Recent events (i.e. 2011-2012’s spread sovereign debt crisis or 2018’s crash in the stock market) forced, in the last decade, the need to enhance financial risk management by quantifying dependencies between extreme events. This turns out to be a mandatory points of emphasis, for instance, for insurance companies, which have to redesign their capital requirement in order to consider these kind of occurencies. Currently, no measure for extreme financial risks is agreed upon. Tail dependence is a very good concept to describe the degree of extreme co-movements, for instance, in asset prices. Estimates of the tail dependence coefficient can be found from different methods. This work aims to review some of the recent concepts for estimating tail dependence and implement those viewed as relevant in practical applications. We are going to concentrate on the analysis of some of these approaches with respect to historical and synthetic time series. We are going to focus on two estimators: the one introduced by Schmidt-Stadtmuller and another one by Caillault-Guégan, with a slight modification done to their algorithm, which defines the main point of this thesis.
Recenti eventi dell’ultimo decennio (i.e. la crisi del debito sovrano del 2011-2012 o il crollo dei mercati azionari alla fine del 2018) hanno evidenziato la necessità di migliorare la gestione del rischio finanziario quantificando le dipendenze tra eventi estremi. Questo risulta essere un punto di attenzione fondamentale, ad esempio, per le compagnie di assicurazione, che devono ridisegnare i loro requisiti patrimoniali per tenere conto di questo tipo di eventi. Attualmente non è stata concordata alcuna misura per rischi finanziari così estremi. La dipendenza di coda è un concetto molto valido per descrivere il grado di co-movimenti estremi, ad esempio, nei prezzi degli asset. Le stime del coefficiente di tail dependence possono essere ottenute con diversi metodi. Questa tesi si propone di passare in rassegna alcuni dei recenti concetti di stima della dipendenza di coda e di implementare quelli ritenuti rilevanti nelle applicazioni pratiche. Ci concentreremo sull’analisi di alcuni di questi approcci in relazione a serie storiche e sintetiche. Ci concentreremo su due stimatori: quello introdotto da Schmidt-Stadtmuller e un altro di Caillault-Guégan, con una leggera modifica apportata al loro algoritmo, che definisce il punto principale di questa tesi.
Estimating tail dependence in financial risk management: a comparative analysis of Schmidt-Stadtmuller and Caillault-Guégan approaches
Donadoni, Gabriele
2023/2024
Abstract
Recent events (i.e. 2011-2012’s spread sovereign debt crisis or 2018’s crash in the stock market) forced, in the last decade, the need to enhance financial risk management by quantifying dependencies between extreme events. This turns out to be a mandatory points of emphasis, for instance, for insurance companies, which have to redesign their capital requirement in order to consider these kind of occurencies. Currently, no measure for extreme financial risks is agreed upon. Tail dependence is a very good concept to describe the degree of extreme co-movements, for instance, in asset prices. Estimates of the tail dependence coefficient can be found from different methods. This work aims to review some of the recent concepts for estimating tail dependence and implement those viewed as relevant in practical applications. We are going to concentrate on the analysis of some of these approaches with respect to historical and synthetic time series. We are going to focus on two estimators: the one introduced by Schmidt-Stadtmuller and another one by Caillault-Guégan, with a slight modification done to their algorithm, which defines the main point of this thesis.File | Dimensione | Formato | |
---|---|---|---|
2024_12_Donadoni.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
1.85 MB
Formato
Adobe PDF
|
1.85 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/229517