Quantitative Methods in Investment and Risk Management 2007 09 20 Leo de Bever Chief Investment Officer Victorian Funds Management Corporation Victorian Funds Management Corporation
Main Points Superannuation funds are growing rapidly: how well are we managing long-term assets? Short-term evaluation horizons are an impediment to rational decision-making Investing long-term in a short-term world: what does it take to be exceptional Quantitative risk management can help deliver excellent results to our clients Alternative investments: superior returns or just poorly defined risks? 2
Global Retirement Savings Estimated global superannuation fund assets (DB & DC) and growth rates Country US$bn 2005 1 year 6 year 10 year Per Capita 2005 2000-2005 1995-2005 Assets Growth Growth p.a. Growth p.a Switzerland 464 62 20% 11% 11% Netherlands 764 46 20% 13% 12% Canada 1,022 31 14% 7% 10% Australia 592 29 17% 14% 14% US 8,123 27 7% 2% 6% UK 1,621 27 20% 6% 8% Japan 3,235 25 22% 6% 6% Ireland 90 22 21% 14% 15% Hong Kong 49 7 5% 9% 9% Germany 287 4 19% 10% 7% France 133 2 19% 11% 9% Source: Watson Wyatt, various sencondary sources 3
It is Hard to Do Better than Market Returns 280% 260% 240% 220% VFMC Passive Benchmark +1% VFMC Passive Benchmark VFMC Actual net of 6% effective tax and fees Intech upper quartile Intech median growth fund 200% 180% 160% 140% 120% 100% Jun-97 Jun-98 Jun-99 Jun-00 Jun-01 Jun-02 Jun-03 Jun-04 Jun-05 Jun-06 Jun-07 4
What it Takes to be Exceptional Independent governance model Strong board of trustees under prudent man model Long term investment horizon Willingness to invest in unusual opportunities Cost-effective mix of internal and external mgt Internal staff must be paid on commercial term Strong portfolio and risk systems Doing the basics better Mgt focus on maximising return/risk Management language for contrasting opportunities 5
Balancing Internal and External management Ontario Teachers VFMC Current VFMC in 3 yrs Investment staff Salaries ~2 bps ~1 bps ~2 bps STI and LTI in bps of assets ~10 bps ~1 bps ~4 bps External Fees (includes embedded fees) ~3 bps ~35 bps ~18 bps Total Investment Costs ~15 bps ~37 bps ~24 bps % managed externally ~10-15% ~100% ~50% IT ~3 bps ~0.5 bps ~3 bps Rent, Legal, Financial, Compliance, Custody, etc ~4 bps ~4bps ~4bps Total Investment cost bps ~22bps ~42bps ~31bps Value added/assets last 5 yrs ~4% none ~1% Assets Under Mgt AUD$110B AUD$40 AUD$60 6
Simple Quant: Doing Basics Better The easiest way to make money is not having to spend it Not lowest cost, but aiming for best return/cost Make sure passive index benchmarks are consistent with liability funding objectives Benchmarks motivate behaviour: they separate active (alpha) from passive (beta) returns Managing cash effectively Exploiting economies of scale Eliminate unnecessary trading May be as high a $50 million/year at VFMC 7
Improving Efficiency with Derivatives Can be far more cost-efficient than buying and selling physical securities Must be supported by Board, senior management, corporate culture Must be supported by strong accounting and risk control systems Must address counterparty credit risk Do not buy what you do not understand 8
Motivating Better Asset-Liability Management Objective before the fact: Maximise long-term (asset growth) (liability growth) within a risk tolerance limit What gets measured after the fact: Annual return on assets Investment Heisenberg uncertainty principle : How outcomes are measured affects strategy direction Managers respond to how they are evaluated Will factor in career risk in case of conflict with long-term goals 9
Investment Management is Risk management All investments have return on risk Only some have return on assets Same return/assets can imply very different return/risk What ultimately matters is marginal return/risk Includes the effects of correlation and diversification Must consider all active and passive risks together Assets allocation imperfect way fix risk Same physical asset can have very different return/risk depending on contracts and gearing 10
