Under One Roof: An Study of Simultaneously Managed Hedge Funds (HF) and Funds of hedge - PowerPoint Presentation

Slide1

Under One Roof: An Study of Simultaneously Managed Hedge Funds (HF) and Funds of hedge funds (FOF)

Sugata

Ray (University of Florida)

Joint work with

Vikas

Agarwal (Georgia State University)

Yan Lu (University of Florida)Slide2

Motivation I

What is simultaneous management?

A HF and a FOF managed by the same parent company are defined as being simultaneously managed

Anecdotal evidence that prominent HFs and FOFs engage in the practice

David

Einhorn

of

Greenlight

Capital and

Greenlight

Masters

Dubin

&

Swieca

and

Highbridge

Capital (now bought by JP Morgan, but D&W still active)Slide3

Motivation II

¼ of all USD AUM

in HF/FOF space today is simultaneously managed – and this seems to be increasingSlide4

Research Questions

How prevalent is simultaneous management of HFs and FOFs?

What are some of the determinants of simultaneous management?

What are some of the effects of simultaneous management?

Does it matter if a FOF company starts HFs or vice versa? Slide5

Hypotheses (HF  FOF)

H1. There are

o

pportunities for value creation through simultaneity

Value creation for the HFs through

Discussing ideas with portfolio funds of FOFs

Tighter control on flows to manage any diseconomies of scale

Increased access for newly created FOFs

H2. Simultaneous FOF is an opportunity for added fees

HFs with good returns want to increase fees (similar to Agarwal and Ray (2012))

Doing it through FOFs allows greater control of flows than simply raising fees or starting another HF with higher fees (may also allow value creation)

H3. Simultaneous management leads to agency problems

Strategic diversion of flows across internal portfolio HFs by FOF managed HFs to prop up poorly performing HFsSlide6

Hypotheses (FOF  HF)

H1. Starting a HF allows value creation

Having a FOF allows better networks with other HF managers (more access)

The newly started HF will be able to discuss ideas with HFs in FOF’s portfolio

H2. Having a HF allows a ready source of supply (vertical integration)

HFs are essentially suppliers to FOFs

Having your own HF allows you to have your own supply

H3. Starting HF is an opportunity for added fees

Higher fees on HF business (2/20 versus 1/10)Slide7

Contributions to relevant literature

Managerial or organizational structure in asset management industry

Side-by-side management

Cici, Gibson, and

Moussawi

(2010),

Nohel

, Wang, and

Zheng

(2010)

Multitasking by mutual fund managers

Agarwal and Ma (2012) and Choi, Kahraman, and Mukherjee (2012)

Multiple offerings within fund families

Nanda, Wang, and

Zheng

(2004) – star funds

Gaspar, Massa, and Matos – cross-fund subsidization

Team management

Baer, Kempf, and Ruenzi (2011)

Bliss, Potter, and Schwarz (2008)

Massa, Reuter, and

Zitewitz

(2010)Slide8

Relevant literature (cont’d)

Fund of mutual funds

Liquidity insurance for other funds in the family

Bhattacharya

, Lee, and Pool

(2013)

Investing in poorly performing and high-fee funds

Sandhya (2012

)

Funds of hedge funds

Underperform hedge funds due to the second layer of fees

Amin and Kat (2003) and Brown,

Goetzmann

, and Liang (

2004

)

Better managers going to multi-strategy funds

Agarwal

and Kale (2007

)

Overweighting in local hedge funds can help them perform better

Sialm, Sun, and

Zheng

(2012)

Add value through effective monitoring but not fund selection

Aiken, Clifford, and Ellis (2012

)Slide9

Data

TASS HF and FOF data from 1994:01 to 2011:12

Management firm data used to determine simultaneous management of FOFs and HFs

Both cases of HFs first and FOFs later and vice versa

Performance and flow data used for analysis

Registered FOF holding data

Examine holdings for evidence of investing in simultaneously managed HFsSlide10

Summary statistics (HFs only)

SHF (2,928)

non-SHF

(8,245)

