Comment%20on%20

1
Comment on Mutual Fund Tournament Test Do
Shareholders Benefit from Fund Managers
Risk-Taking Behavior

• Pei-Gi Shu
• Department of Business Administration
• Fu-Jen Catholic University

2
Summary

• Extreme interim losers reduce risk while medium
  winners increase risk inconsistent with the
  tournament hypothesis proposed by Brown,
  Harlow, and Starks (1996)
• Career risk account for the risk-deduction
  behavior of extreme losers. Overconfidence and a
  call-option payoff scheme bounded by a cap
  account for the risk-increasing behavior of
  medium winners.
• The interim risk shift of fund managers is not to
  the best interest of fund investors.

3
Contributions

• Use quartile-based instead of median-based
  performance categorization to further contrast
  the risk-taking behavior of mutual fund managers.
• Include alternative hypotheses to discuss the
  performance-risk relation of mutual fund
  managers.
• Well-structured to link the relationship among
  performance, risk-shifting behavior, and ex-post
  returns.
• Document various types of interim risk changes
  for different fund categories.

4
Comments and Suggestions (1)

• How to include and contrast different arguments
  (career concern, attribution theory,
  overconfidence, prospect theory, and option-like
  payoff scheme) into testable hypotheses?
• For example, it would be more applicable for
  repeated or long-term losers to take serious
  concern of being fired. Overconfidence is also
  somewhat acquired or accumulated through
  consecutive successes.

5
Comments and Suggestions (1) Con.

• In contrast, prospect theory or option-like
  payoff scheme come by nature is independent from
  performance evaluation period.
• Therefore, the frequency and time-frame of
  performance evaluation could be further taken
  into account in experiments as to discern the
  performance-risk relationship for mutual fund
  managers.

6
Comments and Suggestions (2)

• If the concern of being fired dominates
  tournament argument for extreme losers, it is
  suggested to provide evidences that losers are
  more likely to be replaced.
• Moreover, I see no reason why the extreme losers
  who reduce portfolio risk still suffer inferior
  ex-post returns? A further discussion is needed.

7
Comments and Suggestions (3)

• Table 2 Why the cutoff of RAR (s2/s1) is set at
  sample median instead of 1 that could be more
  intuitively appearing to illustrate whether fund
  managers increase or decrease their portfolio
  risk after interim performance evaluation?

8
Comments and Suggestions (4)

• Table 4 Regression Analysis The dependent
  variable, stdi,t/stdi,t-1, is presumably to be
  negatively correlated with stdi,t-1 (independent
  variable). I would suggest alternative dependent
  variable such as a risk-increasing dummy that
  takes the value of 1 if the fund manager
  increases risk and 0 otherwise.