Stress Testing under CCAR and DFAST

1. CCAR and DFAST: Regulatory Context

The Comprehensive Capital Analysis and Review (CCAR) and Dodd-Frank Act Stress Testing (DFAST) are US Federal Reserve programs requiring large banking organizations to demonstrate capital adequacy under adverse economic scenarios. While Brazilian institutions are not directly subject to these frameworks, understanding CCAR/DFAST methodologies offers valuable insights for:

• ICAAP (Internal Capital Adequacy Assessment Process) under Basel III Pillar 2.
• Bacen stress testing requirements for systemically important institutions.
• IFRS 9 forward-looking scenarios incorporating severe downturns.
• Strategic capital planning and risk appetite calibration.

2. Scenario Architecture: Baseline vs. Adverse

Federal Reserve provides three scenarios annually with 9-quarter projection horizons:

• Baseline: Consensus macroeconomic forecast (modest growth, stable unemployment, gradual rate normalization).
• Adverse: Moderate recession (GDP contraction -2%, unemployment +3%, equity decline -25%).
• Severely adverse: Global financial crisis-level stress (GDP -5%, unemployment +5%, equity -50%, housing prices -25%).

Each scenario includes 28 variables: GDP, unemployment, interest rates, equity indices, house prices, commodity prices, VIX, exchange rates—enabling granular portfolio impact modeling.

3. Credit Loss Projection Methodology

Loan-level modeling:

• PD stress models: Logistic regression or survival models linking borrower default to macro variables (unemployment, housing, income).
• LGD stress adjustments: Collateral values decline under adverse house price scenarios; recovery rates compress as workout capacity saturated.
• EAD dynamics: Credit line utilization increases as liquidity tightens (CCF multipliers for revolving products).

Portfolio aggregation:

• Segment portfolios by product (mortgage, credit card, auto, commercial) and risk band.
• Apply stressed PD/LGD/EAD to exposures; aggregate losses by quarter over 9-quarter horizon.
• Incorporate new originations at stressed credit quality (adverse selection during downturns).

4. Pre-Provision Net Revenue (PPNR) Projection

PPNR represents revenue and expenses before credit losses—critical for determining capital generation capacity under stress:

• Net interest income (NII): Balance sheet projections by product × stressed rates × margin compression.
• Non-interest income: Fee income, trading revenue, servicing fees—typically decline during stress.
• Non-interest expense: Personnel, technology, occupancy—model as fixed plus variable components.
• Operational risk losses: Scenarios may include litigation or fraud spikes.

PPNR projection quality separates sophisticated institutions from peers—granular business line models with macro sensitivities outperform simplistic historical averages.

5. Capital Ratio Projections

Translate losses and PPNR into projected capital ratios:

• Starting capital position: CET1, Tier 1, Total Capital at stress test start date.
• Capital depletion: Credit losses + operational losses + trading/counterparty losses - PPNR - taxes.
• RWA dynamics: Risk-weighted assets increase as credit quality deteriorates (migration to higher risk weights).
• Capital actions: Dividends, buybacks (constrained under stress), issuances—model per planned capital policy.

Key output: Minimum capital ratio during 9-quarter stress period. Pass threshold: CET1 >4.5% plus stress capital buffer.

6. Qualitative Assessment: Model Risk and Governance

Beyond quantitative projections, supervisors evaluate:

• Model documentation: Clear specification of PD/LGD/PPNR models, variable selection, estimation methodology.
• Validation rigor: Independent review of stress models, back-testing against historical crises, sensitivity analysis.
• Data quality: Completeness of borrower attributes, historical default data spanning full cycles.
• Governance processes: Board oversight of stress testing, integration into capital planning, remediation of prior-year findings.
• Business integration: Stress results inform risk appetite, pricing, portfolio limits—not just compliance exercise.

7. Common Pitfalls and Supervisor Expectations

• Mechanical application: Plugging macro variables into models without considering second-order effects (customer behavior shifts, market dysfunction).
• Data limitations: Sparse default data for low-default portfolios—supervisors expect expert judgment overlays with documentation.
• Concentration risk: Failing to model correlated defaults within sectors (energy, real estate) during tailored scenarios.
• Optimistic PPNR: Assuming flat expenses or unrealistic revenue resilience—expense flexibility and business model stress must be credible.
• Capital action rigidity: Maintaining dividends under severe stress—regulators expect conservatism when ratios approach minimums.

8. Adapting CCAR Principles to Brazilian Context

ICAAP integration:

• Develop institution-specific severely adverse scenario reflecting Brazil macro risks (exchange rate shock, commodity collapse, political crisis).
• Project credit losses over 3-year horizon per Bacen ICAAP guidelines, linking to IFRS 9 ECL models.
• Assess capital adequacy including Pillar 2 add-ons for concentration, interest rate risk, operational risk.

Forward-looking provisioning:

• Use CCAR-style scenarios (base, optimistic, pessimistic) for IFRS 9 ECL probability weighting.
• Calibrate satellite models linking Selic, GDP, unemployment to PD/LGD adjustments by segment.
• Document scenario selection rationale and governance approval for auditor/supervisor review.