Context and objectives
Climate risk assessment for large credit portfolios lies at the intersection of supervisory climate stress testing, internal risk management, and
model risk governance. Supervisory exercises rely on scenario-based climate narratives and portfolio-level aggregation of losses. Supervisory guidance explicitly emphasizes
governance and robustness of climate-related risk measurement.
A standard modelling chain starts from NGFS/SSP scenario trajectories and maps physical and transition drivers to corporate credit risk. In particular,
linking socioeconomic transition pathways (SSP-like narratives) to firms' dynamics and credit metrics has been formalized in portfolio credit settings. Beyond the ``mean path'', practical risk analysis requires a panel of stochastic scenarios (variability, correlations, extremes) rather than a single deterministic trajectory.
A second layer of complexity arises from production defaults: temporary or persistent breakdowns in productive capacity induced by extreme physical events, energy supply disruptions, or binding transition constraints. Such events affect cash flows upstream of credit defaults and introduce regime switching, discontinuities and path dependence.
Finally, transition risk is subject to deep (Knightian) uncertainty, especially regarding future carbon pricing. Ambiguity-aware corporate credit models have been proposed
to capture this dimension.
From a mathematical viewpoint, the resulting framework combines high-dimensional stochastic control and switching dynamics with robustness requirements. The objective of this project is not to introduce new economic mechanisms, but to use existing climate-to-credit model structures while developing numerically tractable, scalable, and auditable methods for (i) stochastic scenario generation, (ii) large-scale simulation, and (iii) robustness quantification of tail risk measures.
Project domain: sustainable finance, applied mathematics
Sponsors: MIRTE project of PEPR Math-VivES, PdR GRESC