Duration GAP Analysis, stress testing, and backtesting are foundational tools in a bank’s Asset-Liability Management (ALM) toolkit to manage interest rate risk and protect both earnings and economic value of equity (EVE). Structured measurement reports connect these analytics to governance, limits, and action. This article outlines a cohesive approach suitable for policy frameworks and ALCO reporting in modern banking environments.
Duration GAP Analysis
Duration GAP measures the sensitivity mismatch between asset and liability cash flows by comparing their effective durations to infer the impact of parallel and non-parallel rate shocks on EVE. A positive GAP (assets’ duration > liabilities’ duration) indicates equity value is more sensitive to rate increases, while a negative GAP suggests vulnerability to falling rates. Duration should be effective, accounting for embedded options (prepayments, early withdrawals), convexity, and behavioral adjustments. The leverage-adjusted duration GAP, together with the size of the balance sheet and assumed shock magnitudes, provides a tractable approximation of EVE change under rate moves. This method complements repricing gap and simulation approaches and is widely used in ALM for interest rate risk governance.
Measurement System Reports
ALM measurement reports should translate analytics into decision-ready management information. Well-designed packs generally include:
- Risk profile overview: Summary of EVE and NII sensitivities under prescribed supervisory and internal scenarios, limit utilization, and trend lines.
- GAP views: Duration GAP by major books and at consolidated level; repricing buckets with RSA/RSL and cumulative gaps; option-adjusted metrics where relevant.
- Scenario analytics: Parallel up/down shocks, steepeners/flatteners, basis shifts, and key rate durations; behavioral assumptions, prepayment vectors, and deposit betas.
- Hedging and positioning: Derivatives inventory mapping (pay/receive profiles), hedge effectiveness, basis and convexity effects, and residual exposures versus risk appetite.
- Liquidity and funding linkages: Overlap to liquidity coverage, structural funding metrics, and contingency capacity.
- Governance artifacts: Limit breaches/exceptions, model changes, assumption overlays, and action items for ALCO.
These reports must be consistent, reconciled to the GL/balance sheet, version-controlled, and timely, with clear ownership and data lineage.
Stress Testing
Stress testing probes portfolio resilience to severe but plausible rate and market structure shifts, extending beyond base sensitivities. A robust regime covers:
- Rate path stresses: Large parallel shocks, twists (bear steepener, bull flattener), key-rate shocks, and basis dislocations across benchmark curves.
- Behavioral stresses: Accelerated mortgage prepayments, surge in early deposit run-off, altered deposit betas/asymmetry, and option exercise under stress.
- Market/liquidity overlays: Wider spreads, lower hedge market depth, increased funding costs, and reduced collateral valuations.
- Multi-risk coherence: Interaction with credit migration (through discount rates and customer behavior), liquidity outflows, and margining requirements on hedges.
- Reverse stress tests: Identify scenarios that would breach capital or risk limits, quantifying buffers and management actions required to restore within appetite.
Outputs should quantify EVE and NII impacts, convexity effects, hedge performance under stress, and management levers (rebalancing, hedging, pricing actions). Documentation should record scenario construction, parameter sources, expert overlays, and validation status.
Backtesting
Backtesting evaluates the accuracy and stability of IRRBB models and key assumptions by comparing realized outcomes with prior projections. A practical framework includes:
- EVE/NII forecast versus actual: Attribution of variances into rate path differences, volume/mix changes, behavioral divergences (prepayments, betas), hedge P&L, and model error.
- Assumption validation: Periodic tests of deposit decay and beta models, prepayment curves, early redemption rates, and option exercise; recalibration triggers and governance.
- Model performance metrics: Out-of-sample errors, stability of parameter estimates, sensitivity to data windows, and challenger model comparisons.
- Controls and change management: Independent model validation, performance thresholds, remediation plans, and ALCO escalation for material deviations.
Backtesting should feed directly into model risk management, with tracked remediation actions and effective dating of model changes.
Putting It Together: A Cohesive ALM Cycle
- Measure: Maintain reconciled duration GAP, repricing gaps, and simulation-based NII/EVE sensitivities; compute convexity and key-rate durations for richer insight.
- Report: Deliver standardized measurement packs to ALCO with clear narratives, trend analytics, and limit consumption; ensure drill-downs by product and hedge.
- Stress: Run severe but plausible scenarios, reverse stresses, and combined risk overlays; quantify management actions and timings.
- Validate: Backtest forecasts and assumptions; recalibrate and re-approve models per policy.
- Act: Execute hedges, rebalance portfolios, adjust pricing/betas, and refine funding mix within risk appetite and capital constraints.
Practical Considerations for Banks
- Behavioral modeling matters: Non-maturity deposits and prepayable assets dominate IRRBB; option-adjusted durations and robust beta/decay models are critical.
- Convexity and basis risk: Duration-only views can understate exposure; incorporate convexity and curve/basis stresses to avoid hedge slippage.
- Data and lineage: Golden sources, reconciliation to finance, and transparent assumption inventories raise trust and auditability.
- Governance: Define risk appetite limits for both EVE and NII; enforce breach protocols, with ALCO accountability and Board-level oversight.
- Integration: Link ALM outputs to pricing, funds transfer pricing (FTP), budgeting, and strategic planning to convert risk insight into performance outcomes.
If a formatted Word-ready version with a model policy appendix, sample ALCO dashboard outlines, and a backtesting checklist is needed, that can be provided.
Articles related to Risk Management in ‘Model ‘A’ of CAIIB –Elective paper:
