Asset Allocation in a non-normal Framework using PortfolioChoice A New Approach to Portfolio Selection

Transcription

1 Asset Allocation in a non-normal Framework using PortfolioChoice A New Approach to Portfolio Selection Paul Spence, Director Kenneth Lassner, CFA, Director April 2004 Deutsche Asset Management is the marketing name in the US for the asset management activities of Deutsche Bank AG, Deutsche Bank Trust Company Americas, Deutsche Asset Management Inc., Deutsche Asset Management Investment Services Ltd., Deutsche Investment Management Americas Inc. and Scudder Trust Company

2 Agenda Deutsche Asset Management overview Modern portfolio theory: where it all started PortfolioChoice: what is it? PortfolioChoice: how does it work? Examples: PortfolioChoice in action Appendix: disclaimers and disclosures Deutsche Asset Management 1

3 Deutsche Asset Management Applying these advances to asset management: Global capabilities, local expertise Over $700 billion* in assets under management More than 500 investment professionals globally Investment insights based on local market intelligence and global outlook Montreal London Chicago San Francisco Boston New York Philadelphia Frankfurt Milan Seoul Hong Kong Singapore Tokyo Disciplined pursuit of performance Full spectrum of fundamental, quantitative and alternative investment products Expertise across asset classes, industry sectors and geographic regions Value-added asset management services As of December 31, 2003 *Note: balanced mandates are not broken out into underlying asset classes. Assets shown in US dollars. Sydney Real-world approach to innovation New product initiatives linked to clients evolving investment challenges and new market realities Deep quantitative resources to support portfolio management, process enhancements, attribution and alpha generation Applications-oriented research to address both the tactical and theoretical challenges facing the discipline of asset management Deutsche Asset Management 2

4 Quantitative teams bring a critical skill set Advanced Research & Quantitative Strategies (ARQS) team Manage quantitatively-based client portfolios (eg, active quantitative equity, igap, GrOWE) Assist clients with strategic investment issues (eg, optimization, allocation) Create and implement custom alpha strategies (eg, portable alpha structured products) Develop and enhance proprietary portfolio models and analytics Advanced Research & Quantitative Strategies Advanced Research & Strategy Development Portfolio Management Economics, Investment & Product Strategy Global alpha platform -- portfolio modeling and construction Risk budgeting Investment strategy research & development Enhancements to existing investment processes Client-directed research Portfolio management -- global asset allocation, active quantitative equity, enhanced global passive, tax-advantaged equity Trading for quantitatively based strategies Infrastructure management and development Economic outlook and analysis Product development Product management Liaison between quantitative staff and clients Deutsche Asset Management 3

5 ARQS products and services integrated Global Alpha Platform Quantitative Equities Strategic Asset Allocation Macro based global alpha platform Portable alpha / Overlay strategy Global tactical asset allocation using derivatives or physical portfolios Active quantitative equity management - US - Europe GrOWE - Global and Global ex-us enhanced passive Tax advantaged equity - Customized harvesting of gains and losses for US and international equity PortfolioChoice Client advisory Strategic asset allocation funds Deutsche Asset Management 4

6 Modern Portfolio Theory: where it all started Two theories of diversification (1952) Mean-variance objective (Markowitz) Probabilistic objective (Roy) Due to limitations on computing power Markowitz s theory was tractable Roy s was not In recent decades, finance has built on Markowitz s work: 1960s CAPM (Sharpe) 1970s Option pricing (Black Scholes Merton) 1980s Multi-period Finance (Merton) 1990s Re-sampled efficient frontier (Michaud) But there has been no evolution in the theory, nor a return to Roy s work Until now Deutsche Asset Management 5

7 What it is all about In an uncertain world, the best way to conduct your affairs is through probabilistic decision-making. Success came by evaluating all the information available to try to judge the odds of various outcomes and the possible gains or losses associated with each. Former Treasury Secretary Robert Rubin s book: In an Uncertain World: Tough Choices from Wall Street to Washington : 2003 Deutsche Asset Management 6

8 PortfolioChoice is made possible by advances in: Computing power Finance Statistics Econometrics Operations research Deutsche Asset Management 7