Difference

Variable

Mean

Mean

Mean

T-test

Management Fee

1.53

1.50

0.03

2.17***

Incentive Fee

15.36

17.92

-

2.56-16.14***High-water Mark0.450.65-0.20-19.11***Lockup Period (years)1.533.16-1.63-11.79***Log Size (Inception)15.6115.300.314.76***

SHFs (especially FOF -> SHF) have lower

inc

Fee, HWM, lockupSlide11

Summary statistics (FOFs only)

SFOF (2,843)

non-SFOF (

3,333)

Difference

Variable

Mean

Mean

Mean

t-test

Management Fee

1.36

1.37

-0.01

-0.64

Incentive Fee

8.25

7.08

1.17

5.85***

High-water Mark0.450.440.010.48Lockup Period (years)1.121.080.040.31Log Size (Inception)15.23

15.44

-

0.21-2.15***

Both FOFs and SFOFs are generally similar, except

SFOF (especially HF -> SFOF) have higher

inc

fee & smaller sizeSlide12

Summary statistics (MFs only)

SMFs are larger and have

more funds than non-SMFs

(all differences significant)

SMF (461)

non-SMF

(

4,093)

Difference test

Variable

Mean

Mean

(1)-(2)

Average Owned

11.00

2.50

26.54***

No. of Hedge Funds

5.34

1.7816.35***No. of FOFs5.660.7222.93***Total Log Size (Inception)70.8326.8221.07***Slide13

Strategies – SHF vs Non-SHF

Primary Category (percent)

SHF

non-SHF

SHF

start as HF

SHF

start as

SHF

Convertible Arbitrage

1.95

2.64

3.77

1.37

Dedicated Short Bias

0.2

0.59

0.29

0.17

Emerging Markets7.559.413.488.52Equity Market Neutral5.056.216.094.84Event Driven4.997.885.074.75Fixed Income Arbitrage4.783.587.684.11Global Macro7.457.749.138.22

Long/Short Equity Hedge

22.71

6828.41

20.89

Managed Futures

7.17

10.35

13.19

5.52

Multi-Strategy

33.64

10.02

18.99

37.03

Options Strategy

0.14

0.51

0.29

0.09

Other

4.37

3.85

3.62

4.49Slide14

When did simultaneity occur? Mid to late 2000sSlide15

How many simultaneous funds among all newly incepted funds?Slide16

How many non-simultaneous funds become simultaneous each year?Slide17

Fixing timings and notationSlide18

Methods

How prevalent is simultaneity?

Counts and AUM-weighted percentages

Determinants of becoming simultaneous

Comparing A versus B (what are the differences between funds that become simultaneous and those that don’t?)

Effects of becoming simultaneous

Comparing (C-A) to (D-B)

Matched sample analysis, matching A and B

Comparing E to F (how the newly started simultaneous funds do compared to other newly started non-simultaneous funds?)Slide19

Linking empirical tests to hypothesesHF  FOF

No.

Hypothesis

Implication

I

1a

Value creation for the

HFs through

discussing ideas with portfolio funds of

FOF

Returns: C-A > D-B

1b

Value creation for the

new FOF

through

access

Returns: E > F

2a

Starting a FOF is an opportunity for added fees

Superior

returns are a determinant of simultaneity2bThe FOF structure allows tighter control of flows to HFClosing to new investment around s-dateFlows: C-A < D-B3Strategic diversion of flows across internal portfolio HFs by FOFsFlow-performance: C-A < D-BSlide20

Linking empirical tests to hypothesesFOF -> HF

No.

Hypothesis

Implication

I

1a

Value creation for the

FOFs through better access

Returns: C-A > D-B

1b

Value creation for the

new HF through idea

discussion

Returns: E > F

2

Having a HF allows ability to ability

to deploy capital without external suppliers

High flows

are a determinant of simultaneity

3

Starting a HF is an opportunity for higher/added feesSuperior returns are a determinant of simultaneitySlide21

Results: Simultaneous management is quite prevalent

Simultaneous management of HF and FOFs appears to be relatively prevalent

26% of HFs

46% of FOFs

10% of Management Firms (MFs)

