Hedge Fund Market Neutral Strategies: Distinguishing Financial
and Operational Risk Factors
Stephen J. BrownNYU Stern
The Joint 14th Annual PBFEAand 2006 Annual FeAT Conference
第十四屆亞太財務經濟及會計會議暨2006台灣財務工程學會聯合研討會
Distinguishing operational and financial risk
Historical perspectiveOperational risk
Characterized by conflicts of interest
Financial riskThe myth of market neutrality
Robust measure of tail risk neutrality
Conclusion
The History of Hedge Funds
The first hedge fund: Alfred Winslow Jones (1949)Limited Partnership (exempt from ’40 Act)Long-short strategy20% of profit, no fixed feeUsed short positions and leverage
“Hedge Fund” (Fortune magazine 1966) Tiger Fund (Institutional Investor 1986) George Soros $3.2Billion raid on the ERM
(1992) CalPERS (2000)
Institutional concern about risk
Fiduciary guidelines imply concern for riskFinancial riskOperational risk
Institutional demandGrowing popularity of market neutral
stylesExplosive growth of funds of fundsDemand for “market neutral” funds of
funds
Operational Risk
0
0.2
0.4
0.6
0.8
1
7 15 23 31 39 47 55 63 71 79 87
Duration (Months)
Frac
tion
of Fu
nds
Sur
viving
CTAs
HedgeFund
Source: Tremont TASS (Europe) Limited
Hedge fund failure is highly predictable …
Measuring operational risk
SEC registration requirement (Feb 2006)
2270 of TASS Funds that registeredHad better past performanceHad larger assets under management15.8% had prior legal/regulatory problems
What are the correlates of operational risk?
Correlates of operational risk
“Problem” Funds
“Non-Problem” Funds
N Mean
Median
N Mean
Median
Diff
Avg Return 356
0.89
0.80 1898
0.98 0.84 -0.09*
Std Dev 354
2.60
1.79 1897
2.74 2.08 -0.14
Sharpe Ratio 354
0.33
0.29 1897
0.39 0.30 -0.06*
AUM ($mm) 325
218.2
58.74 1647
180.2
54.00 38.00
Age (Years) 358
5.65
4.50 1912
4.99 3.92 0.66**
Management Fee (%)
358
1.37
1.25 1912
1.38 1.50 -0.01
Incentive Fee (%)
358
15.23
20.00 1912
17.52
20.00 -2.29**
High Water Mark
358
0.69
1.00 1912
0.82 1.00 -0.13**
Lockup Period (months)
358
4.07
0.00 1912
4.48 0.00 -0.41
External conflicts
Problem funds
Non problemfunds
With: N % Yes N % Yes
Broker/Dealer 359 73.8 1912 24.8
Investment Comp
359 50.4 1912 16.0
Investment Advisor
359 74.7 1912 41.3
Commodities Broker
359 53.5 1912 20.3
Bank 359 40.4 1912 9.8
Insurance 359 39.8 1912 9.4
Sponsor of LLP 359 56.8 1912 22.2
Internal conflicts
Problem funds Non problemfunds
With: N % Yes N % Yes
Trade securities with clients
359 30.1 1912 8.4
Allow trading on own account
359 85.2 1912 69.6
Recommend own securities 359 74.9 1912 50.8
In-house broker dealer 359 31.2 1912 2.3
Recommends own underwriting service
359 69.4 1912 46.8
Recommends commission fee items
359 22.6 1912 15.7
Recommends brokers 359 45.7 1912 38.4
Use broker provided external research
359 81.3 1912 69.9
Towards a univariate index of operational risk
TASS Variables SEC Variables
Previous Returns -0.27 In-house broker dealer 0.06
Previous Std. Dev. -0.36 Associated with broker dealer 0.24
Fund Age -0.10 Investment company association
0.25
Log of Assets 0.09 Investment advisor association
0.24
Reports Assets 0.07 Commodity trader association 0.44
Incentive Fee -0.89 Associated with bank or thrift 0.39
Margin -0.29 Associated with insurance co 0.42
Audited -0.21 Associated with ltd. partner syndicator
0.27
Personal Capital -0.26 Trade securities with clients 0.06
Onshore -0.11 Allow trading on own account -0.12
Open to Inv. 0.04 Recommend own securities 0.32
Accepts Managed Accts -0.13 Recommends own underwriting service
0.24
Recommends commission fee items
0.28
Recommends brokers -0.35
Use broker provided external research
-0.69
Correlation Between Fraction of owners who hold 75% of firm
0.17
TASS and ADV Panels 0.41 Fraction of domestic ownership
0.28
Financial Risk
0%10%20%30%40%50%60%70%80%90%
100%
0 10 20 30 40 50 60 70 80 90 100
Size of portfolio
Per
cent
of risk
Equities
S&P500 risk
Source: Elton and Gruber 1995. Risk is measured relative to the standard deviation of the average stock
Financial Risk
0%10%20%30%40%50%60%70%80%90%
100%
0 10 20 30 40 50 60 70 80 90 100
Size of portfolio
Per
cent
of risk Equities
Hedge Funds
S&P500 risk
Hedge Fund risk
Caught by the tail
“S&P500 returns at Treasury Bill risk”Most new funds claim to be “market
neutral”Zero correlation with benchmark
Zero correlation is not a strategyZero correlation is an outcome of a
strategy These strategies fail in liquidity crises
Risk is considerably understated New concept: “tail risk neutrality”
A market neutral strategy
Data
TASS hedge funds – both dead and alive
US funds with at least 10 returns, average of 40 max of 120.
