Survival and Performance of Pension Fund Money Managers

1
Survival and Performance of Pension Fund Money
Managers

• Janis Berzins
• Charles Trzcinka
• Indiana University
• T. Daniel Coggin
• UNC Charlotte
• Acknowledgment and disclaimer
• The authors are grateful for the assistance of
  the Virginia Retirement System, and The Mobius
  Group.
• The opinions in this presentation are solely
  those of the authors.

2
Why Study Institutional Money Managers?

• Definition An Institutional Money Manager is a
  Manager who is not covered by the Investment
  Company Act of 1940
• They manage more money for fewer clients
• Institutional managers are larger than mutual
  funds and their customers typically have more
  money than the managers
• The average minimum investment in 2003 for our
  sample is 8.9M.
• While the average account size for a mutual fund
  is about 28,000 in 2003 according to the Mutual
  Fund Factbook, Investment Company Institute.
• They are less regulated
• managers do not own the securities they manage
• beyond what is required by clients there are no
  reporting requirements, no diversification
  requirements, no restrictions on leverage or
  fees, no advertising rules
• ERISA rules apply to the client, not to the
  manager
• All Previous Studies have small samples,
  survivorship bias and selection bias.

3
The Mobius Survey of Institutional Money Managers

• The Mobius Group,(located in North Carolina)
  surveys money managers every quarter. The 1,000
  or so clients of Mobius use the data in their
  search for better managers. The clients are
  pension fund sponsors, endowment funds,
  investment banks, consultants, mutual funds and
  high net worth individuals.
• If a manager does not fill out the survey there
  is a high probability that the manager will not
  be included in a search
• If the manager gives inaccurate information there
  is a high probability that the market will
  penalize the manager
• June 1993 Survey
• 871 Domestic Equity Portfolios managed 0.66
  trillion in 79,280 accounts
• 608 active managers with 0.49 trillion in 62,559
  accounts
• 25 passive managers with 0.02 trillion in 486
  accounts
• December 2003 Survey (actual institutional
  portfolios only)
• 2058 Domestic Equity Portfolios managed 2.55
  trillion in 412,909 accounts
• 1889 active managers with 2.30 trillion in
  395,907 accounts
• 68 passive managers with 0.18 trillion in 1,138
  accounts

4
Data Quality Checks

• Economic Quality Check Mobiuss products must
  meet the market test. Data is their only product.
• Consultant data only receives an internal check
  and is not sold publicly
• Data of Mobius and its competitors receives
  internal and external scrutiny.
• Empirical Quality Checks
• 10 Mobius Managers were used by Virginia
  Retirement System Mobius Returns Match the
  internal records of the VRS
• Compare worst decile Mobius Managers with worst
  decile Piper Managers
• 50 Common Managers in Decile
• Returns Identical
• Revision of History between each survey (for
  whole data base)

5
Hypotheses

• Can Managers add value?
• Lakonishok, Shleiffer and Vishny (LSV) argue that
  the money management industry is a cottage
  industry riddled with agency costs
• Coggin, Fabozzi and Rahman (CFR) conclude that
  money-managers add substantial value over style
  benchmarks
• Christopherson, Ferson and Glassman (CFG)
  conclude
• persistence of poor performance (negative alphas)
  but not of good performance.
• Bad managers survive
• If LSV are correct about agency costs there will
  be an effect on performance
• Managers are measured by factors other than
  returns. (e.g. supplying style stories). This
  means survival bias should be low or negligible.
• Loadings on factors will be non-stationary as
  managers change portfolio weights to match client
  perceptions.

6
The Money Manager UniversePortfolios must meet
the following screens

• The portfolio reports gross returns for at least
  one quarter
• Net returns are less commonly reported. Managers
  use fee schedules with break points and the
  charged clients depends on the size of the
  account
• Managers routinely give discounts.
• The manager includes terminated accounts in the
  portfolios return.
• Manager can simply report the return of all the
  accounts under management at the end of the
  period, not at the beginning.
• Portfolio returns can be restated for terminated
  accounts.
• The manager has full discretion over the account

7
The Mobius Universe of Institutional Money
Managers of Domestic Equity
8
Flexible Regression Model

• Following Kalaba and Testfatsion(1989)
• assume and E(et)
  cov(et, e t-k) 0 and Var(et) s2
• assume the initial density g(ß1) is
  constant(diffuse) and choose those coefficient
  values that maximize the conditional expectations
  E(ß ty1,,yt)
• algorithm models two sources of error
• measurement error (the usual regression error)
• dynamic error (the result of changing betas)
• Kalaba and Testfatsion show that the collection
  of all possible sse,ssd is bounded away from
  the origin by an envelope. To estimate a point on
  the envelope they propose minimizing a weighted
  sum
• ?sse
  (1- ? )ssd
• smoothness in beta changes is used in the
  estimation algorithm, simulation evidence
  suggests the model is robust against large shifts
  in beta.

9
What we think the tables show

• The survival rate, defined as the number of
  quarters a portfolio exists, is lower than in the
  mutual fund industry and depends on the date the
  portfolio is created
• There is equally-weighted average survivorship
  bias of 31.4 basis point bias per quarter in
  gross return and an asset weighted bias of 6.4
  basis points.
• They beat the SP500.
• The mean SP500 alphas are 65 basis points per
  quarter and a median 26 basis points per quarter.
  
• These managers are using size, book-to-market and
  momentum to earn a higher return than the SP500.
• If managers are penalized for taking momentum
  bets, alphas are negative.
• The average time-varying alpha is smaller than
  the OLS alpha.
• SP500 alphas are much more volatile than Fama
  French alphas.
• The market beta and book to market betas appears
  constant and size betas vary over time.

10
Survivor Returns Less Non Survivor Returns(not
in the paper)