9 What is PortfolioChoice? PortfolioChoice is a portfolio selection tool, just like any other, that takes inputs and processes them into outputs: Data inputs Statistical analysis Output: investment portfolios Yet, PortfolioChoice is also different from traditional portfolio selection tools: Larger universe of inputs Advanced statistical processes Different investment portfolios Deutsche Asset Management 8

10 Asking the traditional investment question Investors are presented with a direct trade-off between risk and return*: Assumes all investors have the same objective: maximizing the risk-return relationship Each point on the frontier seeks to maximize return per unit of risk This rigid definition of the investor s objective allows only one optimal portfolio for each objective (ie, return target) Expected Return 14% 12% 10% 8% 6% 4% Exp Return 8% Std Devn 3.9% MV Efficient Frontier Exp Return 10% Std Devn 8.2% Exp Return 12% Std Devn 14.7% 2% 0% 0% 2% 4% 6% 8% 10% 12% 14% 16% Std Deviation *For illustrative purposes only. Please see notes and disclosures in the appendix for more information. Deutsche Asset Management 9

11 PortfolioChoice takes a different approach PortfolioChoice re-frames the question with new parameters that we believe better reflect investors true objectives: Maximizing the probability of outperformance versus Minimizing the potential shortfall Probability of Outperformance 70% 60% 50% 40% 30% 20% 10% Target return = 8% X% Probabilistic Efficient Frontier Goal is to minimize expected shortfall relative to the return target Goal is to maximize the Target probability return of outperformance relative to the return target Target return 0% 0% 1% 2% 3% 4% 5% 6% Expected shortfall This is a hypothetical example being used for illustrative purposes only. The scenarios noted are not exhaustive to the various outcomes to the markets. Please see notes and disclosures in the appendix for more information. Deutsche Asset Management 10

12 PortfolioChoice offers a greater range of choices Each investor can choose from a range of optimal portfolios Optimal is defined according to the investor s objective Depends on resolution of the trade-off between maximizing probability of outperformance versus minimizing potential shortfall Probability of Outperformance 70% 60% 50% 40% 30% 20% Target return return 8% = X% Probabilistic Efficient Frontiers Target Target return return = X 10% + 2% Target Target return return = X + 12% 4% 10% 0% 0% 1% 2% 3% 4% 5% 6% Expected shortfall This is a hypothetical example being used for illustrative purposes only. The scenarios noted are not exhaustive to the various outcomes to the markets. Please see notes and disclosures in the appendix for more information. Deutsche Asset Management 11

13 PortfolioChoice: advantages over mean-variance Investor objective Definition of risk Data issues Estimation risk Platform Mean-Variance Not able to handle customized objectives Typically stated as maximize expected return per unit of total volatility Standard deviation used as the risk measure which encompasses both upside and downside deviations Historic data typically used Can t deal with incomplete data sets Ignores features that may be found in non-normal return distribution, such as skewness Data inputs of returns and volatilities treated as certain Small changes in assumptions or inputs can lead to wide swings in optimal allocations Most traditional mean-variance optimizations are quick due to simplicity of objective PortfolioChoice Objectives are customized to solve investor s needs Maximize the probability of reaching an objective (return or benchmark) Minimize the shortfall from an objective (return or benchmark) Risk measure is focused on the downside and is defined as minimizing a shortfall from an objective Blends history with forecasts through enhanced Monte Carlo methods Can fill in missing or short data sets Can deal with non-normal return distributions typically found in alternative asset classes Allows investor to add uncertainty to data input Results are more stable - less sensitive to small changes in inputs and assumptions PortfolioChoice optimization module generates large numbers of scenarios as it seeks the optimal solution Results Results expressed as maximum expected Sharpe ratio portfolio (total expected return per unit of total risk) Full distribution of possible outcomes are the same for all portfolios with the same expected return and risk Results expressed as a probability of reaching an objective and the average shortfall expected Distribution of possible outcomes are different - can exhibit positive or negative skewness and other non-normal properties Deutsche Asset Management 12