Investment Optimization Process Determine the nature of the liabilities Typically looks like some kind of nominal or index-linked bond Find the closest risk free asset match to liability 100% match would eliminate any shortfall from capital target The ideal long-dated asset looks like a bond Pursue higher return by investing in risky assets typically index exposure to listed equities Accepting risk of a shortfall because of poor returns Governance Issues: How much investment risk is acceptable? Is that risk earning an appropriate return? 11
Understanding Downside Risk VFMC Risk Profile at $40 Billion 1 in 6 Worst Outcome 1 in 100 Worst Outcome Annual 4000 10.0% 12000 30.0% Monthly 1155 2.9% 3464 8.7% Weekly 553 1.4% 1658 4.1% Daily 250 0.6% 750 1.9% Hourly 88 0.2% 265 0.7% 12
Expected Return on Passive and Active Risk 4 Asset Growth Liability Growth 3 2 1 0-1 Return from Active Risk 25% Stocks 50% Stocks Total Risk Tolerance Or Risk Budget Return from Passive (Market Index) Risk 100% Stocks -2 0 2 4 6 8 10 12 14 16 18 20 Volatility of Assets/Liabilities If the active return on risk is realistic, a larger active risk allocation is warranted 13
Traditional Split of Total Risk Management Risk total = (Risk 2 passive + Risk2 active + Risk passive x Risk activex Correlation x 2) Passive risk: funding loss from poor markets Fixed by policy asset mix managers are not responsible Assumes stable correlations, volatility, efficient markets Active risk: falling below passive market benchmarks Poor result from attempt to add value by deviating from passive asset mix or security selection within asset class Benchmarks determine active, passive risk split Benchmarks should serve portfolio objectives They are typically selected for ease of judging managers 14
What s Wrong with that Model? Passive risk is large, variable, and neglected Market indices have changing volatility Trend return on passive risk variable due to mean reversion Active risk is small and may not add to total risk Tracking error is not the same as incremental risk Separate mgt of active, passive risk suboptimal No incentive to am for higher total return with less total risk Volatility in constant passive risk is larger than marginal contribution of active risk to total risk Why concentrate on managing noise? 15
Passive Risk is Variable (Vix: S&P volatility) 50 45 40 35 30 25 20 15 10 5 0 16 Jan-90 Jan-91 Jan-92 Jan-93 Jan-94 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05 Jan-06 Jan-07
Managing The Right Risks (with real liabilities) CPI Sensitive Liability Risk CPI sensitive Asset Risk + + Equity Asset Risk + Active Risk = Diversified Asset Liability Risk Is almost linear in equity risk 17
Asset Mix Does Not Fix Passive Risk Risk as 1% Worst Outcome 22.0% Passive Risk 19.0% Equity Benchmark Weight reduced from 65% to 60% Total Risk Equity benchmark Weight reduced from 60% to 50% 16.0% Oct-98 Feb-99 Jun-99 Oct-99 Feb-00 Jun-00 Oct-00 Feb-01 Jun-01 Oct-01 Feb-02 Jun-02 Oct-02 Theory says passive risk line should be flat between asset mix changesc Perception is that total risk should always be higher than passive risk 18
Active Risk Negatively Correlated with Passive Risk 0.4 0.2 0-0.2-0.4-0.6-0.8-1 Dec-93 Jun-94 Dec-94 Jun-95 Dec-95 Jun-96 Dec-96 Jun-97 Dec-97 Jun-98 Dec-98 Jun-99 Dec-99 Jun-00 Dec-00 Jun-01 Dec-01 Jun-02 Dec-02 Jun-03 Dec-03 Intuitively: good managers select assets that have higher return on risk than passive index 19
Asset/Liability Mgt Incorporates Joseph Effect 30% 20% Change in Funding Ratio 10% 0% -10% -20% -30% Fat Years 0 1 2 3 4 5 6 7 8 9 10 If the expected return on risk just meets long run needs Any surplus improvements will be transitory Lean Years Trend + Cycle Trend Years Fat years make up for lean years 20
Long-Term Funding Ratio with 50% stocks Assuming Return on Risk Is Realised 160% 140% 120% 100% 80% 60% 1927 1932 1937 1942 1947 1952 1957 1962 1967 1972 1977 1982 1987 1992 1997 2002 Starting at 130% in 1927, and without withdrawing funds or increasing benefits, The funding ratio oscillates (wildly) around 100% in the long run 21
30 Year Change in Assets/Liabilities: Fat Years Balance Lean Years 200% 30 Year Historical Funding Ratio Simulation (Using Data from 1926-2002) 50% Stocks, 20% Bonds, 30% Real Return Assets, Liabilities with 20 year real duration valued at Risk Free Rate + 0.5% 180% 160% Funding Ratio 140% 120% 100% 80% 60% 40% 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Year No change in contributions or benefits 22