¼ of all USD AUM today

Anecdotal evidence that prominent HFs and FOFs engage in the practice

David

Einhorn

of

Greenlight

Capital and

Greenlight

Masters

Dubin

&

Swieca

and

Highbridge

Capital (now bought by JP Morgan, but D&W still active)Slide22

HF

FOF

SMF HF 1

st

SMF FOF 1st

Returnt-1,t-24

0.253***

0.248**

0.128

0.470***

(3.888)

(2.150)

(1.474)

(3.572)

Flowt-1,t-24



1.593

4.615***



2.0983.431**(0.681)(8.603)(0.641)(2.284)Domicile Dummy1.009***0.503**(5.210)(2.419)Management Fee0.1120.431***(0.893)

(4.468)

Incentive Fee

0.018



0.055***

(



1.151)

(



4.174)

High-water Mark

0.152

0.312*

(0.845)

(1.847)

Share Equalization Method



0.473

0.499*

(



1.451)

(1.909)

Leveraged

0.0950.034(0.554)(0.259)Lockup Period0.0020.036**(0.122)(2.244)Log Assets0.0230.0560.0140.051(0.495)(1.479)(0.295)(0.651)Age0.0040.003*0.019***0.007***(1.630)(1.682)(6.809)(2.829)Funds Managed0.085***0.080***(6.355)(8.923)Strategy DummiesYesNoNoNoYear DummiesYesYesYesYesR-squared0.0460.0430.0920.069N242,92498,052137,00532,523

Logisticregression1 = s-date,0 = other

Non-US FundsMore Likelysimul-taneous

Returns



simul

-

taneity

Flows simul-taneityFor FOF onlySlide23

Effect of simultaneity: Matched sample analysis to control for mean reversion

C-A versus D-B

E versus F

Time horizon

12 months

24 months

6-24 months (we use

this, results are robust)

Match on broad strategy, size, returns, and flows for flow analysis, and starting time and strategy only for E versus F

Similar analysis for HFs, FOFs, MF-HF first, MF-FOF first (MFs don’t have the E versus F analysis)Slide24

Format of univariate analysis table

Variables:

Returns or Flows

Mean (

C/D)

Mean (

A/B)

t-stats

C-A

Treated funds after

treatment (C)

Treated funds before treatment (A)

C versus A

t

-stat

D-B

Matched

sample after treatment (D)

Matched

sample before treatment (B)D versus B t-statDIDDifference for treated fund (C-A)Difference for matched sample (D-B)(C-A) versus (D-B) t-statE-FNewly started simultaneous funds after treatment (E)Control group for newly started simultaneous funds (F)E versus F t-statSlide25

Effects of simultaneity (hedge funds -> FOFs): matched sample univariate analysis on returns and flows

Flows

Mean (

C/D)

Mean (

A/B)

t-stats

C-A

-0.0005

0.0460

-8.1691***

D-B

0.0184

0.0459

-4.3865***

DID

-0.0465

-0.0275

-4.5266***

E-F

0.04660.04100.9563ReturnsMean (C/D)Mean (A/B)t-statsC-A0.00590.0087-5.2077***D-B0.00480.0087-5.2490***DID

-0.0028

-0.0039

1.6293

E-F

0.0047

0.0044

0.4874

Alphas

Mean (

C/D)

Mean (

A/B)

t-stats

C-A

0.0075

0.0088

-2.2600***

D-B

0.0065

0.0088

-3.3105***

DID

-0.0013

-0.0023

1.3510

E-F0.00720.00611.5050Sharpe RatioMean (C/D)Mean (A/B)t-statsC-A0.59530.7602D-B0.4560.7529DID-0.2972.6223***E-F0.53360.28373.6241***Strong support for dampened inflows into SHFs after s-date Weak support for better performance for SHF and SFOF after s-dateSlide26

Effects of simultaneity (FOFs -> hedge funds): matched sample univariate analysis on returns and flows

Flows

Mean (

C/D)

Mean (

A/B)

t

-stats

C-A

-0.0028

0.0422

-9.4424***

D-B

0.0067

0.0395

-8.1133***

DID

-0.0450

-0.0328

-3.4461***

E-F0.04060.0500-1.6885*Sharpe RatioMean (C/D)Mean (A/B)t-statsC-A0.3450.4533D-B0.46160.44590.4615DID



