Managing Risk in Multi-Asset Class, Multimarket Central Counterparties: The CORE Approach Luis Antonio Barron G. Vicente Risk Management Officer May/2013 CLASSIFICATION OF INFORMATION (CHECK WITH AN “X”): CONFIDENTIAL AND RESTRICTED CONFIDENTIAL INTERNAL USE X PUBLIC AGENDA RISK MODELING IN MULTI-ASSET CLASS AND MULTIMARKET CLEARINGHOUSES THE CORE MODEL FOR CLEARINGHOUSE RISK CALCULATION HEDGING STRATEGIES THAT BENEFIT FROM THE CORE MODEL CORE MODEL IMPLEMENTATION KEY BENEFITS DERIVED FROM THE CORE MODEL IMPLEMENTATION AGENDA RISK MODELING IN MULTI-ASSET CLASS AND MULTIMARKET CLEARINGHOUSES THE CORE MODEL FOR CLEARINGHOUSE RISK CALCULATION HEDGING STRATEGIES THAT BENEFIT FROM THE CORE MODEL CORE MODEL IMPLEMENTATION KEY BENEFITS DERIVED FROM THE CORE MODEL IMPLEMENTATION RISK MODELING IN MULTI-ASSET CLASS CLEARINGHOUSES DEFINING A ROBUST & EFFICIENT RISK MODEL MULTI-ASSET CLASS AND MULTIMARKET CLEARINGHOUSES OPPORTUNITY TO INCREASE EFFICIENCY VIA RISK-OFFSETTING BUT HOW TO ENSURE THAT EFFICIENCY GAINS ARE ROBUST? Efficiency gains are not considered robust when the assumptions employed by the riskoffsetting model have a low level of adherence to reality, resulting in insufficient resources for the clearinghouse to fulfill its obligations NEED TO BUILD A RISK MODEL THAT REFLECTS, IN A REALISTIC WAY, THE RISK MANAGEMENT PROBLEM FACED BY A CLEARINGHOUSE 4 RISK MODELING IN MULTI-ASSET CLASS CLEARINGHOUSES THE RISK MANAGEMENT PROBLEM FACED BY A CLEARINGHOUSE IN THE EVENT OF A PARTICIPANT DEFAULT, THE RISK MANAGEMENT PROBLEM OF FACED BY A CLEARINGHOUSE IS TO HAVE THE RESOURCES AND LIQUIDITY NEEDED TO PROVIDE AN ORDERLY CLOSEOUT FOR THE SET OF POSITIONS HELD BY THE PARTICIPANT, UNDER CURRENT MARKET CONDITIONS, CONSIDERING A MINIMUM HOLDING PERIOD PORTFOLIO CLOSEOUT PROCESS T+0 T+1 T+2 T+3 T+4 ... T+N MAJOR ASPECTS THAT SHOULD BE TAKEN INTO ACCOUNT BY THE MODEL EVOLUTION (INTERTEMPORAL DYNAMICS) OF THE RISK FACTORS THAT DEFINE THE VALUE OF THE ASSETS AND CONTRACTS INCLUDED IN THE PORTFOLIO, AS WELL AS OF THE PORTFOLIO COMPOSITION ITSELF FRICTIONS, RESTRICTIONS AND OPERATIONAL FEATURES ASSOCIATED WITH EACH ASSET INCLUDED IN THE PORTFOLIO TRADING MODEL – ELECTRONIC VS OTC SETTLEMENT MODEL – RTGS VS DNS LIQUIDITY/MARKET DEPTH CASH FLOW STRUCTURE OF THE ASSET POSSIBILITY OF A FRACTIONAL SETTLEMENT 5 RISK MODELING IN MULTI-ASSET CLASS CLEARINGHOUSES A MORE COMPLEX APPROACH THAN THAT OF MODELS BASED ON VAR WHEN MODELLING THE RISK MANAGEMENT