Actuarial Valuation : Every Year is Another Year 30% Change in Funding Ratio 20% 10% 0% -10% -20% Actuarial valuations assume that at any time Investment gains are as certain as a checking account balance Assets in excess of liabilities are surplus Fat Years 0 1 2 3 4 5 6 7 8 9 10 Just Bad Years Years -30% Trend + Cycle Trend Creates a long-term bias to being under funded 23
30 Yr Change in Assets/Liabilities: Surplus over 110% is Withdrawn 200% 30 Year Historical Funding Ratio Simulation (Using Data from 1926-2002) 50% Stocks, 20% Bonds, 30% Real Return Assets, Liabilities with 20 year real duration valued at Risk Free Rate + 0.5% 180% 160% Funding Ratio 140% 120% 100% 80% 60% 40% 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Year No change in contributions or benefits 24
Risk Based Asset-Liability Management Conceptual split into passive and active return Passive: what a representative index would deliver Active: incremental return from security selection Benchmark index selection determines the split Absolute return strategies can never dominate Pension managers own a large part of assets in that market: they cannot collectively be market neutral Management must focus on what matters Avoiding distraction, maintaining consistency with objective Finding asset classes with superior return/risk Building expertise for superior security selection 25
Shiller: Starting Conditions Matter 25% Shiller's P/E vs Future 10-Yr S&P Real Total Return Data: 1871-2004 Annualized Real Total Return 20% 15% 10% 5% 0% Feb 2005 Shiller's P/E = 28 Implied Real T. Return = -2% Historical Range -2% to 2% -5% -10% Real Total Return = 16.5-0.66 * (Shiller's P/E) R-squared = 32%, S.D. = 5% 5 10 15 20 25 30 35 Shiller's P/E The very case for equities (mean reversion of returns) suggests that passive mgt is suboptimal 26
Management Perfection not always obvious 120% 100% 80% 60% 40% 20% 0% 27 Jan-56 Jan-58 Jan-60 Jan-62 Jan-64 Jan-66 Jan-68 Jan-70 Jan-72 Jan-74 Jan-76 Jan-78 Jan-80 Jan-82 Jan-84 Jan-86 Jan-88 Jan-90 Jan-92 Jan-94 Jan-96 Jan-98 Assume you know whether stocks or bonds will do better over next 5 years What % of intervening months; does this look like the right decision? 75%
Alternative Investments Includes wide range of assets and instruments Property, Infrastructure and Timberland Commodities Private equity Absolute Return Strategies Alternatives: a matter of perspective Investments that promise better return on risk than long-only holdings of stocks and bonds Investments in inefficient markets lacking history to get a good fix on return/risk 28
Alternative Assets: Stages of Maturity Return On Risk Timber Real Estate Equilibrium return on risk Commodities? Irrational Exuberance Infrastructure, Hedge Funds First Mover Stage Mature Stage Time 29
Excess Alternative Return: Inefficiency and Risk Market Inefficiency First Mover Advantage: can disappear or turn negative Illiquidity: Reward for Committing Patient Capital Premium gets smaller over time: Real Estate, Timberland and Infrastructure Taking Operating Risk Return on Human Capital akin to running a listed company Offering insurance against price fluctuation Commodity futures Better active strategies.or just new risks Hedge funds and risk-based long-short management 30
Private Equity Returns: Gross vs. Net Internal Poor Man s Equity Portfolio Transfer $100 from S&P holdings expected S&P return: 10% Borrow $100 internally to fund another $100 S&P at bond portfolio return cost of 5% Portfolio Return: $200 x 10% - $100 x 5% = $100 x 15% Private Equity Alternative Sell $100 S&P to fund Private Equity Manager Money (Before 2 and 20) earns 20% after 50% leverage Net return = 20% - {2% - 20% x (20% - 8%)} = 20% - 4.4% = 15.6% You Need to Be Really Good To Earn Your Keep in Private Equity 31
Summary Quantitative methods can help preserve maximum net return for clients in low return world Must Manage long-term total return on total risk Doing Better than market is hard to do Beware of quantitative fads designed to benefit managers Investment management is risk management and risk is not easy to understand Quantitative approaches can be very helpful to improving long-term return/risk optimisation Measuring and managing basic operations as important as complicated active strategies 32