0.0157

E-F

0.6784

0.4752

1.8297*

Alphas

Mean (C/D

)

Mean (

A/B)

t

-stats

C-A

0.0035

0.0055

-4.6577***

D-B

0.0061

0.0055

1.3653

DID

0.0020

0.0006

-5.4674***

E-F0.00730.0109-2.9058***ReturnsMean (C/D)Mean (A/B)t-statsC-A0.00140.0049-6.5282***D-B0.00520.00490.7012DID-0.00340.0003-7.2712***E-F0.00530.0105-6.4190***FOFs becoming simultaneous is bad all aroundReturns and flows are lower compared matched sample for both existing FOFs & for the newly started HFs (every relevant t-stat is significantly negative)Slide27

Effects of simultaneity matched sample univariate

analysis – MF (starts HF first)

Flows

Mean (

C/D)

Mean (

A/B)

t

-stats

C-A

0.0319

0.0403

-1.0645

D-B

-0.0001

0.0413

-5.0114***

DID

-0.0111

-0.0417

4.4625***Sharpe RatioMean (C/D)Mean (A/B)t-statsC-A0.5710.7162D-B0.16630.7394DID5.1487***

Alphas

Mean (

C/D)

Mean (

A/B)

t

-stats

C-A

0.0065

0.0083

-1.2120

D-B

0.0032

0.0083

-3.6594***

DID

-0.0018

-0.0051

2.5047***

Returns

Mean (

C/D)

Mean (

A/B)

t

-statsC-A0.00630.0078-1.3124D-B0.00150.0078-4.8994***DID-0.0015-0.00633.3626***MFs with HFs get benefits from starting FOFsReturns and Alphas better compared to matched sampleEven combined flows better compared to matched sample (due to the added FOFs)Slide28

Effects of simultaneity matched sample univariate

analysis – MF (starts FOF first)

Flows

Mean(C/D)

mean(A/B)

t-stats

C-A

0.0212

0.0357

-1.8429*

D-B

0.0138

0.0353

-2.6669***

DID

-0.0143

-0.0206

1.0771

Sharpe Ratio

Mean(C/D)

mean(A/B)t-statsC-A0.52670.7532D-B0.57650.7488DID

Alphas

Mean(C/D)

mean(A/B)

t-stats

C-A

0.0031

0.0061

-2.3479***

D-B

0.0070

0.0062

0.5927

DID

-0.0030

0.0008

-2.6053***

Returns

Mean(C/D)

mean(A/B)

t-stats

C-A

0.0038

0.0055

-1.6574*

D-B

0.00780.00552.1218***DID-0.00170.0023-3.5151***MFs with FOFs are actually hurt by starting HFsReturns and Alphas worse compared to matched sampleSlide29

(C-A) vs (D-B)Y

after

-Y

before

=



s

dummy

+ controls + 

(only  presented)

Sample

Y = Flows

Y =

Returns

Y =

Alpha

Y = Sharpe Ratio

HF

−1.583*

0.0870.08914.144***(1.805)(0.887)(0.858)(3.470)FOF0.7370.371***−0.124*10.953**(1.019)(5.283)(1.956)(2.562)Univariate analysis is confirmed in multivariate regressionsControls include: (1) incentive and management fees charged by the funds, (2) other compensation contract details, such as lockup length, use of the high-water mark, etc. (3) size of the fundFor HFs becoming simultaneous, flows are dampened after becoming simultaneousFor FOFs becoming simultaneous, returns and alphas are worse after becoming simultaneousSlide30

E vs. FYafter

=



s

dummy

+ controls + 

(only  presented)

Y = Flows

Y = Returns

Y = Alpha

Y = Sharpe

Ratio

New

FOF

0.998

0.082

0.126

-11.180

(1.524)

(1.356)(1.316)(-1.377)New HF0.095-0.308***-0.333***-13.478(0.165)(3.417)(7.988)(-1.443)For newly started FOFs by HFs, flows are higher compared to other non-simultaneous newly started FOFsFor newly started HFs by FOF, returns and alpha worse compared to other non-simultaneous newly started HFsSlide31