PROBLEM FACED BY A CLEARINGHOUSE, ONE MUST CONSIDER, IN A JOINT FASHION, THE EVOLUTION OF THE MARKET VARIABLES (PRICES & RATES) AND THAT OF THE PORTFOLIO COMPOSITION, RESPECTING A SET OF SIGNIFICANT RESTRICTIONS IMPOSED BY THE CHARACTERISTICS OF EACH ASSET UNDER CONSIDERATION PORTFOLIO CLOSEOUT RISK P&L CALCULATION T+0 T+1 T+2 T+3 T+4 ... T+N DYNAMIC PROCESS WITH FRICTIONS THIS TYPE OF MODELLING REQUIRES CONCEPTS AND TOOLS MORE COMPLEX THAN THOSE TYPICALLY EMPLOYED BY THE FINANCIAL INDUSTRY (I.E. MODELS BASED ON VAR). IN FACT, THESE MODELS OFTEN FOCUS ON MEASURING THE POTENTIAL VALUE OF A STATIC PORTFOLIO, WITHOUT TAKING INTO ACCOUNT A DYNAMIC CLOSEOUT PROCESS WITH FRICTIONS VARIATION RISK OF THE PORTFOLIO VALUE P&L CALCULATION T+0 T+N STATIC PROCESS WITHOUT FRICTIONS 6 RISK MODELING IN MULTI-ASSET CLASS CLEARINGHOUSES A MORE COMPLEX APPROACH THAN THAT OF MODELS BASED ON VAR (CONT’D) ALTHOUGH THE MODELS BASED ON VAR MAY BE ADAPTED TO ESTIMATE THE CLOSEOUT RISK, THEIR PLAUSIBILITY IS COMPROMISED WHEN MULTI-ASSET AND MULTIMARKET PORTFOLIOS (I.E. HIGHLY HETEROGENEOUS) ARE CONSIDERED UNDERLYING HYPOTHESIS: ALL ASSETS & CONTRACTS ARE TO BE SETTLED AT THE SAME TIME WITHOUT ANY FRICTIONS, WITH FULLY COINCIDING CASH FLOWS IMPLICIT CLOSEOUT MODEL T+0 T+N AN ALTERNATIVE APPROACH CONSISTS IN THE USE OF A MODEL BASED ON MULTIPLE SILOS, WHERE EACH SILO CONTAINS ONLY ASSETS AND/OR CONTRACTS WITH COMMON FEATURES (I.E. HOMOGENEOUS). IN THIS CASE, THE TOTAL PORTFOLIO RISK IS GIVEN BY THE ALGEBRAIC SUM OF EACH SILO. IMPLICIT CLOSEOUT MODEL SUM OF RISKS T+0 T+N SILO 1 T+0 T+0 T+N SILO 2 T+N SILO 3 7 ... RISK MODELING IN MULTI-ASSET CLASS CLEARINGHOUSES SILO MODELLING & SYSTEMIC RISK INCREASE EVEN A MODEL BASED ON SILOS, WITH SUPERCOLLATERALIZATION VIA SUM OF RISKS, DOES NOT NECESSARILY ENSURE A MORE ROBUST SYSTEM. IN FACT, A MODEL BASED ON SILOS MAY HIDE IMPORTANT RISKS OF LIQUIDITY FRAGMENTATION AND REDUCE INCENTIVES TOWARDS THE ADOPTION OF A DILIGENT BEHAVIOR IN TIMES OF CRISIS. ORIGINAL SITUATION, AGENTS “A” & “B” COLLATERAL (RISK) = 100 T+0 T+N T+0 SILO 1 T+N SILO 2 INCREASED MARKET VOLATILITY AGENT “A” HEDGES SILO 2 RISK ON THE MARKET COLLATERAL(RISK) = 200 T+0 T+N T+0 SILO 1 T+N LIQUIDITY RISK INCREASES IN THE SYSTEM SILO 2 AGENT “B” DOES NOT HEDGE AT ALL COLLATERAL (RISK) = 100 T+0 T+N SILO 1 T+0 DISINCENTIVE TOWARDS A DILIGENT BEHAVIOR T+N SILO 2 8 LTCM SCENARIOS (1998) & NTN-D CRISIS (2002) AGENDA RISK MODELING IN MULTI-ASSET CLASS