Not a lot of data per fund, but plenty when the universe is combined – nearly 50,000 fund-month observations.
An example of ‘market neutrality’Fu
nd
Retu
rns
Market Returns0.2
0.2
0.4
0.6
0.8
0.6
0.8
0.4
1.5%
1.1%
0.8%
0.4%
Beta = .28, rho = .24
Assuming MVN returns
2.5%
1.9%
1.3%
0.6%
Fu
nd
Retu
rns
0.2
0.4
0.6
0.8
S&P500 Returns0.2
0.6
0.8
0.4 Beta = .28, rho
= .24
Market neutrality in the ‘real world’
Using TASS data
2.5%
1.9%
1.3%
0.6%
Fu
nd
Retu
rns
0.2
0.4
0.6
0.8
S&P500 Returns0.2
0.6
0.8
0.4 Beta = .28, rho
= .24
Market neutrality in the ‘real world’
Long Short Equity Funds
2.9%
2.2%
1.3%
0.6%
Fu
nd
Retu
rns
0.2
0.4
0.6
0.8
S&P500 Returns0.2
0.6
0.8
0.4 Beta = .50, rho
= .37
Event driven style
3.1%
2.3%
1.5%
0.8%
Fu
nd
Retu
rns
0.2
0.4
0.6
0.8
S&P500 Returns0.2
0.6
0.8
0.4 Beta = .20, rho
= .23
Dedicated Short Sellers
4.5%
3.4%
2.3%
1.1%
Fu
nd
Retu
rns
0.2
0.4
0.6
0.8
S&P500 Returns0.2
0.6
0.8
0.4 Beta = -.91, rho =
-.61
Fixed income arbitrage
1.5%
1.1%
0.8%
0.4%
Fu
nd
Retu
rns
0.2
0.4
0.6
0.8
S&P500 Returns0.2
0.6
0.8
0.4 Beta = 0.01, rho =
0.02
Funds of Hedge Funds
Hedge Fund 1 Hedge Fund 2 Hedge Fund 3
Fund of Funds
Funds of Hedge Funds
Provides
Hedge Fund 1 Hedge Fund 2 Hedge Fund 3
Fund of Funds
Funds of Hedge Funds
ProvidesDiversification – lower value at risk
Hedge Fund 1 Hedge Fund 2 Hedge Fund 3
Fund of Funds
Funds of Hedge Funds
ProvidesDiversification – lower value at riskSmaller unit size of investment
Hedge Fund 1 Hedge Fund 2 Hedge Fund 3
Fund of Funds
Funds of Hedge Funds
ProvidesDiversification – lower value at riskSmaller unit size of investmentProfessional management / Due
diligence
Hedge Fund 1 Hedge Fund 2 Hedge Fund 3
Fund of Funds
Funds of Hedge Funds
ProvidesDiversification – lower value at riskSmaller unit size of investmentProfessional management / Due
diligenceAccess to otherwise closed funds
Hedge Fund 1 Hedge Fund 2 Hedge Fund 3
Fund of Funds
Institutions love FoF
Spectacular growth of Funds of Funds
2000: 15% of all Hedge funds were FoF2003: 18% of all Hedge funds were FoF2005: 27% of all Hedge funds were FoF
Institutional attraction of Funds of Funds
Risk management Due diligence
Funds of Funds
2.9%
2.2%
1.3%
0.6%
Fu
nd
Retu
rns
0.2
0.4
0.6
0.8
S&P500 Returns0.2
0.6
0.8
0.4 Beta = .14, rho
= .22
Relationship to LIBOR
1.0%
0.8%
0.5%
0.3%
Fu
nd
Retu
rns
0.2
0.4
0.6
0.8
LIBOR return0.2
0.6
0.8
0.4 Beta = 0.0, rho =
0.0
Fixed income arbitrage
2.0%
1.5%
1.0%
0.5%
Fu
nd
Retu
rns
0.2
0.4
0.6
0.8
LIBOR return0.2
0.6
0.8
0.4 Beta = -.02, rho =
-.05
Simple measures of tail risk exposure
