Transcript Slide 1

Hedge Fund Alpha and Beta
John H. Cochrane
University of Chicago GSB
Betas
•
•
Alpha = Average return – beta * E(factor)
HF data is pretty hopeless for ER, alpha
1.
2.
3.
4.
•
•
•
•
Survivor / backfill / self-reported
σ/√T , made worse by option-like returns
Alpha is boring. Who bought HF for 1-2% alpha?
“Sure, average alpha is low. My alpha is big.” Untestable.
HF data still useful for betas.
Betas are interesting on their own! Compensation
benchmark, risk management, how to put HF in
portfolio all depend on beta.
HF marketing: “Market neutral”, “Absolute return.”
HF reality: Big betas!
Hedge fund alphas and betas – lags and stale prices
Style
ER (%/mo)
a
b
a3
b3
Index
0.64
0.46
0.28
0.36
0.44
Std. errors
0.20
0.17
0.04
Short
-0.53
0.10
-0.94
0.13
-0.99
Emerg mkts
0.39
0.00
0.58
-0.07
0.69
Event
0.61
0.46
0.22
0.38
0.37
Global Macro
0.93
0.82
0.17
0.74
0.31
Long/Short Eqty
0.73
0.42
0.47
0.32
0.65
Not zero!
Bigger with lags
Smaller with lags
Really not zero.
“Alternative asset?”
Long-short doesn’t
mean zero beta!
rti  a  brts& p500  ti
rti  a3  b1rt sp  b2 rt sp1  b3rt sp2  b4 rt sp3   ti
b3  b1  b2  b3  b4
•Lags are important – stale prices or lookback option
•Betas are big!
Source: my regressions using CFSB/Tremeont indices at hedgeindex.com, idea from Asness et al JPM
•Correlation with the market is obvious.
•Getting out in 2000-2003 was smart! (Mostly due to Global/Macro group)
Monthly returns on Global Macro HF and US market
•“Global macro” yet you see the correlation with US market
•Lagged market effect is clear in 1998. Is Nov/Dec 1998 unrelated to Oct?
•Dramatic stabilization / change of strategy in mid 2000
Monthly returns on Emerging Market HF and US market
•“Emerging markets diversify away from US investments, give us
access to a new asset class?”
•Names: yes. Betas: no.
“Short volatility” and option-like returns, betas on option strategies
•Source: Mitchell and Pulvino, using CFSB/Tremont merger-arb index
•News: 1) “occasional catastrophes’’ 2) catastrophes more likely in market declines
Hedge fund up/down betas
Style
b3
b up
b down
Index
0.44
0.08
0.77
Short
-0.99
-0.22
-1.82
Emerg mkts
0.69
0.08
1.16
Event
0.37
0.18
0.47
Global Macro
0.31
-0.08
0.66
Long/Short Eqty
0.65
0.19
1.18
rti  a3  b1rt sp  b2 rt sp1  b3rt sp2  b4 rt sp3   ti
b3  b1  b2  b3  b4
Example: if the market goes
up 10%, the HF index goes up
0.8%. But if the market goes
down 10%, the HF index goes
down 7.7%!
rti  a  bup (rt sp  0)  bdown (rt sp  0)  ti
(Includes 3 lags)
•Many near, or above 1. These are big betas!
•Many HF styles are much more sensitive to down markets = write puts.
•Source: my regressions using hegefundindex.com data; following Asness et al JPM
Option return benchmarks
ER
(%/mo)
alpha
SPPo
(puts)
SMB
(size)
HML
(value)
Event Arb
1.03
0.04
-0.92
0.15
0.08
Restructure
1.29
0.43
-0.63
0.24
0.12
Event driven
1.33
0.20
-0.94
0.31
0.12
Rel. value arb
1.15
0.38
-0.64
0.17
0.08
SPPo = return from rolling over out-of-the-money puts
Source: Agarwal and Naik RFS, using HFR data
• Morals:
1. Including option benchmarks can reveal big betas.
2. And hence alphas a lot less than average returns.
Additional benchmarks matter too!
