Measurement Bias ()
• Skewness Bias (-)
• New Listing Bias ()

• With BHAR
• Rebalancing Bias (-)
• Skewness Bias (-)
• New Listing Bias ()

17
Control Firm Approach

• Most or all of these biases go away
• New Listing Bias?
• No. (Control firm is chosen ex-ante)
• Skewness Bias?
• No. (Both firms have similar skewness)
• Rebalancing Bias?
• No. (Neither is rebalanced)
• Measurement Bias?
• Not if you use BHAR

18
How they pick control firms

• Pick firm closest in size
• Pick firm closest in B/M
• Pick a firm which is similar in both
• Identify firms in 70,130 of market cap
• Of those, pick the one with closest B/M
• Other approaches use industry, use beta,
  propensity matching

19

20

21

22

23
Conclusion

• CARs biased upward
• This is a big deal for test size when using
  reference portfolios
• For control firm approach, theres no obvious
  effect
• Reference portfolio approach has more power
• Fama-French generally least favored