14 PortfolioChoice addresses a wide range of issues Overall, PortfolioChoice uses advanced statistical techniques to address wide range of issues that are beyond the capability of traditional techniques: Objectives -- PortfolioChoice can express a range of objectives, from conservative to aggressive Investor views -- PortfolioChoice can blend historical data with a range of investor views about potential market performance (ie, views from optimistic to pessimistic) Data challenges -- PortfolioChoice can address difficult data issues such as short manager histories, missing data and non-normal distributions Market risks -- PortfolioChoice can express uncertainty as well as the investor s level of confidence in market views (eg, views on returns,volatilities, correlations) Deutsche Asset Management 13

15 How does it work? Stage 1: analyzing beliefs and views about the potential performance of assets Stage 2: portfolio selection based on investor preferences Each stage follows a similar format: INPUTS ANALYSIS OUTPUTS Deutsche Asset Management 14

16 Stage 1: assembling and analyzing raw data Stage 1 considers these inputs: Historical returns Forecasts/views Modifies historical data to handle issues such as: Short data histories/missing data Non-normal distributions Estimation risk Stage 1 INPUTS Historical data Manager returns Benchmark returns Asset returns Forecasts/views Means Volatilities Correlations Deutsche Asset Management 15

17 Stage 1: assembling and analyzing raw data Statistical analysis: Uncertainty is a risk to investors Explicitly treated here as probability Markov-Chain-Monte-Carlo module: Advanced computational technique Robust problem-solving capacity Stage 1 INPUTS STATISTICAL ANALYSIS Historical data Manager returns Benchmark returns Asset returns Forecasts/Views Means Volatilities Correlations Markov Chain Monte Carlo Module Deutsche Asset Management 16

18 Stage 1: assembling and analyzing raw data Predictive distribution: A representation of investor s overall beliefs Projection is based on historical data and investor views 20,000 potential return scenarios Stage 1 INPUTS STATISTICAL ANALYSIS OUTPUTS Historical data Manager returns Benchmark returns Asset returns Forecasts/Views Means Volatilities Correlations Markov Chain Monte Carlo Module Investor beliefs/predictive distribution Many potential scenarios for future returns for each asset Summary statistics Deutsche Asset Management 17

19 Stage 2: portfolio optimization and selection The results from Stage 1 are used as inputs for Stage 2 The other set of inputs are our goals and preferences Stage 1 Historical data Views Markov Chain Monte Carlo Module Beliefs/predictive distribution Stage 2 Goals/preferences Traditional mean variance Probability of outperformance Minimze expected shortfall Deutsche Asset Management 18

20 Stage 2: portfolio optimization and selection Identifies an optimal portfolio based on the predictive distribution and the investor s objectives Performs approximately calculations Produces output statistics Stage 1 Markov Chain Monte Carlo Module Beliefs/predictive distribution Stage 2 Output Goals/preferences Traditional mean variance Probability of outperformance Minimze expected shortfall Optimization Module Optimal portfolio weights Prob. of outperformance Expected shortfall Expected return Portfolio volatility Distribution of returns Deutsche Asset Management 19

21 Stage 2: portfolio optimization and selection Portfolio selection process can be run multiple times to test various scenarios: Objectives: conservative to aggressive Views: optimistic to pessimistic Create probabilistic efficient frontier (showing trade off based on shortfall-aversion) Stage 1 Markov Chain Monte Carlo Module Beliefs/predictive distribution Stage 2 Goals/preferences Traditional mean variance Probability of outperformance Minimze expected shortfall Optimization Module Optimal portfolio Potential performance Deutsche Asset Management 20

22 An example: distribution Target return Average return Expected shortfall Probability of outperformance -10% -5% 0% 5% 10% 15% 20% Deutsche Asset Management 21

23 How can this new approach benefit investors? We believe that it provides a better approach to objective setting More flexible More customized Greater range of options Absolute return targets, benchmark targets, excess return targets Probability of outperformance, minimize shortfall, or a combination of the two More realistic assessment of potential market risks and behavior Solutions are more appropriate to client s true objectives Deutsche Asset Management 22

