ECB Economists Unveil Quantitative Toolkit to Measure Real-World Relevance of Alternative Forecast Scenarios

Economists at the European Central Bank are trying to answer a recurring policy problem: alternative scenarios are routinely published alongside the ECB/Eurosystem staff’s baseline projections, yet there is little systematic evidence on how “relevant” those scenarios are when judged against real-world macroeconomic behaviour. In a box prepared by Niccolò Battistini, Mohammed Chahad and Luís Fonseca for the ECB Economic Bulletin (Issue 7/2026), the authors set out a quantitative toolkit designed to evaluate scenarios by testing whether their implied economic outcomes look empirically plausible.

In the ECB’s projection exercises, the baseline is treated as the most probable path for the economy, given current conditions and a set of technical assumptions. Alternative scenarios are then used to explore policy-relevant narratives—such as potential shocks tied to geopolitical tensions—including, in this application, narratives related to the Middle East conflict. While such narratives can be useful for risk communication, research that formally assesses scenario plausibility remains limited, as noted in earlier work including Cordova-Pozo and Rouwette (2023).

The ECB authors propose a two-step framework. First, they build joint predictive distributions for key macroeconomic variables, meant to capture both how those variables evolve over time and how they move together. In their application, the variables are real GDP growth and HICP inflation. The aim is a single joint density for these outcomes, centred on the baseline projection, while allowing for the historical correlation between growth and inflation—both across time and between variables.

To construct that joint distribution, the framework uses two complementary approaches that start from univariate (single-variable) predictive distributions. One is a model-based route: Bayesian quantile regressions are used to estimate conditional distributions for each variable, and these are then combined into a joint density using the probability chain rule. The other is a copula-based route: a copula is used to link univariate marginal distributions into a joint distribution that reflects historical dependence patterns. For those marginals, the authors draw on two sources—historical macroeconomic data and a survey-based distribution from the ECB Survey of Professional Forecasters—so that forward-looking expectations from professionals are incorporated alongside past outcomes.

Once estimated, these joint densities are re-centred so that their most likely point (their mode) matches the baseline projection. The authors then compute these centred joint distributions over 20-year rolling windows beginning in the early 1980s. That rolling-window design serves two purposes: it allows the relationships between growth and inflation to change over time, and it lets each scenario be judged against the historical subsample most “favourable” to it, rather than against a single fixed period.

The second step translates the joint density into an empirical measure of plausibility. Following the logic of Doan et al. (1984), the authors treat plausibility as the extent to which a scenario’s projected path is unusual compared with the most likely trajectory. Concretely, for each historical subsample they compute the likelihood of the scenario’s joint path for growth and inflation and compare it with the likelihood of the baseline path. The resulting plausibility index for a scenario is the highest relative-likelihood score across all subsamples.

The index runs from 0 to 100: a value near 0 indicates that the scenario’s joint macroeconomic outcomes have essentially no historical precedent, while a value near 100 indicates that the scenario’s outcomes are about as plausible as the baseline’s. Importantly, the score is not meant to validate the underlying narrative events—such as the precise nature of an energy supply disruption or a particular tariff change. Instead, it evaluates whether the combination and co-movement of the macroeconomic outcomes implied by the scenario have been observed before, making it applicable even to narratives that have never literally occurred.

Why it matters

Scenario analysis influences how central banks communicate risk and how markets interpret policy choices. A tool that quantifies whether scenario outcomes resemble historically observed macroeconomic patterns can help users distinguish between informative stress tests and implausible storylines. The next step to watch is how such indices are used in practice—whether they shape which scenarios are published, and how much weight policymakers and investors give to them.