AND MULTIMARKET CLEARINGHOUSES THE CORE MODEL FOR CLEARINGHOUSE RISK CALCULATION HEDGING STRATEGIES THAT BENEFIT FROM THE CORE MODEL CORE MODEL IMPLEMENTATION KEY BENEFITS DERIVED FROM THE CORE MODEL IMPLEMENTATION THE CORE MODEL FOR RISK CALCULATION THE CORE MODEL THE CORE MODEL WAS SPECIFICALLY DEVELOPED BY BM&FBOVESPA TO ALLOW FOR ROBUST AND EFFICIENT RISK ESTIMATION IN A MULTI-ASSET CLASS, MULTIMARKET CLEARINGHOUSE MAJOR FEATURES CONSIDERS THE INTERTEMPORAL DYNAMICS OF THE PORTFOLIO CLOSEOUT PROCESS CONTEMPLATES IMPORTANT FRICTIONS & RESTRICTIONS ASSOCIATED WITH THE SETTLEMENT PROCESS OF ASSETS AND CONTRACTS – TRADING DYNAMICS, MARKET LIQUIDITY AND DEPTH, CASH FLOW STRUCTURE, ETC ESTIMATES, IN BOTH A JOINT AND A CONSISTENT MANNER, THE MARKET AND LIQUIDITY RISKS ASSOCIATED WITH A PORTFOLIO CLOSEOUT PROCESS 10 THE CORE MODEL FOR RISK CALCULATION OVERVIEW: CLOSEOUT RISK CALCULATION IN THREE STEPS 1. DETERMINING THE CLOSEOUT STRATEGY T+0 T+1 T+2 T+3 T+4 ... T+N 2. RISK EVALUATION T+0 T+1 T+2 T+3 T+4 ... T+N T+1 T+2 T+3 T+4 ... CLOSEOUT RISK PERMANENT LOSS TRANSIENT LOSS 11 Defines the (stress) scenarios associated with the dynamics of each risk factor relevant to the portfolio. All assets and contracts are reevaluated considering the scenarios defined in this step (full valuation). Calculates and aggregates intertemporally P&L associated with each scenario, considering the defined closeout strategy 3. POTENTIAL P&L CALCULATION T+0 Defines the portfolio closeout strategy which, respecting the settlement restrictions of the portfolio of assets/markets, should minimize the risk of a loss associated with the closeout process, preserving existing hedge strategies T+N Result: Two risk measures—market and liquidity—that are estimated both jointly and consistently THE CORE MODEL FOR RISK CALCULATION OVERVIEW: PERMANENT & TRANSIENT LOSS 3. POTENTIAL P&L DETERMINATION T+0 + V1 + V2 + V3 + T+3 V4 + ... T+4 VN V0 ... T+N CASH NEED ON T+N EQUALS PERMANENT LOSS CASH NEED BY T+0 V0 + V1 V0 + V1 + V2 V0 + V1 + V2 + V3 V0 + V1 + V2 + V3 + V4 V0 + V1 + V2 + V3 + V4 CASH NEED BY T+1 CASH NEED BY T+2 CASH NEED BY T+3 CASH NEED BY T+4 + ... VN 12 CASH NEED BY T+N MAXIMUM BETWEEN CASH FLOW AMOUNTS V0 T+2 T+1 TRANSIENT LOSS THE CORE MODEL FOR RISK CALCULATION DETAIL: CLOSEOUT STRATEGY DEFINITION T+0 T+1 T+2 T+3 T+4 T+5 CLOSEOUT PORTFOLIO FUTURES, BUY, IMMEDIATE SETTLEMENT OPTIONS, SELL, SETTLEMENT ON T+3 ONLY 1 NAIVE STRATEGY 2 OPTIMAL STRATEGY