Style
Rm+
Rm-
Term+
Term-
Corp+
Corp-
Index
0.07
0.72
0.27
0.55
1.01
0.94
Conv arb
0.25
0.20
0.10
0.35
-0.22
1.03
Short
-0.32
-1.75
-0.05
0.19
0.62
-0.66
Emerg mkts
0.13
1.03
-0.44
0.20
0.73
2.48
Event
0.21
0.39
0.15
0.32
0.30
1.43
BondArb
0.08
0.12
-0.02
0.35
0.44
1.40
Global Macro
-0.09
0.66
0.66
1.17
2.29
1.53
Long/Short Eqty
0.16
1.05
0.09
0.22
0.38
0.84
•Term = long term gov’t bond return – t bill rate
•Corp = corporate bond return – long term gov’t
•Big betas, especially on corp (default spread)
•Often much more for bad news than for good news
•Source: my regressions using hegefundindex.com data
Implications
•
Need lots of factors.
1.
2.
3.
4.
•
Standard regression method is completely inadequate.
1.
2.
3.
•
Market, value, size, momentum, term, default, currency…
Plus options on all of these.
(Next: + mechanical timing strategies that change all exposures.)
(Next 2: + mechanical rules that update the coefficients.)
More RHV than data points.
HF styles shift – betas not constant over time.
HF style groups mean little. “Small cap growth’’ vs. “Global macro.”
Whole style/selection concept makes little sense anymore.
1.
2.
3.
4.
5.
6.
Does “style” (beta x E(f), passive, no fee) vs. “selection” (alpha, active,
fee) make any sense with 27 factors, time-varying premia?
“You could reproduce HF return with xyz mechanical strategy.” (e.g.
write put.) -- But you can’t, and the investor dosn’t. So what?
The only beta, alpha that matter are those on the investor’s portfolio. If
the investor has not optimized on the extra factors, it’s alpha!
Theorem: There is no alpha, there is only style. Trade = f(information)
The theorem is silly.
“Style” is “selection,” worth a fee!
HF as part of a portfolio, not a standalone investment
Standard “passive” plus smaller “active” including multiple HF
•The Absolute Return portion of the portfolio is primarily invested in non-directional hedge funds. That is, returns should be
independent of the direction of global equity, fixed income or currency markets. Strategies include Global Convertible Arbitrage,
Global Merger Arbitrage, Long/Short Equity and Blended Strategies….
Hedge funds as part of a portfolio
•
Problem 1: Risk management. Must know betas!
–
•
How much are you overall short volatility?
Problem 2: Transaction cost and fee explosion.
1. Overall portfolio is what matters.
a. Portfolio is (10 A, 10 B). HF is long A short B.
b. Is (11A, 9 B) really worth short cost, 2+20 fee?
2. “Diversify with multiple HF.” (SG, WSJ 9/29)
a.
b.
c.
d.
Is HF #1 long A, short B, HF #2 short A, long B?
You pay ½ ( 2 + 20 ) for sure, plus short costs for nothing.
HF are not taking true idiosyncratic risk. (If so, 2+20 is a disaster!)
Hedge (style betas) with passive, not multiple active investments!
3. Cost explosion – portfolio of options ≠ option on portfolio.
a. 100 mean zero stocks in one fund: 2% for sure.
b. 100 stocks in 100 funds: 2% + ½ (20%) for sure!
•
If funds of funds can solve these problems, maybe
they’re worth another 2 + 20!
Bottom line:
• HF must figure out and disclose betas (and tail
probabilities), based on holdings not regressions.
(Compensation for accuracy?) – Or alpha, beta to you.
• It’s OK to “shop for bargains” (earn high risk premia)
not just alpha, arbitrage, magic.
• Honesty might also stop panicked withdrawals.
• Alpha is boring. “Style” vs. “Selection” is dead.
• Understanding HF: A brilliant marketing success in a
marketing business.
– “Absolute Returns,’’ ”Market-Neutral,” “Alternative asset,”
“Near-Arbitrage”… “Alternative beta,” “Entrepreneur”
– Whatever they mean, they separate rich people, money.
– 2% + 20% “We only charge if we win.”