Satellite models.
Turning macro scenarios into risk parameters.

The link between GDP, unemployment, rates and house prices on one side, and default and loss rates on the other. For the EBA stress test, ICAAP and climate scenarios.

Talk to a model specialist Feeds the EBA Filler →

Default rate · history and projectionIllustrative

ObservedBaselineAdverse

Regulatory basis

EBA EU-wide stress test methodology · SSM expectations for ICAAP

Portfolios

By segment, country and sector

Outputs

PD and LGD paths under baseline and adverse scenarios

Used for

EBA stress test, ICAAP, climate scenarios, IFRS 9

Approach

Statistically sound, and economically sensible.

Driver shortlist

Candidate macro variables per segment, matched to the scenario variables provided.

Specification

Transformations and lag structures searched by agents, ranked by fit and parsimony.

Economic plausibility

Sign constraints and sensitivity checks rule out models that fit but don't make sense.

Projection

Paths under each scenario, with the methodology's constraints applied.

Data

What you need. And what happens if you don't have it.

Data	Needed	If missing or incomplete
Default rate time series	Required · at least one full cycle	Pooled segments, with added conservatism
Macro history	Required	Public statistical series
Scenario paths	Required	Supervisory or internal scenarios
Segment and sector mapping	Required	Proposed by the agent for approval

In Model Studio 0.1.0

Every choice is a decision for a person. Every test has its threshold on the page.

Decisions the agents propose

Each comes with a recommended option, the reason and any data gap. The model owner approves the set at the design gate.

Driver shortlist
Which macro variables may enter the model?
Lag structure
Up to how many quarters may a driver lead the default rate?
Model size
How many drivers per model?
Economic plausibility
Are sign constraints enforced?
Out-of-sample test
How many recent quarters are held out?
Segments
Which segments get their own model?

Tests the engine runs

Shown next to each result. A test outside its threshold stays visible and becomes a limitation in the documentation.

Test	Group	Threshold
In-sample fit and residual diagnostics	Fit	Residuals stationary (ADF p < 0.10); autocorrelation handled by HAC errors
Out-of-sample performance	Forecast	RMSE reported; champion ≤ 1.2 × challenger
Sign consistency with economic theory	Plausibility	No coefficient with the wrong sign
Stability across sample periods	Stability	No sign change between halves
Comparison with challenger models	Challenger	Reported
Scenario sensitivity	Plausibility	Reported