6.3. Ongoing monitoring and expert judgment

To enhance the robustness of their models, institutions should consider challenging their calibration approach by:

• a. comparing their results and assumptions with external, including supervisory, observations from credible sources to assess the consistency of their own assumptions and results;
• b. using sensitivity analyses to test the degree of stability and consistency of their models’ outputs or to identify the effect of potential non-linearities not included in the scenarios;
• c. where a third-party model is used, verifying that the validation framework of the external suppliers complies with the EBA Guidelines on outsourcing arrangements.

To address the residual shortcomings of their stress test models, institutions should consider accounting for the impacts of the factors that could not, at this stage, be otherwise integrated (e.g. risks stemming from counterparties’ value chain, tipping points, contagion effects, etc.), by adjusting conservatively the results of their models on the basis of expert judgement.

More generally, institutions should use expert judgement when carrying out quantitative analyses to compensate for incomplete or approximate environmental data, the absence of observed historical correlations and other model limitations.

Institutions should ensure regular monitoring of significant developments in their environment (including counterparties’ strategy to cope with environmental risks) so that the scenarios and modelling approaches used remain relevant. The frequency with which scenario analyses are carried out should be adapted to the needs and practices of the institutions.

Scenario analysis should be designed with adaptability and modularity in mind to allow for ongoing refinements as the environment and knowledge evolve. Institutions should keep abreast of the latest scientific knowledge. FINAL REPORT ON GUIDELINES ON ENVIRONMENTAL SCENARIO ANALYSIS