Identifying scenarios of interest under deep uncertainty

Working paper 860
Working Papers
Read aloud

Published: 20 April 2026

By: Jan Kwakkel Willem L. Auping Jan Willem van den End

We introduce Exploratory Modelling and Analysis (EMA) as a framework for macroeconomic scenario analysis under deep uncertainty. EMA traces scenarios of interest from policy‑relevant outcomes. These outcomes are forecasted by a model, based on a large set of uncertain conditioning variables. This approach flips the conventional method of scenario analysis, by sampling many scenarios without imposing strong ex‑ante priors on the choice, combination, value, or distribution of the conditioning variables. Applying EMA with an interacted VAR model to financial stress scenarios for the euro area shows that scenarios of interest can be uncovered that may be overlooked by common approaches for scenario analysis.

Keywords: Financial shocks; Scenarios; Exploratory modelling; Deep uncertainty
JEL codes E52; E58; G12

Working paper no. 860

860 - Identifying scenarios of interest under deep uncertainty

1.8MB PDF
Download 860 - Identifying scenarios of interest under deep uncertainty

Research highlights:

·       We introduce Exploratory Modelling and Analysis (EMA) as tool for macroeconomic scenario design.

·       We show how EMA enables systematic identification of policy relevant scenarios under deep uncertainty.

·       EMA is applied by sampling from a wide space of uncertain financial conditioning variables and tracing back scenarios of interest from the outcome space.

·       Based on different non‑linear VARX model specifications, we show that larger models with higher goodness of fit may omit scenarios of interest.

·       Moreover, restricting the set of conditioning variables too much increases the likelihood of overlooking scenarios of interest.

Discover related articles