Financial Modeling of Climate-related Risks

Thesis defense presentation by Théo Le Guenedal

Abstract:

This research project aims at estimating financial risks related to climate change. Beyond the applications and quantitative findings, the main objective of the chapters of this thesis is to provide a structural and methodological framework that is generalizable, in order to facilitate their integration by practitioners.

The first chapter proposes a bottom-up measure of transition risk, which can be incorporated with classical risk models (Merton or credit risk model). This "cost-based" approach is limited to the directly polluting sectors, which leads to the second chapter, which allows for the diffusion of transition risk through the value chain. These approaches offer a static structure that allows for a fixed scenario stress-test but not for pricing the bonds by considering heterogeneous scenarios and the probability of realization. To this end, chapter three proposes a pricing model that integrates a Bayesian approach in updating scenario probabilities based on observed jumps in carbon pricing mechanisms. Finally, the last chapter proposes a Monte-Carlo methodology for simulating annual damages caused by tropical cyclones. The conversion of raw climatic data into a synthetic database of losses is achieved by coupling statistical and thermodynamic relationships. The exposure of physical assets, the dynamics of socio-economic factors, local population densities and specific vulnerabilities in different regions of the world are borrowed from different segments of the literature, and combined to obtain a complete model of the classical triptych necessary for the study of physical hazards: hazard intensity x exposure x vulnerability generalizable and homogeneous across countries. The resulting signal can then be simply included in credit risk models equating annualized damages with additional debt.