(C-A) versus (D-B)Yafter



Y

before

=



s

dummy

+ controls + 

(only  presented)

Sample

Y = Flows

Y =

Returns

Y =

Alpha

Y = Sharpe

Ratio

MF

HF first3.183**0.538***0.506**36.148*(2.562)(2.903)(1.996)(1.781)MFFOF first0.744−0.324**−0.352*−2.076(0.577)(−2.108)(−1.698)(−0.161)Similar findings when examining the effects of simultaneity on the management firmFor MFs with only HFs, starting a FOF is associated with improved returns and flowsFor MFs with only FOFs, starting a HF is associated with decreased returns and alphasSlide32

Summary so farHF  FOF

 FOF

 HF



HFs start FOFs which do well; HFs themselves

do weakly better

FOFs start HFs which do less well; FOFs themselves do worse

In both cases, flows decrease for original funds (significant for HFs, insignificant for FOFs)

for HFs, newly started FOFs have higher than expected flows

for FOFs, flows into the started HFs in line with other

H

Fs started at the same time

Mgmt. firm results echo these broad themesSlide33

H2. HFs want to add another layer of fees through FOFs and close to new investment: Further analysis on closing to new investment

Look for evidence of closing to investment around

s

-dates

Overall HF “closed to new investment” rate = 5.8%

SHF “closed to new investment” rate = 10.8% (significantly greater than 5.8%)

32% of the closing dates are within 6 months of s-date

Consistent with dampened flows to the HFs after the

s

-dateSlide34

H3. What about the flow-performance relation before and after simultaneity?

Regression: comparing periods A, B, C and D



5

is the DID estimate of SHF and non-SHF before and after the

s

-date

(1)

(2)

Return t-13, t-24

0.771**

0.734**

(2.183)

(2.133)

Return*After

-0.515

-0.499

(-1.352)(-1.348)

Sdummy

0.118

0.082(0.802)

(0.579)

Return* Sdummy

-0.653

-0.782

(-0.720)

(-0.860)

Return*Sdummy*After

1.274

1.509*

(1.566)

(1.869)

Age

-0.004***

(-4.816)

R-squared

0.014

0.038

N

532

532

Actually see increased flows performance sensitivity for simultaneous funds afterSlide35

H3. What about the flow performance relationship before and after simultaneity?

Regression: E

versus F



2

is the DID estimate of SHF and non-SHF after

the

s

-date

(3

)

Match on same

inception date

(4

)

Match on inception

date and broad category

Return t-13, t-242.949***3.475***(6.821)(3.919)Return* Sdummy-1.197*-1.350(-1.694)(-1.121)R-squared0.0960.105N292152

Weak evidence of newly started HF by FOF having less flow performance sensitivitySlide36

Linking empirical tests to hypothesesHF  FOF

No.

Hypothesis

Implication

I

1a

Value creation for the

HFs through

discussing ideas with portfolio funds of

FOF

Returns: C-A > D-B

1b

Value creation for the

new FOF

through

access

Returns: E > F

2a

Starting a FOF is an opportunity for added fees

Superior

returns are a determinant of simultaneity2bThe FOF structure allows tighter control of flows to HFClosing to new investment around s-dateFlows: C-A < D-B3Strategic diversion of flows across internal portfolio HFs by FOFsFlow-performance: C-A < D-BSlide37

Linking empirical tests to hypothesesFOF -> HF

No.

Hypothesis

Implication

I

1a

Value creation for the

FOFs through better access

Returns: C-A > D-B

1b

Value creation for the

new HF through idea

discussion

Returns: E > F

2

Having a HF

allows ability to control

capital deployment

High flows

are a determinant of simultaneity

3Starting a HF is an opportunity for higher/added feesSuperior returns are a determinant of simultaneitySlide38

ConclusionsHFs starting FOFs create value and share in the value they create through second layer of fees

FOFs destroy value when they start HFs

Why do they continue doing this?

Overconfidence? Greed? Opportunism?

Minimal evidence of agency problems, such as flow diversion

Overall, this practice seems benign/beneficial, except for investors investing when FOFs start HFs