We provide a comprehensive review across all key components of the credit stress testing framework:
Scenario Development and Risk Alignment
We assess the design and severity of macroeconomic and idiosyncratic scenarios to ensure they are tailored to your credit portfolio’s risk profile and business model. Scenarios typically include stressors like rising unemployment, declining property values, rate shocks, and sector-specific downturns (e.g., CRE, C&I, consumer).
Loan-Level Credit Loss Forecasting
Our team evaluates the stress testing engine’s ability to forecast expected credit losses at the loan level, including:
Probability of Default (PD) modeling under stress
Loss Given Default (LGD) assumptions and haircuts
Exposure at Default (EAD) projections
Incorporation of behavioral assumptions such as prepayments, curtailments, and utilization of unfunded lines
Segmentation and Risk Drivers
We review the appropriateness of loan segmentation, risk drivers, and model inputs such as LTV, DSCR, FICO, internal risk ratings, and industry classifications. Our validations ensure that credit deterioration is realistically modeled based on historical relationships and forward-looking macroeconomic paths.
Assumptions and Model Transparency
We examine the credibility and support for all key assumptions, including credit migration paths, cure rates, recovery lags, and portfolio dynamics under stress. We assess whether overlays or qualitative adjustments are documented, repeatable, and grounded in analysis.
Capital Impact and Integration
We validate the linkage between loan-level losses and projected capital impacts, ensuring stress results appropriately flow into earnings, allowance, and capital forecasts. This includes alignment with CECL provisioning, capital buffers, and internal planning thresholds.
Governance, Documentation & Policy Alignment
Our review includes the model’s alignment with your credit risk management framework, stress testing policies, and capital planning processes. We assess governance controls, version tracking, change management, and the clarity and completeness of model documentation.
Modeling Platforms & Environments
We work with a variety of modeling platforms, including:
In-house loan-level stress testing engines
CECL-integrated stress testing models
Vendor tools embedded within credit or ALM systems
Python, R, SAS, or Excel-based models with statistical or rule-based approaches
