Performance
P
f
and
dC
Capital
it l Fl
Flows
in Private Equity
q y
Q Group Fall Seminar 2008
November 2008
November,
Antoinette Schoar, MIT and NBER
O
Overview
i
Is private equity an asset class?
‹ True story lies beyond the aggregates
‹
♣ Large variance and persistence in returns
♣ Idiosyncratic
Idi
ti risk
i k matters
tt
‹
Implications for risk management
Aggregate PE returns are
pro-cyclical
Venture returns versus 6-year cumulative S&P
Buyout and Venture Returns Relative to S&P
Net Annualized IRR to Funds vs. 5 Year S&P
80
70
60
50
40
Buyout Returns
30
Venture Returns
20
S&P 5 Year
10
0
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
-10
-20
-30
30
Source: Venture Econom ics
1994
1995
1996
1997
1998
1999
2000
M lti l are pro-cyclical
Multiples
li l
Venture multiples versus 6-year S&P
Buyout and Venture Multiples Relative to
Multiple on S&P for next 5 years
5
4.5
4
3.5
3
Buyout Multiples
2.5
Venture Multiples
S&P 5 Years
2
1.5
1
0.5
0
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
Source: Venture Econom ics
1995
1996
1997
1998
1999
2000
N t ttoo surprising
Not
i i …
‹
Valuations are driven by exit markets
♣ IPO and M&A valuations are strongly
gyp
pro-cyclical
y
‹
Large
g capital
p
inflows in yyears after high
g returns
♣ Funds raised in years with large capital inflows have
poor subsequent performance and vice versa
♥ Young funds and funds with worse track record are
especially negatively affected
♥ Established funds are better able to weather industry
cycles
Idi
Idiosyncratic
ti risk
i k matters
tt
‹
The three pillars of public equity markets
♣ Non predictability of returns
♣ Diversification
♣ Arbitrage
‹
All three are violated in Private Equity
♣ Entry limitations
♣ Ticket size
♣ Illiquidity of investments
♣ Large information asymmetries
P i t
Persistence
off performance
f
‹
Bottom
Medium
Bottom
Tercile
61%
22%
17%
Medium
ed u
Tercile
25%
45%
30%
Top
Tercile
Top
High likelihood that the
next funds of a given
partnership stays
sta s in the
same performance
bracket
ÎPersistence
♣ Larger for the top quartile
27%
24%
48%
Kaplan and Schoar, Journal of Finance, 2005
♣ More persistence for top
quartile funds in market
downturns
I di id l ffunds
Individual
d returns
t
All Funds
VC Funds
Buyout
y
IRR
0.12
(0.17)
[ 04 0
[.04,
0.20]
20]
0.11
(0.17)
[ 03 0
[.03,
0.19]
19]
0.13
(0.19)
[ 06 0
[.06,
0.24]
24]
PME
0.74
(0.96)
[ 45 1
[.45,
1.17]
17]
0.66
(0.96)
[ 43 1
[.43,
1.13]
13]
0.80
(0.97)
[ 46 1
[.46,
1.14]
14]
Kaplan and Schoar, Journal of Finance, 2005
‹
‹
‹
The average fund does
nott outperform
t f
the
th S&P
500
Enormous differences in
performance between the
top and bottom quartile
Skewness of returns
Sharp and persistent differences
in LP returns over 1990s
Banks
Ad i
Advisors
Public Pensions
Insurance Companies
Private Pensions
Endowments
-5%
5%
Source: Lerner, Schoar and Wang [2007]
0%
5%
10%
15%
20%
But Idiosyncratic risk is lowest for
high return investors
Value at Risk: % of funds in lowest quartile
Banks
Advisors
Public Pensions
Insurance Companies
Private Pensions
Endowments
0%
5%
Source: Lerner, Schoar and Wang [2007]
10% 15% 20% 25% 30% 35% 40%
I t
Interpreting
ti the
th differences
diff
‹
Do these just reflect timing or investment mix?
♣ No… robust to various controls.
‹
Are these differences due to fund access?
♣ Look at reinvestment process, where access is
much less critical:
♥ Pension funds and advisors tend to invest when current
returns are high.
♥ But much more dramatic difference in future returns from
endowments.
Reinvestment and returns
40
30
IRR of next
fund
20
10
0
Advisors
Banks
Corporate
Pensions
Reinvested
Endowments
Public
Pensions
Did Not Invest
40
30
IRR of previous
fund
20
10
0
-10
Ad i
Advisors
B k
Banks
Reinvested
Corporate
C
t
Pensions
E d
Endowments
t
Did Not Invest
P bli
Public
Pensions
What do endowments know
that others don’t?
Suggests differences in ability to identify
or act on inside information
information.
‹ Systematically obey a few simple rules
that can be observed by most investors
‹
♣ Monitor fund size and partnership growth
Fund Size
‹
Relation IRR and Fund Size
14
12
‹
10
IRR
8
6
‹
4
2
0
1
2
3
4
5
6
7
Fund size in $100 m illion
8
9
10
11
Concave relationship
between IRR and fund
size
Fund size is measured as
capital committed at
closing
Regression results control
for vintage year, fund
category
Change
g in Fund Size
‹
Relationship of Change in IRR to
Change in Fund Size
0
-20
1
2
3
4
5
6
‹
-40
40
-60
‹
-80
-100
100
Logarithm of Size
Negative relationship
between change in IRR
and change in fund size
for a given firm
Fund size is measured as
capital committed at
closing
l i
Regression results control
for vintage
g yyear effect,
fund category, and firm
fixed effects
Fund Sequence
q
Number
‹
IRR and Fund Sequence Number
25
‹
20
IR R
15
‹
10
5
0
1
2
3
4
5
6
7
Sequence Number
8
9
10
11
Positive relationship
between IRR and fund
sequence number
Fi t time
First
ti
funds
f d perform
f
especially bad
Regression results control
for vintage year effect,
fund category and fund
size
Partner to Size Ratio
‹
IRR and Partner to Size Ratio
30
25
20
IR R
‹
15
10
5
0
0
0.2
0.4
Number of Partners to $100 million in committed capital
0.6
Positive relationship
between IRR and the ratio
of partners to committed
capital
Regression results control
for vintage year effect
effect,
fund category and fund
size
Summary
‹
PE differs from other asset classes
♣ Vast heterogeneity across funds
♣ Strong persistence of returns
♣ Top quartile funds have less idiosyncratic risk than the rest of the
funds
‹
Focus on relationships with top funds not ex ante asset
allocation
♣ Manage investment process opportunistically
♣ Bottom-up approach rather than fixed asset allocation targets
♣ Incentives for deployment of capital are detrimental to returns
Th k you!!
Thank