Stress Testing Models
Executive Summary
Key Takeaways
- ✓ A stress test should vary the same volume, rate, and credit-loss drivers already present in the base model under a defined adverse scenario, reusing the base model's structure rather than building a separate, disconnected stress workbook.
- ✓ A macroeconomic scenario (a defined GDP contraction, unemployment rise, and rate shift, for example) needs to be translated into specific, quantified changes to the model's own volume, rate, and loss-rate assumptions before it can actually be applied.
- ✓ Credit loss provisions typically respond most materially to a stress scenario, since deteriorating macroeconomic conditions directly elevate segment-level expected loss rates in the loan portfolio.
- ✓ Stress test results should present the resulting capital ratios and liquidity metrics against the base case, showing the specific driver changes responsible for the difference, not just the stressed outcome in isolation.
- ✓ A stress test that cannot be reconciled back to the base model's structure cannot be relied upon, since a reviewer cannot confirm the stress result reflects a genuine, consistent variation of the same underlying model logic.
Objective¶
This guide covers how to build a bank stress test, within Banking Scenario Analysis, reusing the base model's structure rather than constructing a separate, disconnected workbook.
Reuse the Base Model Structure¶
A stress test should vary the same volume, rate, and credit-loss drivers already present in the base model under a defined adverse scenario. Building a separate stress workbook that does not share the base model's structure makes it impossible to confirm the stress scenario is a genuine, consistent variation of the same underlying logic, rather than a parallel, potentially inconsistent build that happens to produce a plausible-looking number.
Translating a Macroeconomic Scenario into Model Inputs¶
A macroeconomic scenario — a defined GDP contraction, unemployment rise, and interest rate shift, for example — is a narrative until it is translated into specific, quantified changes to the model's own drivers:
| Scenario Element | Model Driver Affected | Typical Direction of Change |
|---|---|---|
| GDP contraction / unemployment rise | Segment-level loan loss rates | Increase |
| Interest rate shift | Reference rates driving asset yields and deposit costs | Increase or decrease, per scenario design |
| Market stress / confidence shock | Deposit run-off and wholesale funding availability assumptions | Deposit outflow increase; wholesale funding availability decrease |
| Asset price decline | Collateral values supporting secured lending, loss given default | Decrease in collateral value; increase in loss given default |
Credit Loss Provisions Under Stress¶
Credit loss provisions typically respond most materially to a stress scenario, since deteriorating macroeconomic conditions directly elevate the segment-level expected loss rates built into Credit Loss Provisions and extended for scenario purposes in Loan Loss Forecasting. This is usually where the largest single driver of a stress scenario's capital impact originates.
Presenting Results Against the Base Case¶
Stress test output should present the resulting capital ratios (see Capital Adequacy Models) and liquidity metrics side by side with the base case, explicitly identifying which driver changes are responsible for the difference — not presenting the stressed figure alone, which gives a reader no way to assess which assumption changes actually produced the result.
Common Construction Pitfalls¶
- Building a separate, disconnected stress workbook rather than applying parameter variations to the existing base model.
- Applying a stress scenario narrative without translating it into specific, quantified driver changes.
- Understating the credit loss response to stress relative to other drivers, when it is typically the largest single component of the capital impact.
- Presenting the stressed outcome without a side-by-side base-case comparison identifying the specific drivers responsible for the difference.
Continue Reading¶
Prerequisites¶
- Banking Scenario Analysis — the parent guide
Related Technical Guides¶
How OXXON tests thisRun a free structural check with FMAE
Frequently Asked Questions
How should a bank stress test be built?
As a set of parameter variations applied to the same base model structure — flexing the same volume, rate, and credit-loss drivers already present in the base case under a defined adverse scenario — rather than as a separate, structurally disconnected stress workbook.
How is a macroeconomic scenario translated into model inputs?
By quantifying the specific effect the scenario implies for the model's own drivers — a defined GDP contraction and unemployment rise translated into a specific increase in segment-level loan loss rates, a specific change in deposit run-off behaviour, and a specific shift in reference interest rates, rather than left as a qualitative narrative disconnected from the numbers.
Which part of a bank model typically responds most to stress?
Credit loss provisions, since deteriorating macroeconomic conditions directly elevate segment-level expected loss rates in the loan portfolio, which is why the stress scenario's effect on loss rates needs particular care and granularity — see Loan Loss Forecasting.
How should stress test results be presented?
Against the base case, showing the resulting capital ratios and liquidity metrics alongside the base-case figures and identifying the specific driver changes responsible for the difference, not presenting the stressed outcome in isolation without that context.
Why does a stress test need to reconcile back to the base model?
Because a stress test built as a separate, disconnected workbook cannot be confirmed to reflect a genuine, consistent variation of the same underlying model logic — a reviewer cannot verify the stress result is meaningful if it was produced by an entirely different, potentially inconsistent calculation structure.
How does this guide relate to Banking Scenario Analysis?
Stress testing is a specific, typically regulator-defined application of the broader scenario analysis discipline — see Banking Scenario Analysis for the general approach to building and comparing multiple forward-looking scenarios in a bank model.
Related Articles
Banking Scenario Analysis
Scenario analysis in a bank model means building multiple forward-looking cases — a base case and one or more alternative cases — as parameter variations of the same underlying model structure, not as separate, disconnected workbooks. This guide covers how to structure a bank's scenario framework generally, how scenarios should be selected and switched cleanly, and how stress testing and loan loss forecasting fit as specific, more prescriptive applications of this same underlying discipline.
Loan Loss Forecasting
Loan loss forecasting extends the segment-level credit loss provisioning build into a forward-looking exercise, projecting how expected loss rates evolve across the forecast period as macroeconomic conditions and portfolio composition change. This guide covers how to structure that forward-looking loss-rate projection, how it should respond to defined economic scenarios, and how it connects the credit loss provisioning module to the base and stressed forecasts elsewhere in the model.
Capital Adequacy Models
Capital adequacy modelling represents the constraint regulatory capital requirements place on how much risk-weighted balance sheet a bank can carry against its available capital base. This guide covers how to structure a capital adequacy model — the capital tiers, the risk-weighted asset base they are measured against, minimum ratio and buffer requirements — and how it should be built as a live check against the balance sheet forecast rather than a standalone reporting exercise calculated after the forecast is already complete.
Balance Sheet Forecasting
Balance sheet forecasting is the central forward-looking exercise in a bank model: forecasting segmented asset volumes (loans, securities) and liability volumes (deposits, wholesale funding) period by period, then reconciling the two through an explicit funding plan. This guide covers how to structure that forecast, how to build the funding plan that closes any gap between asset growth and deposit growth, and how the forecast should be checked against capital adequacy and liquidity constraints rather than produced in isolation from them.
Basel Capital Ratios
The Basel III framework defines three core capital ratios — Common Equity Tier 1, Tier 1, and total capital — each measured against risk-weighted assets, layered with additional capital buffers above the hard minimums. This guide sets out the ratio definitions, the minimum and buffer levels the framework establishes, and how a bank model should represent each ratio and buffer as a distinct, named threshold rather than a single blended capital requirement.