24 Examples: PortfolioChoice in action Deutsche Asset Management 23

25 Value/growth example Imagine it is January 2000 (after four years of relatively strong growth performance) and we are comparing two managers:* A growth manager with a short history A value manager with a long history How do we compare them fairly? Truncate the longer history? -- that throws away valuable information! Extend the short history? -- what do we use as a reasonable proxy? Value manager Growth manager 17.0% (1/79-12/99) 19.5% (1/96-12/99) 31.2% Dec-79- Dec-80 Dec-81 Dec-82 Dec-83 Dec-84 Dec-85 Dec-86 Dec-87 Dec-88 Dec-89 Dec-90 Dec-91 Dec-92 Dec-93 Dec-94 Dec-95 Dec-96 Dec-97 Dec-98 Dec-99 *Russell 1000 Value and Growth indices used to represent value and growth manager. Deutsche Asset Management 24

26 Truncation leads to over-emphasis on growth 35% Zephyr StyleADVISOR Manager Risk/Return Single Computation January December 1999 Zephyr StyleADVISOR: Deutsche Asset Management Return 30% 20% 10 % Value manager Russell 1000 Value Growth manager Russell 1000 Growth Market Benchmark: S&P 500 Index Cash Equivalent: Citigroup 3-month T-bill Capital Market Line Much higher Sharpe ratio influences meanvariance optimizer -- despite significant bias in the growth history. 0% 0% 5% 10% 15% 18% Standard Deviation Risk-Return Table January December 1999: Annualized Summary Statistics Return (%) Std Dev (%) Sharpe Ratio Observs. Value manager Growth manager S&P 500 Index Past performance is not an indication of future results. Russell 1000 Value and Growth indices used to represent value and growth manager. Deutsche Asset Management 25

27 PortfolioChoice is designed to avoid disaster results 1 PortofolioChoice estimates the missing history by statistical extrapolation. Mean-Variance portfolio PortfolioChoice portfolio 2. Estimate is constructed using the growth manager s actual correlation to the benchmark and the value manager. 0% 45% 55% 3. Allows us to preserve the full performance history of the value manager. 100% 4. Provides a more realistic allocation between the two managers. Grow th Value Grow th Value Deutsche Asset Management 26

28 Historical excess return (annualized) 15.0% 10.0% 9.3% 5.0% 0.0% -5.0% -10.0% -15.0% -20.0% -25.0% -30.0% -14.3% -23.6% PC portfolio MV portfolio Difference As of December 31, 2002 Source: DeAM analysis Past performance is not an indication of future results. Russell 1000 Value and Growth indices used to represent value and growth manager. Deutsche Asset Management 27

29 Appendix: disclaimers and disclosures Deutsche Asset Management 28

30 Paul Spence, Director Head of Quantitative Investments for the Asia-Pacific Region. Joined Deutsche Asset Management from Zurich Scudder Investments. Previously worked at Barra International as a research consultant. B.App.Sc (Mathematics) from University of Technology, Sydney. Member of the Bachelier Finance Society. Deutsche Asset Management 29

31 Kenneth Lassner, CFA, Director Portfolio investment specialist representing Deutsche Asset Management s global asset allocation group: New York Joined Deutsche Asset Management in 2002 after 13 years of experience as vice president/investment specialist for JP Morgan Fleming Asset Management, plan sponsor/manager of investments for the Federal Reserve and the Interpublic Group of Companies and treasury analyst for Dime Savings Bank BS from Babson College; MBA from Baruch College Member of AIMR and the New York Society of Securities Analysts Deutsche Asset Management 30

