A Framework of Liquidity Risk Assessment for JP

A Framework of Liquidity Risk Assessment for JP Morgan Private Bank $500 -600 Billion Fund of Funds Multi-Asset Portfolio Construction & Optimization Presentations to Dr. Georgiy Zhikharev, JP Morgan Global Head of Quant Research & Analytics, JP Morgan US Head of Portfolio Construction, & Team of Managing Directors & Portfolio Managers June 21, 2012 and July 27, 2012 JP Morgan Global Headquarters: 270 Park Ave, New York, NY 10017 www. yogeshmalhotra. com Copyright, Yogesh Malhotra, Ph. D, 2012

Framework of Liquidity Risk Assessment for JP Morgan Private Bank $500 B Fund of Funds Multi-Asset Portfolio Construction & Optimization “With market risk and credit risk, you could lose a fortune. With liquidity risk, you could lose the bank!” LIQUIDITY Liquidity Risk / Cost Portfolio Illiquidity Cost Liquidity Premium Sharpe Ratio Liquidity Risk Liquidity Index RETURN MODEL Vehicles and Asset Classes Measures at Portfolio Level Liquidity Score Correlations Across Liquidity Insurance Funding Liquidity Risk Idiosyncratic Risk (Endogenous) Market Risk (Exogenous) Quantitative Qualitative - Stress Testing Aggregates Stochastic Assets f (size, complex, convertible) Measures for Various Asset Classes Capture Diversity of Risk Assets Preserve Core Parameters for Aggregates Deterministic - Scenarios Minimize Model Risk LIQUIDITY RISK Time Horizon Maturity Duration Convexity Volatility Autocorrelations Returns Stress Tests Discovery Price "Market risk can hurt you, but liquidity risk can kill you. " www. yogeshmalhotra. com Copyright, Yogesh Malhotra, Ph. D, 2012

Current State of Liquidity Research "Liquidity is a great metaphor, but we still don't have an unambiguous definition of it. " Regulators Liquidity and Solvency Systemic Risk of Failure Two Minimum Standards (Funding Liquidity Buffers) Liquidity Coverage Ratio, 30 - High Quality Liquid Assets Net Stable Funding Ratio - Minimum Stable Funding Academics Practitioners Measure, Monitor, Manage Currencies | Stress Testing Forward Liquidity Exposure Counter-Balancing Capacity - ‘Liquidity Buffer’ (Fiedler) - BSL = CBC - FLE - Transfer Pricing of Liquidity Liquification Algorithm/Score - Term Structure, Repo, FAI Interest Rate Risk Gaps & Liquidity Risk Gaps (Matz) LG: A&L |CF (Time) RG: A&L |Re-Price (Time) LVa. R = position ($) * [-drift (%) + volatility *deviate + LC]; 0. 5*spread " No single measure captures the various aspects of liquidity in financial markets. " www. yogeshmalhotra. com Copyright, Yogesh Malhotra, Ph. D, 2012

Current State of Liquidity Risk Measures “Comparing individual assets’ liquidities is problematic because one asset could be more liquid along one dimension of transaction costs while the other is more liquid in a different dimension. ” Assets: Specific Popular Liquidity Models (Price, Quantity, Time) Bid-Ask Spread Exogenous Position Size Endogenous Resiliency (Rare) Trading Volume (Flawed) - Flash Crash Tightness, Depth, Resilience (Kyle, 1985) Aggregates of Multiple Basic Liquidity Measures Bid Ask, Volume, Turnover, Loeb Price Impact (Lo 2010) (Amount of Trading, Cost) Serial Correlation as proxy. Liquidity Adjusted CAPM (Acharya & Pedersen, 2005 JFE) Risky and has Commonality Constant Trading Frictions Volume Weighted Spreads 1. Bid Ask Spread, 2. Transactions (Volume), 3. Volume Weighted Spreads Bangia et al. (2001) Berkowitz (2000) Giot & Gramming (2005) Assets: General Coherent Measures Aggregates of Multiple Basic Liquidity Measures Long-Short Index Measure Chacko, Das, Fan (2012) Long the ETFs Short components of ETFs Liquidity as Shadow Asset: Kinlaw et al. (2012) - Identify optimal weights by maximizing expected utility - Substitute illiquid asset for liquid equity - Solve for illiquid asset return based on σ and ρ - Estimate AR(1) Model using least squares - De-smooth the time series Liquidity as Real Option: Ang & Bollen (2008) Investor’s decision to withdraw capital: real option Lockups & Notice periods: exercise restrictions: reduction Basel Committee in the value of liquidity option. Jarrow & Protter (2005) Stange and Kaserer (2008 ) "No bank can ever afford to hold enough liquidity during normal times www. yogeshmalhotra. comto be able to survive a severe or prolonged funding disruption. " Copyright, Yogesh Malhotra, Ph. D, 2012

Autocorrelation Measure of Relative Liquidity & Setup for a Wall Street Asset Management Firm "Liquidity risk today is where credit risk was 10 years ago. " Assets: Specific Scale for Relative Liquidity Assets: General Portfolio How fast price (t) decays - Faster decay, more liquid How long stale price (t) lasts - More it affects, less liquid - Longer it lasts, less liquid Y Proxy for Liquidity Risk Plot serial correlations Use Model to find Decay Measure Decay Factor " In times of financial crisis, asset prices in some markets may reflect the amount of liquidity available in the market rather than the future earning power of the asset. " www. yogeshmalhotra. com Copyright, Yogesh Malhotra, Ph. D, 2012