Frequency of falling into lower decile for both fund and benchmark
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
-0.5 -0.3 -0.1 0.1 0.3 0.5
rho
Pro
bab
ility
MVN
MVt (3 df)
Independence an unrealistic benchmark
ConsiderMV Normal with
the same sample correlation
MV Student with 3 df
Simple measures of tail risk exposure
Frequency of falling into lower decile for both fund and benchmark
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
-0.5 -0.3 -0.1 0.1 0.3 0.5
rho
Pro
bab
ility
MVN
MVt (3 df)
Independence an unrealistic benchmark
ConsiderMV Normal with
the same sample correlation
MV Student with 3 df
0.24
0.0188
An example of ‘market neutrality’Fu
nd
Retu
rns
Market Returns0.2
0.2
0.4
0.6
0.8
0.6
0.8
0.4
1.5%
1.1%
0.8%
0.4%
Beta = .28, rho = .24
Assuming MVN returns
An example of ‘market neutrality’Fu
nd
Retu
rns
Market Returns0.2
0.2
0.4
0.6
0.8
0.6
0.8
0.4
1.5%
1.1%
0.8%
0.4%
Beta = .28, rho = .24
WW
LW
WLLL
LL should be 1.88% of sample assuming MVN
returns
Comparison with S&P500 Benchmark
Correlation with
benchmark
Binomial Crash
p-value (ind)
p-value (N)
p-value (t)
All Funds 0.28** 0 0 0
Funds of Funds 0.14** 0 0 0
Convertible Arbitrage 0.09** 0 0.033 0.840
Dedicated Short Bias -0.91** 0.997 0.112 0.838
Emerging Markets 0.66** 0 0.031 0.394
Equity Market Neutral 0.02 0.001 0.006 0.893
Event Driven 0.20** 0 0 0
Fixed Income Arbitrage 0.01 0.395 0.480 0.995
Global Macro 0.08 0.004 0.034 0.752
Long Short Equity 0.50** 0 0 0.006
Managed Futures -0.11** 0.563 0.127 0.999
Comparison with LIBOR Benchmark
Correlation with
benchmark
Binomial Crash
p-value (ind)
p-value (N)
p-value (t)
All Funds 0.00 1 1 1
Funds of Funds 0.01 1 1 1
Convertible Arbitrage 0.00 0 0 0.074
Dedicated Short Bias 0.07 0.006 0.031 0.432
Emerging Markets -0.17** 0.995 0.823 1
Equity Market Neutral 0.07** 0.148 0.567 1
Event Driven -0.04** 1 1 1
Fixed Income Arbitrage -0.05 0 0 0.007
Global Macro -0.03 0.849 0.756 0.999
Long Short Equity 0.00 1 1 1
Managed Futures -0.02 0.525 0.399 1
Logit Specification
Boyson, Stahel and Stulz [2006] suggest running logit regressions of whether a fund index crashes in a month upon the market return and a dummy for market crashes. A positive coefficient on the dummy indicates additional dependence during crashes.
Lacks power when run on a single index. We run the regressions on the cross-section.
Conclusions
Operational riskImportant role for due diligenceCharacterized by internal and
external conflicts of interestFinancial risk
Undiversifiable crash risk lurks in hedge fund returns, despite their seemingly light dependence in normal times.