Credit risk modelling under IFRS 9 : a practical and universal tool for the estimation of lifetime probability of default

Banks are required to analyse the credit risk impact of the macroeconomic scenarios in terms of both the capital available (via impairments and thus P&L) and the Risk Exposure Amount (also known as Risk-Weighted Exposure Amount alias RWA) for positions exposed to risks stemming from the default of counterparties, making use of their models and considering a number of conservative constraints. The estimation of credit impairments requires the use of statistical methods and includes the following main steps: (i) estimating starting values of the risk parameters, (ii) estimating the impact of the scenarios on the risk parameters, and (iii) computing changes in the stock of provisions that will drive the P&L impact. Banks are required to forecast credit impairments resulting from the materialisation of different scenarios on the basis of IFRS 9, that is an International Accounting Standard Board's (IASB) response to the 2008 Global Financial Crisis. The objective is to improve the accounting and reporting of financial assets and liabilities post financial crisis. Therefore, the main focus is to predict loss recognition by avoiding financial issues faced during global recession. Dynamic aspects are crucial though. IFRS 9 requirements raise new issues regarding dynamic and long term risk assessment. Plenty of information is available for calibrating Probability of Default curves (scores, risk classes, risk class migrations, observed defaults, delinquencies) and there is a large set of statistical methods at hand as well. However, there are financial entities that still have big problems on collecting data to be used as inputs in credit risk modelling. This thesis aims to develop a practical tool to compute Lifetime Probability of Default, focusing on retail portfolios, even when enough data are not available.

Le istituzioni finanziarie sono chiamate a detenere capitale per assorbire eventuali perdite causate dall'incorrere di un evento di default. In aggiunta agli accordi di Basilea, dal 2018 è in vigore il nuovo principio contabile IFRS 9. Il modello di impaiment introdotto dall' IFRS 9 prevede il calcolo della perdita attesa seguendo un approccio più dinamico, per l'intera durata del credito. La perdita attesa lifetime e forward looking è basata sulla stima unbiased di parametri di rischio quali Probabilità di Default e Loss Given Default, risultanti quindi da realistiche analisi di scenario. Questa tesi si fornisce una metodologia pratica e universale per la stima delle Probabilità di Default IFRS 9 compliant.