Post-2008 Trends Requiring Research Advances for Wall Street Firm Multi-Asset Portfolio of 16 Asset Classes " People have known since Mandelbrot in 1963 that returns are not normally distributed. " Alternative Investments New Global Risk Indicators Liquidity Fund of Funds Credit spreads Bid-Ask Spreads Trade Volumes Correlation Matrices Forward Duration Curves Positions Outstanding New Macro Risk Indicators Systemic vs. Specific Risk Tradition VIX as Indicator Breakdown of VIX, Volatility Non-Normal Returns ‘Chaotic’ Returns No Risk-Free Rates Inflation Expectations Future Volatility Concerns Future Credit Concerns Sovereign Default "I'd just caution you that models are backward-looking. The future isn’t the past. " www. yogeshmalhotra. com Copyright, Yogesh Malhotra, Ph. D, 2012

Framework of Liquidity Assessment for Wall Street Asset Management Firm’s Multi-Asset Portfolio Construction & Optimization w Liquidity [Risk] Measures: Asset Classes, Aggregates, Portfolio w Framework n 16 -Asset Class Portfolio Liquidity Assessment l l n Defined Asset Classes & Aggregates (Vehicles) Endogenous Liquidity: Time / Maturity / Duration, Volatility / Autocorrelations, Returns / Stress Tests: Price Discovery 16 -Asset Class Portfolio Optimization: Sharpe / Liquidity Risk / Liquidity Insurance l l Funding Liquidity Risk, Asset Liquidity Risk Idiosyncratic (Endogenous), Market (Exogenous) w Models: Quantitative + Qualitative, Stochastic + Deterministic Stress Testing Scenarios Minimize Model Risk w Assets Modeled: Hedge Funds (HF), Alternative Investments, Equities, Commodities, Fixed Income, Bonds, Currencies "Normality has been an accepted wisdom in economics and finance for a century or more. Yet in real-world systems, nothing could be less normal than normality. Tails should not be unexpected, for they are the rule. " www. yogeshmalhotra. com Copyright, Yogesh Malhotra, Ph. D, 2012

Portfolio Construction Framework: An Overview MODELS / MEASURES Aggregate the PORTFOLIO Returns Liquidity Costs Correlations Volatilities AGGREGATES Returns Liquidity Costs Correlations Volatilities Measures Portfolio Normalize the Measures Aggregates Assets ASSETS Returns Liquidity Costs Correlations Volatilities Standardize the Measures Select Most Robust Measures "It is truly an art to build a long-term robust quantitative model that will perform well out-of-sample and through several different types of market environments. ” www. yogeshmalhotra. com Copyright, Yogesh Malhotra, Ph. D, 2012

Implementation Model MODELS / MEASURES PARAMETERS STRESS TESTS SHOCKS VALIDATE Portfolio SIMULATE Aggregate 1 Asset 2 Aggregate 2 Asset 3 REPLICATE Market Data Feeds Other Updates "I think you should be ambitious about your models, and push them as far as you can, but you need to be aware they will fail - and under what circumstances. ” www. yogeshmalhotra. com Copyright, Yogesh Malhotra, Ph. D, 2012

Stress Testing of Asset Classes & Combined Portfolio Integrated Analyses of 16 Different Classes of Alternative Investment Assets Hedge Fund of Funds & Funds of Funds Specific Analyses: Returns, VAMI, Histogram, Empirical vs. Normal Distbn. , Q-Q Plots, Moments (Mean, St. Dev. , Skew, Kurtosis), Descriptive Stats. , Correlations, Jarque-Barra Tests, Regressions & Plots Portfolio Analyses: MVO Solver, Rankings (Sharpe, MSRs, Sortino Ratios, DD Ratios, Information Ratios, M-Squared Ratios, Treynor Ratios, Jensen Ratios) Beyond Va. R: Hist. Simulation, Parametric Va. R, Mod. Va. R (CF), Expected Shortfall w w Portfolio Level and Asset Class Level Empirical Tests yield Consistent Findings verified by triangulation of empirical results from four key outputs n n n Mean Variance Optimization using MS-Excel Solver Stress Testing Results using Parametric Va. R Ranked Comparison of Risk Adjusted Return / Performance Measures VAMI Plots Visual Comparison Across Assets / Reconciliation with Robust Ratios Jarque-Barra Tests & QQ Plots www. yogeshmalhotra. com Copyright, Yogesh Malhotra, Ph. D, 10 2012

Stress Testing of Asset Classes & Combined Portfolio VAMI Plots Visual Comparison Across Assets / Reconciliation with Robust Ratios 700. 00 600. 00 SP RIY RTY 500. 00 400. 00 MXEF LPX 50 TR DJUBS 300. 00 USTW$ 200. 00 100. 00 HFRIEDI HFRIEHI HFRIMAI M on De th c. De 94 c. De 95 c. De 96 c. De 97 c. De 98 c. De 99 c. De 00 c. De 01 c. De 02 c. De 03 c. De 04 c. De 05 c. De 06 c. De 07 c. De 08 c. De 09 c. De 10 c 11 0. 00 www. yogeshmalhotra. com HFRIMI HFRIRVA Copyright, Yogesh Malhotra, Ph. D, 11 2012

Serial Autocorrelation of HF Funds vs. Market HF Returns are higher, auto-correlated, with less volatility www. yogeshmalhotra. com Copyright, Yogesh Malhotra, Ph. D, 12 2012