Reverse Stress Testing Ron Papanek ©2011. All rights reserved. msci.com msci.com Agenda What is Reverse Stress Testing? Why is it important? Practitioner Examples Single-Factor Multi-Factor Historical Monte Carlo Related Stress Tests Historical Unwind Liquidity Hedging ©2011. All rights reserved. msci.com Reverse Stress Testing – What is it? Definitions Traditional Stress Test Reverse Stress Test Shock P&L P&L Shocks Portfolio Specific Incorporating information specific to the fund company, or portfolio. Create an Explanatory Narrative Traditional Stress tests quantify loss from a real or potential event Reverse Stress tests tell a story from the numbers. ©2011. All rights reserved. msci.com Reverse Stress Testing – Why is it Important? Engages Management The methodology leads to a better understanding of firm risk through the process of identifying thresholds and the exploration of macro scenarios Connects the portfolio risk to the business risk Provides a different perspective than traditional stress tests Provides a more thorough examination of tail risk Can yield more efficient hedging strategies Tells a story Advocated by Regulators FSA, BIS, FED, CRMPG have all voiced support for reverse stress testing “The emphasis of a ‘reverse-stress test’ would be on identifying the high impact stress events which would cause the firm to fail and considering the appropriate action, if any, to protect against such failure.” FSA 12/2008 (2.27) Identify systemic risk and trigger points for contagion ©2011. All rights reserved. msci.com Reverse Stress Testing Reverse Stress Test Methodologies Sensitivity Analysis One Factor - Simple, Predictive, Multiple correlated factors Multifactor – two or more independent risk factors Scenario Analysis Historical – Date Range Monte Carlo – Loss simulations ©2011. All rights reserved. msci.com Reverse Stress Testing - One Factor Traditional stress test is a function not necessarily monotonic. Reverse stress test is not a function Different shocks can produce the same P&L ©2011. All rights reserved. msci.com One Two Factor Reverse Stress Test Equity Shift P&L ©2011. All rights reserved. msci.com Sensitivity Analysis Multi-Factor One Factor – One Dimension Two Factors – Two Dimensions Locking two factors together allows us to introduce a third factor ©2011. All rights reserved. msci.com Two Factor Reverse Stress Test ©2011. All rights reserved. msci.com Multi-Factor Sensitivity Grid ©2011. All rights reserved. 20 % 3 0 40 % 50 % msci.com Reverse Stress Testing - Historical Steps Run a traditional Historical stress test and sort by threshold violation. Run same Historical stress test, but widen period from days to rolling periods, widening periods until loss threshold is triggered. The output of the Reverse Historical Stress Test is a series of dates and date ranges. ©2011. All rights reserved. msci.com Reverse Stress Testing Historical Daily Daily Day 22 Day Weekly Monthly ©2011. All rights reserved. msci.com Reverse Historical Stress Test Match date scenario with events to explains the portfolio loss Single Date September 20, 2008 ©2011. All rights reserved. Date Range March 2003 msci.com Reverse Stress Testing – Scenario Generation Monte Carlo How do we simulate a 100 year flood? Run 100 years worth of Monte Carlo simulated returns Run Monte Carlo simulated returns using correlation periods from different market regimes. ©2011. All rights reserved. msci.com Reverse Stress Testing - Scenario Historical / Monte Carlo Hybrid Run long term Historical Simulation Sort sims by loss threshold and use dates to create a period for correlation calculation Run Monte Carlo simulated returns report using new “extreme correlation period” Sort sims by loss threshold and aggregate risk factor scenarios ©2011. All rights reserved. msci.com Reverse Stress Testing - Simulations Historical / Monte Carlo Hybrid Historical Monte Carlo ©2011. All rights reserved. msci.com Monte Carlo Visualization Worst ©2011. All rights reserved. to Best msci.com Reverse Stress Testing - Simulation Monte Carlo by Position ©2011. All rights reserved. msci.com Reverse Stress Testing - Simulation ©2011. All rights reserved. msci.com Reverse Stress Testing - Simulation Monte Carlo Visualization Visualization allows the inspection of hundreds or thousands of scenarios Observe the distribution of individual scenarios (not just the average) Visually identify hedge positions and hedge effectiveness ©2011. All rights reserved. msci.com Agenda What is Reverse Stress Testing? Why is it important? Practitioner Examples Single-Factor Multi-Factor Historical Monte Carlo Related Tools Historical Unwind Liquidity Hedging ©2011. All rights reserved. msci.com Reverse Stress Testing – Related Tools Historical Unwind Historical Stress Test Replays History Historical Unwind Replays History in Reverse Benefits Provides insight into Bubbles Separates buy and hold from arbitrage strategies Identifies increasing leverage Can highlight trend following Separates idiosyncratic trends within an asset class ©2011. All rights reserved. msci.com Reverse Stress Testing Historical Unwind 1 ©2011. All rights reserved. 2 3 4 month month month month msci.com Reverse Stress Testing Historical Unwind ©2011. All rights reserved. msci.com Reverse Stress Testing Liquidity Portfolio specific Risk Risk - based on position not security Test liquidation assumptions with acquisition history Compare Lock-up to liquidity horizon ©2011. All rights reserved. msci.com Agenda What is Reverse Stress Testing? Why is it important? Practitioner Examples Single-Factor Multi-Factor Historical Monte Carlo Related Tools Historical Unwind Liquidity Hedging ©2011. All rights reserved. msci.com Reverse Stress Testing - Hedging Hedge the Business not the Portfolio Focus on hedging the tail Identify and aggregate the most commonly occurring risk factor scenarios Finding the cheapest and most efficient hedge Strategies Costless Collar Correlated Asset Put Spread Macro Scenario overlay ©2011. All rights reserved. msci.com Reverse Stress Testing - Conclusions Engage management as much as the risk department The information flow between CEO and risk department is two way There is a significant amount of firm specific data that should be incorporated into stress testing and risk management. Connect a story to the risk The event has more meaning than the date or the scenario # or the P&L Firm risk is not portfolio risk – Hedge Appropriately P&L is linear Risk is not Reverse Stress Testing helps to identify the conditions for failure allowing for early warning as wells as prevention. ©2011. All rights reserved. msci.com http://www.riskmetrics.com/events/risk2009 ©2011. All rights reserved. msci.com msci.com