1. Modelling Redemption
Risk in Funds of Hedge
Funds
Drago Indjic
Hedge Fund Centre
6 November 2008, Quant Europe
Indjic – 6 November 2008 Page 1
2. Overview
• Origin of the study: hedge fund data quality surveys
• Fund of hedge funds products
o Inaccurate redemption terms
o Unenthusiastic industry response in 2007
• Autumn 2008
o Rapid shrinkage of hedge fund industry
• Market mismatch
• Discussion
Indjic – 6 November 2008
3. Is is too late to save FoHF?
• Quant Europe 2003: hedge fund risk factors
• Today: liquidity
• Quant tools became irrelevant in new regime
Indjic – 6 November 2008
4. Fund of Hedge Funds Products
• Distribute over 40% of HF assets ($0.8t)
o Regulatory artefact: exclusive retail distribution (risk)
o Not “absolute return” - losing money and AUM
o Not agile, almost passive exposure to Beta factors
o ~$20b in fees shared between >1000 products
o Over-concentration: AUM of Top 50 FoF firms > $450b
yet limited capacity
• We know about their Beta - what about liquidity risk?
Indjic – 6 November 2008
5. IOSCO – June 2008
• “The provision of a regulatory requirement that there be a
real consistency between a fund of hedge funds’ liquidity
and that of its underlying hedge funds (..) It is noteworthy
that in the opinion of a few experts, the funds of hedge
funds’ liquidity is not always an issue and may be a
means to protect retail investors’ interests”
• Protecting FoHF from “fund runs”? Or clients?
Indjic – 6 November 2008
7. October 2008
• Fund is legal entity with contractual liquidity obligations
• “Pre-emptive redemptions”
o FoHF have put redemption notices anticipating the
redemptions from their retail clients or as requested by
leverage providers
o Hedge funds have responded
“gates”, fund restructurings, liquidating accounts, in-
kind redemptions etc
• In many cases redemptions suspended
Indjic – 6 November 2008
8. FoHF - What If?
Assets Liabilities
If (with non-zero probability)
HF 1 FoF Class A Net Flows <0 D(Assets)>D(Liabilities)
HF 2 And
Shock
… Then
… Liquidity crisis
Reputation risk
… FoF Class B Possibly
… insolvency → bankruptcy
…
HF n-1 FoF Class C
HF n
Indjic – 6 November 2008
9. No Big Picture
• Firms focus on currently invested funds only
o Internal data rarely reconciled with external data
sources; private Placement Memorandums not
accurately mapped
o No entity is responsible for reference data
• Fragmented, delayed and contradictory data universe
• London Business School and Imperial College London
have better data analytic systems than industry
o Multi-year academic research partnerships produced
tailor made software tools
Indjic – 6 November 2008
10. Risky Data
• Aggregated, multi vendor data
base
o Macro (#management companies; %
Strategies) and micro data (flows,
%counterparties)
o “According to X, Y and Z vendors”
• Leading “Rosetta stone” created at
LBS using tailor made Soft
Finance tools (since 2005)
Indjic – 6 November 2008
11. Status and Strategy Data
• “Alive” or not?
• Doing exactly what?
• Hedge fund index “averages” do not make sense
Indjic – 6 November 2008
12. Data quality
• Redemption conditions
o Notice/Redemption Frequency/Lockup periods
o Inaccurate and insufficient: >10% fields are blank
o No FoHF holdings or strategy allocation data
• Visualise stylised data
o Simple “duration” = notice + redemption frequency
o Mature FoHF portfolios – outside lock-up
Indjic – 6 November 2008
13. HFR Database Example
• Most FoHF are designed around long/short equity
Indjic – 6 November 2008
14. CISDM Database Example
• As in HFR it appears that “durations” are matched
Indjic – 6 November 2008
17. Strategies & Durations
• FoHF median “duration” is 97 days(14 days mismatch)
but FoHF <> “market”
no standard hedge fund strategy classification, de-funct
AIMA committee (2003)
Indjic – 6 November 2008
22. A Real Case Study (t)
Indjic – 6 November 2008
23. Liquidity Stress
• Unexpected funding risk of FoHF – beyond investor redemptions
%AUM Loss Event and remedy
<10% Redemptions were normally offset by
subscriptions
10% Covered by portfolio cash or credit facility
25% Limits of (expensive) cash facility
50% Leverage halved at daily notice
100% Liquidation – Board request
Indjic – 6 November 2008
24. Instant Liquidity
• Cash, marketable assets or robust facility
• (Re)action preferences
o (i) Credit facility / Leverage
o (ii) Activate gates
o (iii) Sell most liquid funds
o (iv) Suspend redemptions
o (v) Sell funds on Hedgebay
o (vi) Seek parent (bank) help
o (vii) Restructure or close fund
Indjic – 6 November 2008
25. Fair Deal for Beta?
○ β(“first out”) ○ β(all others)
Indjic – 6 November 2008
26. Shipping
• Navigate in the long shadow of illiquidity
risk and optimisation tools are irrelevant
Indjic – 6 November 2008
27. Impact
• FoHF – unrolling funding crisis
• Investors - Portfolio rebalancing “agility”
o Strategies: δ(weights)/δ(time) → 0
o => weights * risk_factors ~ constant for many months
o Quality of portfolio – loss of control over Beta
• Market – collateral damage of forced selling, influencing
many other funds and markets
Indjic – 6 November 2008
28. FSA – September 2008
• “What has been highlighted for individual funds is the
importance of monitoring redemption risk profiles and
managing mismatches between funding/notice periods
and the underlying liquidity of funds”
Hector Sants, 17 Sep 2008
Indjic – 6 November 2008
29. Quantitative Methods
• Formal language to describe redemption
• Simultaneous risk and liquidity optimisation
o “cheapest, fastest Beta”
o Partial and optimal liquidation vs risk transfer
• Stochastic analysis and optimisation
Cash flow timing and fees scenario analysis
Contingency finance and collateral management
Indjic – 6 November 2008
30. Review
• The end of traditional FoHF business model
o Focus on [Liquidity, Beta, Costs] products
• Lessons
o Extensive stress testing necessary
o Audit contingency asset redemption schedules: NAV
liquidity adjustments in stressed conditions
o Alpha/Beta separation: liquid (possibly synthetic
“clones”) and illiquid Alpha satellites
• Future of open ended, listed products investing in illiquid
funds – private equity “capital call” model?
Indjic – 6 November 2008