DEFINITION 3 OPTIMAL STRATEGY RISK SWAP, SELL, SETTLEMENT ON T+5 ONLY MINIMUM RISK ITERATION 13 THE CORE MODEL FOR RISK CALCULATION DETAIL: PORTFOLIO COMPOSITION & RISK FACTOR EVOLUTION T+1 T+2 T+3 T+4 T+5 T+6 T+4 T+5 T+6 P&L ALONG THE PROCESS T+0 CLOSEOUT PORTFOLIO FACTOR 1 FACTOR 2 FACTOR n T+0 MARKET T+1 T+2 T+3 RISK FACTOR EVOLUTION 14 THE CORE MODEL FOR RISK CALCULATION DETAIL: RISK FACTOR EVOLUTION & MULTIVARIATE SCENARIO GENERATION FACTOR 1 FACTOR 1 FACTOR 2 MULTIVARIATE FACTOR 2 SCENARIO ... GENERATOR ... ... ... ... ... FACTOR n FACTOR n T+0 – T+1 – T+2- ... – T+N T+0 – T+1 – T+2- ... – T+N # SCENARIO ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... SCENARIOS TO DETERMINE P&L DURING THE CLOSEOUT PROCESS ... ... ... ... ... ... ... T+0 – T+1 – T+2 – ... – T+N 15 THE CORE MODEL FOR RISK CALCULATION DETAIL: P&L DETERMINATION DURING THE CLOSEOUT PROCESS PERMANENT LOSS TRANSIENT LOSS PERMANENT LOSS TRANSIENT LOSS #2 PERMANENT LOSS TRANSIENT LOSS #3 ... ... ... ... ... PERMANENT LOSS TRANSIENT LOSS #nSCN SCENARIOS T+1 T+2 T+3 T+4 16 T+5 T+6 POSITIVE FLOW NEGATIVE FLOW WORST CASE SCENARIO #1 AGENDA RISK MODELING IN MULTI-ASSET CLASS AND MULTIMARKET CLEARINGHOUSES THE CORE MODEL FOR CLEARINGHOUSE RISK CALCULATION HEDGING STRATEGIES THAT BENEFIT FROM THE CORE MODEL CORE MODEL IMPLEMENTATION KEY BENEFITS DERIVED FROM THE CORE MODEL IMPLEMENTATION HEDGING STRATEGIES BENEFITING FROM THE CORE MODEL MAIN EXAMPLES HEDGING AN OTC DERIVATIVES POSITION ON THE LISTED DERIVATIVES MARKET CLOSEOUT RISK CORE RISK T+0 T+1 T+2 T+3 T+4 ... T+N SUM OF RISKS CURRENT MODEL T+0 T+T T+0 T+T T+0 T+N SILO 1 SILO 2 SILO 3 OTC POSITION LISTED DERIVATIVES COLLATERAL CORE RISK: PORTFOLIO CLOSEOUT COST (POSITIONS + COLLATERAL) MUST BE EQUAL TO OR LESS THAN ZERO CURRENT MODEL: COLLATERAL-HAIRCUT EQUAL TO OR GREATER THAN RISK (OTC) + RISK (LISTED DERIVATIVES) 18 HEDGING STRATEGIES BENEFITING FROM THE CORE MODEL MAIN EXAMPLES (CONT’D) ASSET BEING HEDGED IS POSTED AS COLLATERAL CLOSEOUT RISK CORE RISK T+0 T+1 T+2 T+3 T+4 ... T+N SUM OF RISKS CURRENT MODEL T+0 T+T T+0 T+N SILO 1 SILO 2 LISTED DERIVATIVES COLLATERAL CORE RISK: PORTFOLIO CLOSEOUT COST (POSITIONS + COLLATERAL) MUST BE EQUAL TO OR LESS THAN ZERO CURRENT MODEL: COLLATERAL-HAIRCUT EQUAL TO OR GREATER THAN RISK (LISTED DERIVATIVES) 19 HEDGING STRATEGIES BENEFITING FROM THE CORE MODEL MAIN EXAMPLES (CONT’D) EQUITIES BORROWER HOLDING COLLATERAL IN SHARES OF THE SAME COMPANY, BUT OF A DIFFERENT