casemark/managing-xva-calculations

Managing XVA Calculations

Structures CVA, DVA, FVA, and KVA calculation with methodology selection and counterparty exposure modeling.

When To Use

• Pricing or re-pricing OTC derivatives where counterparty credit risk adjustments are material
• Computing desk-level or portfolio-level XVA charges for P&L attribution
• Selecting or validating methodology choices (SA-CCR vs. IMM, analytic vs. Monte Carlo)
• Reconciling front-office XVA with risk-neutral CVA/DVA reported to finance
• Responding to model validation or audit queries on XVA assumptions
• Assessing incremental XVA impact for new trades, novations, or collateral restructuring

Inputs To Gather

• Trade population: Netting set definitions, CSA/ISDA terms, collateral thresholds, minimum transfer amounts, margin frequency
• Market data: Yield curves, credit spreads (CDS or mapped proxies), FX/equity/commodity vols, correlation matrices
• Counterparty data: Credit ratings, PD term structures, LGD assumptions, wrong-way risk indicators
• Funding curves: Treasury funding spreads, secured vs. unsecured funding basis, internal FTP rates
• Capital parameters: RWA methodology (SA-CCR or IMM), cost-of-capital rate, regulatory alpha multiplier [VERIFY — jurisdiction-specific]
• Collateral terms: Eligible collateral types, haircut schedules, rehypothecation rights, variation margin vs. initial margin treatment
• Simulation parameters: Number of Monte Carlo paths, time grid granularity, diffusion model choices, close-out period assumption

Workflow

1. Define netting and margining structure

   • Map trades to legal netting sets under enforceable ISDA/CSA agreements
   • Confirm collateral terms: threshold, MTA, rounding, independent amounts
   • Identify any one-way CSAs, break clauses, or optionality affecting exposure profiles
   • Flag netting opinions that are missing or expired [VERIFY — enforceability by jurisdiction]

2. Generate exposure profiles

   • Select simulation approach: full Monte Carlo revaluation, regression-based (e.g., Longstaff-Schwartz for American-style features), or analytic approximation for vanilla books
   • Define time grid — finer near term (weekly to 1Y), coarser long term (quarterly/annual beyond 5Y)
   • Simulate risk factor paths under risk-neutral measure for pricing, real-world measure if needed for KVA/regulatory
   • Compute per-path, per-time-step MtM; apply netting and collateral modeling to produce Expected Exposure (EE), Expected Positive Exposure (EPE), and Expected Negative Exposure (ENE)
   • Derive Potential Future Exposure (PFE) at required quantiles (typically 95th or 97.5th)

3. Compute individual XVA components

   • CVA: Integrate discounted EPE against counterparty default probability and LGD. For Basel III SA-CVA, apply prescribed risk weights and correlation parameters [VERIFY — SA-CVA vs. BA-CVA eligibility under local regulation]
   • DVA: Mirror calculation using ENE and own-default probability. Determine whether DVA is recognized in regulatory capital vs. only in accounting P&L [VERIFY — prudential treatment varies by jurisdiction]
   • FVA: Decompose into Funding Benefit Adjustment (FBA) on negative exposure and Funding Cost Adjustment (FCA) on positive exposure. Apply unsecured funding spread net of collateral offsets. Address controversy: ensure FVA is not double-counting DVA
   • ColVA: Calculate cost/benefit of posting collateral — opportunity cost of segregated IM, carry on rehypothecable VM
   • KVA: Estimate lifetime cost of regulatory capital (RWA × cost-of-capital rate) discounted over the portfolio's expected life. Requires forward projection of capital under chosen methodology (SA-CCR alpha factor = 1.4 unless internal model approved)
   • MVA: Compute cost of posting initial margin under UMR/cleared margin rules; project IM using ISDA SIMM or CCP schedule

4. Aggregate and attribute

   • Roll up netting-set-level XVA to counterparty, desk, business line, and firm-wide totals
   • Perform incremental XVA analysis: marginal contribution of individual trades to portfolio-level CVA/FVA
   • Calculate XVA sensitivities (CS01, IR delta, vega) for hedging desk consumption
   • Reconcile front-office XVA (risk-neutral, continuous) with accounting CVA (fair-value, quarterly) and regulatory CVA capital charge

5. Report and escalate

   • Produce XVA P&L explain: decompose period-over-period changes into market risk, credit migration, new trades, trade maturity/settlement, and methodology changes
   • Flag counterparties where CVA exceeds materiality thresholds or PFE breaches credit limits
   • Identify wrong-way risk concentrations (e.g., FX derivatives with EM sovereign counterparties)
   • Document all model choices, overrides, and proxy mappings for audit trail

Output

• XVA summary table: CVA, DVA, FVA (FCA + FBA), ColVA, KVA, MVA by netting set and counterparty, with total portfolio XVA
• Exposure profile charts: EPE, ENE, and PFE term structures per netting set
• P&L attribution: Period-over-period XVA movement broken into risk drivers
• Sensitivity report: CS01, IR01, and vega for CVA hedging
• Methodology memo: Simulation parameters, model choices, proxy mappings, and known limitations
• Exception log: Counterparties with material wrong-way risk, missing netting opinions, or stale credit data

Quality Checks

• Confirm netting set definitions match signed ISDA/CSA documentation — mismatched netting inflates or deflates exposure
• Validate that EPE converges with increasing Monte Carlo paths (run convergence test; target coefficient of variation < 2% on portfolio CVA)
• Cross-check CVA against market-implied CDS spreads × EPE as a sanity benchmark
• Verify FVA does not double-count DVA — if both are reported, confirm the exposure legs are correctly separated
• Ensure KVA capital projection is consistent with the regulatory methodology actually approved for the entity [VERIFY — IMM vs. SA-CCR status]
• Reconcile sum of incremental XVAs to standalone portfolio XVA (non-linearity is expected but should be documented)
• Confirm collateral modeling reflects actual operational margin call frequency, not just contractual terms
• Review proxy credit spread mappings — ensure sector/region/rating proxies are reasonable and documented