Skip to Main content Skip to Navigation
Theses

Risque de crédit et interdépendance

Abstract : The aim of this thesis is the study of some important problems in credit risk, namely the monotony of transition matrices and interdependence in credit portfolios. On the first topic, we provide a new way to idealise completely and optimally empirical transition matrices in a geometric fashion. We study the produced monotone matrices through the distance from their associated empirical matrices using historical data. We prove in addition some theoretical results on the stability of monotony under classical transformations. On the study of interdependence, we introduce a Markov field model on a graphical formalism that takes into account exogenous factors and local interactions between the nodes that represents the firms of a credit portfolio. Our main idea is the introduction of relations that are functions describing the nature of the interaction between two firms. In that scheme, we study how risk parameters can be deformed under a type of relations, the topology of the considered network or a macroeconomic shock that can be local or global. Furthermore, we study numerically how critical phenomenon can arises in some sufficiently connected network configurations. We underline the role of this phenomenon in stress periods, and how contagion in that formalism can explain the observed cluster defaults in such periods. We prove many theoretical results on how risk can spread on certain class of networks, the explicit expression of loss distribution on regular networks and how the later can be used in the study of general large scale networks.We developed in addition some technics to overcome the limits that rise on large scale networks for calibration or default probability computing.
Document type :
Theses
Complete list of metadata

https://tel.archives-ouvertes.fr/tel-03407792
Contributor : Abes Star :  Contact
Submitted on : Thursday, October 28, 2021 - 4:26:26 PM
Last modification on : Thursday, November 18, 2021 - 10:31:10 AM

File

HJIRT.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-03407792, version 1

Collections

Citation

Mustapha Hjirt. Risque de crédit et interdépendance. Gestion et management. Université Panthéon-Sorbonne - Paris I, 2021. Français. ⟨NNT : 2021PA01E010⟩. ⟨tel-03407792⟩

Share

Metrics

Record views

68

Files downloads

55