TYPE (PREFERRED VS COMMON) CLOSEOUT RISK CORE RISK T+0 T+1 T+2 T+3 T+4 ... T+N SUM OF RISKS CURRENT MODEL T+0 T+T T+0 T+N SILO 1 SILO 2 EQUITIES LENDING COLLATERAL CORE RISK: PORTFOLIO CLOSEOUT COST (POSITIONS + COLLATERAL) MUST BE EQUAL TO OR LESS THAN ZERO CURRENT MODEL: COLLATERAL-HAIRCUT EQUAL TO OR GREATER THAN RISK (LENDING) 20 AGENDA RISK MODELING IN MULTI-ASSET CLASS AND MULTIMARKET CLEARINGHOUSES THE CORE MODEL FOR CLEARINGHOUSE RISK CALCULATION HEDGING STRATEGIES THAT BENEFIT FROM THE CORE MODEL CORE MODEL IMPLEMENTATION KEY BENEFITS DERIVED FROM THE CORE MODEL IMPLEMENTATION CORE MODEL IMPLEMENTATION MODEL COMPONENTS & IT ARCHITECTURE OPTIMAL CLOSEOUT STRATEGY DEFINITION SPECIFIC SOFTWARE TO DEAL WITH OPTIMIZATION ISSUES PRICE GENERATION BASED ON MULTIVARIATE SCENARIOS VERY HIGH PERFORMANCE PARALLEL ARCHITECTURE USING GRAPHIC UNITS WITH MULTIPLE PROCESSORS (GPUs) RISK AGGREGATION & CONTROL HIGH PERFORMANCE SOFTWARE DEVELOPED IN C++ BY BM&FBOVESPA INTERFACE WITH THE RTC PLATFORM (CINNOBER) RISK PLUG-IN DEVELOPED BY BM&FBOVESPA IN TANDEM WITH CINNOBER 22 CORE MODEL IMPLEMENTATION TEAMS INVOLVED MODEL DEFINITION, PROTOTYPE CONSTRUCTION, DEFINITIVE MODEL TESTING FINANCE CONCEPTS (MR. MARCO AVELLANEDA/NYU & MR. RAMA CONT/COLUMBIA) RISK MANAGEMENT OFFICE IT OFFICE POST-TRADING CORE MODEL DEVELOPMENT 23 INDEPENDENT ASSESMENT, FEASIBILITY ANALYSIS, SUPPORT TO MODEL DEFINITION CORE MODEL IMPLEMENTATION PROJECT STATUS - MACRO CONCEPTUAL MODEL MATHEMATICAL MODEL PROTOTYPE RISK PLUG-IN/CORE DEC2010 JUL2010 DEC2011 JUL2011 PROTOTYPE PRESENTATION 24 DEC2012 MAR2013 AGENDA RISK MODELING IN MULTI-ASSET CLASS AND MULTIMARKET CLEARINGHOUSES THE CORE MODEL FOR CLEARINGHOUSE RISK CALCULATION HEDGING STRATEGIES THAT BENEFIT FROM THE CORE MODEL CORE MODEL IMPLEMENTATION KEY BENEFITS DERIVED FROM THE CORE MODEL IMPLEMENTATION KEY BENEFITS DERIVED FROM THE CORE MODEL IMPLEMENTATION DEVELOPED SPECIFICALLY TO DEAL WITH THE RISK MANAGEMENT PROBLEM FACED BY CLEARINGHOUSES ROBUST MODELLING PROVIDING EFFICIENCY GAINS WITHOUT GIVING UP SAFETY TRANSPARENT & INTUITIVE MODEL – ASSUMPTIONS CA BE EASILY VALIDATED MARKET & LIQUIDITY RISKS ARE TREATED IN BOTH A JOINT AND A CONSISTENT MANNER GREATER EFFICIENCY IN CAPITAL ALLOCATION FOR PORTFOLIOS WITH RISK MITIGATION STRATEGIES (HEDGE) INCENTIVES TO THE ADOPTION OF PRUDENTIAL MEASURES TO MITIGATE RISKS CIRCUMVENTS THE SILO APPROACH, SO LIQUIDITY FRAGMENTATION IS AVOIDED AND SYSTEMIC RISK MITIGATED 26