32 Index definitions Index returns assume reinvested dividends and, unlike Composite returns, do not reflect any fees or expenses. It is not possible to invest directly into an index. Citigroup 3-month T-bill Index is an average of the last 3-month US Treasury Bill issues. S&P 500 Index is an unmanaged index widely regarded as representative of the equity market in general. Russell 2000 Index is an unmanaged index that tracks the common stock price movement of the 2000 smallest companies of the Russell 3000 Index, which measures the performance of the 3000 largest US companies based on total market capitalization. Russell 1000 Index The Russell 1000 Index consists of the 1,000 largest securities in the Russell 3000 Index. The Russell 1000 Index represents approximately 89% of the total market capitalization of the Russell 3000 Index. This large cap index is highly correlated with the S&P 500 Index. MSCI EAFE Index is an unmanaged index that tracks international stock performance in the 21 developed markets of Europe, Australasia, and the Far East. MSCI World Index is an unmanaged index of over 1,500 stocks traded in approximately 23 developed world markets. The Lehman Brothers Aggregate Bond Index is an unmanaged index that tracks investment grade bonds including US Treasury and agency issues, corporate bond issues, asset backed, commercial mortgaged backed and mortgage backed securities and Yankee issues. Citigroup World Government Bond Index is an unmanaged index on a US dollar total return basis with all dividends reinvested and is comprised of government bonds from 14 countries. Deutsche Asset Management 31

33 Notes regarding illustrations The illustrations on slides 8, 9 and 10 are based on assumptions for mean return and volatility shown in the table below: US equities Intl equities US bonds Intl bonds Hedge funds Mean 13.1% 7.9% 6.9% 4.7% 8.0% Std devn 19.2% 17.8% 3.7% 9.2% 9.1% US equities are represented by the Russell 1000 Index International equities are represented by the MSCI EAFE Index US bonds are represented by the Lehman Aggregate Bond Index International bonds are represented by the Salomon Government World Bond Index Hedge funds are represented by a proprietary DeAM calculation based on a multi-manager hedge fund portfolio that we believe is broadly representative of the hedge fund market Deutsche Asset Management 32

34 Notes regarding performance The performance tables and charts herein do not reflect the deduction of investment management fees. In the event that such investment management fees and other fees were deducted, the performance of an account would be lower. For example, if an account appreciated by 10% a year for five years, the total annualized return for five years prior to deducting fees at the end of the five-year period would be 10%. If total account fees were.10% for each of the five years, the total annualized return of the account for 5 years at the end of the five-year period would be 9.89%. Past performance is not indicative of future results. Performance numbers used in illustrations are not necessarily indicative of the results you would obtain as a client of Deutsche Asset Management. Results are generally based on security selection, client investment restrictions (if any) market economic conditions and other factors which would all influence portfolio returns. Deutsche Asset Management 33

35 Important information The information is being provided solely as an illustration of how the PortfolioChoice selection process and approach; any quoted PortfolioChoice performance is an estimate that is relevant only to the specified time period, compared with portfolios based on historical extrapolation. The data does not reflect actual performance of any single portfolio, nor was a contemporaneous investment model run. Simulated performance results have inherent limitations. Unlike an actual performance record, simulated results do not represent actual trading and are subject to the fact that they are designed with the benefit of hindsight. Also, since the trades have not actually been made, the results may not reflect the impact that certain material economic and market factors might have had on an investment adviser's actual decision-making. No representation is being made that these or similar results will, or likely, be achieved. Deutsche Asset Management 34

36 Important information No assurance can be given that the investment objectives shown will be achieved or that an investor will receive a return of all or part of its investment, and investment results may vary substantially on an annual or quarterly or monthly basis. An investment is not a deposit and is not insured or guaranteed by the Federal Deposit Insurance Corporation or any other governmental agency or by Deutsche Bank AG, its affiliates or subsidiaries. This material is intended for information purposes only and does not constitute investment advice or an offer or solicitation, and is not the basis for any contract to purchase or sell any security or other instrument or for Deutsche Bank AG to enter into or arrange any type of transaction as a consequence of any information contained herein. We or our affiliates or persons associated with us may maintain a long or short position in securities referred to herein or in related futures or options, purchase or sell, make a market in, or engage in any other transaction involving such securities and earn brokerage or other compensation in respect of the foregoing. In preparing this presentation, we have relied upon and assumed without independent verification, the accuracy and completeness of all information available from public sources. Neither this presentation nor any of its contents may be used for any purpose without the consent and knowledge of Deutsche Bank. The information in this presentation reflects prevailing market conditions and our judgment as of this date, which are subject to change. Deutsche Asset Management 35