In the constantly evolving regulatory landscape of counterparty credit risk management, the Basel Committee on Banking Supervision (BCBS) has recently introduced two new models for the computation of capital requirements for Credit Valuation Adjustment (CVA) risk. This thesis focuses on those newly established approaches: the simple and conservative Basic Approach (BA-CVA) and the more advanced and risk-sensitive Standardized Approach (SA-CVA). SA-CVA offers a more precise capital charge calculation, based on transaction-level sensitivities, which can recognize hedges on all types of financial instruments, but its implementation is highly complex and expensive. As a result, many banks opt for the simpler and cheaper BA-CVA, missing out on the advanced framework's benefits. A major downside of SA-CVA is the computational cost of calculating sensitivities, often achieved using the Bump and Reprice method. In this work, an alternative approach based on the Likelihood Ratio Method (LRM) is presented, hoping to boost efficiency in computing said CVA sensitivities. The method is tested on simple financial instruments, with results highlighting its potential as a viable alternative to Bump and Reprice. Despite these improvements, significant concerns still remain regarding the practical adoption of the advanced approach. Case studies reveal that due to a regulatory supervisor discount factor that reduces capital requirements by 35%, the basic approach often results in lower capital charges than the advanced model. This undermines the intended role of BA-CVA as a conservative fallback option, raising questions about the incentives for banks to adopt SA-CVA. The findings of this thesis highlight both the mathematical feasibility of improving SA-CVA calculations and the broader regulatory inconsistencies that may influence banks' strategic decisions on CVA risk management.
Nel panorama normativo in continua evoluzione della gestione del rischio di controparte, il principale organo regolamentare (BCBS) ha recentemente introdotto due nuovi modelli per il calcolo dei requisiti patrimoniali relativi al rischio di CVA (Credit Valuation Adjustment). Questa tesi si concentra su questi nuovi approcci in vigore da Gennaio 2025: il semplice e conservativo BA-CVA e il più avanzato e più sensibile al rischio SA-CVA. Quest'ultimo fornisce un calcolo più preciso del requisito patrimoniale, basato sulle greche di ogni transazione, permettendo il riconoscimento delle coperture su tutti i tipi di strumenti finanziari. Tuttavia, la sua implementazione è complessa e computazionalmente onerosa. Uno dei principali svantaggi di SA-CVA è l'elevato costo computazionale associato al calcolo di tali greche, spesso ottenute tramite il metodo Bump and Reprice. In questo lavoro viene presentato un approccio alternativo, basato sul Likelihood Ratio Method, con l'obiettivo di migliorare l'efficienza nel calcolo appena citato. Tale metodo è stato testato su strumenti finanziari semplici e i risultati ottenuti evidenziano il suo potenziale come valida alternativa al Bump and Reprice. Gli studi condotti rivelano che, a causa di un fattore di sconto, imposto da BCBS, che riduce i requisiti del 35%, l'approccio base spesso porta a risultati finali inferiori rispetto al modello avanzato. Per tale motivo, molti dubbi sorgono riguardo alla definizione del modello base, nato come un'alternativa semplice ma conservativa, che ora finisce per produrre, nella maggior parte dei casi, risultati addirittura inferiori rispetto a quelli di SA-CVA. In conclusione, questa tesi evidenzia da un lato come i complicati calcoli matematici relativi al modello avanzato possano essere migliorati e ottimizzati, mentre dall'altro come la normativa proposta non sia completamente coerente e, molto probabilmente, influisca sulle decisioni delle banche nella gestione di questo tipo di rischio, indipendentemente da quanto il modello avanzato sia efficiente.
Optimizing CVA sensitivities computations using likelihood ratio method
MONTI, FILIPPO
2023/2024
Abstract
In the constantly evolving regulatory landscape of counterparty credit risk management, the Basel Committee on Banking Supervision (BCBS) has recently introduced two new models for the computation of capital requirements for Credit Valuation Adjustment (CVA) risk. This thesis focuses on those newly established approaches: the simple and conservative Basic Approach (BA-CVA) and the more advanced and risk-sensitive Standardized Approach (SA-CVA). SA-CVA offers a more precise capital charge calculation, based on transaction-level sensitivities, which can recognize hedges on all types of financial instruments, but its implementation is highly complex and expensive. As a result, many banks opt for the simpler and cheaper BA-CVA, missing out on the advanced framework's benefits. A major downside of SA-CVA is the computational cost of calculating sensitivities, often achieved using the Bump and Reprice method. In this work, an alternative approach based on the Likelihood Ratio Method (LRM) is presented, hoping to boost efficiency in computing said CVA sensitivities. The method is tested on simple financial instruments, with results highlighting its potential as a viable alternative to Bump and Reprice. Despite these improvements, significant concerns still remain regarding the practical adoption of the advanced approach. Case studies reveal that due to a regulatory supervisor discount factor that reduces capital requirements by 35%, the basic approach often results in lower capital charges than the advanced model. This undermines the intended role of BA-CVA as a conservative fallback option, raising questions about the incentives for banks to adopt SA-CVA. The findings of this thesis highlight both the mathematical feasibility of improving SA-CVA calculations and the broader regulatory inconsistencies that may influence banks' strategic decisions on CVA risk management.File | Dimensione | Formato | |
---|---|---|---|
2025_04_Monti.pdf
accessibile in internet per tutti a partire dal 10/03/2026
Descrizione: Tesi di Laurea Magistrale
Dimensione
1.97 MB
Formato
Adobe PDF
|
1.97 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/10589/236119