A Critical Review Of The Basel Margin Of Conservatism Requirement In A Retail Credit Context

Abstract

The Basel II accord (2006) includes guidelines to financial institutions for the estimation of regulatory capital (RC) for retail credit risk. Under the advanced Internal Ratings Based (IRB) approach, the formula suggested for calculating RC is based on the Asymptotic Risk Factor (ASRF) model, which assumes that a borrower will default if the value of its assets were to fall below the value of its debts. The primary inputs needed in this formula are estimates of probability of default (PD), loss given default (LGD) and exposure at default (EAD). Banks for whom usage of the advanced IRB approach have been approved usually obtain these estimates from complex models developed in-house. Basel II recognises that estimates of PDs, LGDs, and EADs are likely to involve unpredictable errors, and then states that, in order to avoid over-optimism, a bank must add to its estimates a margin of conservatism (MoC) that is related to the likely range of errors. Basel II also requires several other measures of conservatism that have to be incorporated. These conservatism requirements lead to confusion among banks and regulators as to what exactly is required as far as a margin of conservatism is concerned. In this paper, we discuss the ASRF model and its shortcomings, as well as Basel II conservatism requirements. We study the MoC concept and review possible approaches for its implementation. Our overall objective is to highlight certain issues regarding shortcomings inherent to a pervasively used model to bank practitioners and regulators and to potentially offer a less confusing interpretation of the MoC concept.

A More General Quantum Credit Risk Analysis Framework

“…As stated in Section 1 , the single-factor model, as implemented in the Basel II framework, is intentionally designed to be conservative [ 14 ]. However, it has been recognized that this model has limitations, prompting large financial institutions to seek alternatives to measure risk more accurately.…”

“…While this model is useful, it is not optimal, especially for complex credit risk portfolios. In fact, though it helps to reserve an EC amount that suits every default scenario, it is intended to be a standard tool for CRA and therefore it is deliberately conservative [ 14 ]. Large financial institutions use custom models that consider several risk factors instead of just one since this refinement allows them to reserve a more precise amount to cover potential losses [ 15 ].…”

Towards practical Quantum Credit Risk Analysis

“…However, as already stated in Section 1, the Basel II single-factor model is deliberately conservative [13] and thus an extension of this approach is commonly used by large financial institutions to estimate default probabilities. The fundamental difference is the use of multiple systemic risk factors in place of just one, with the aim of directly attributing default correlations and, furthermore, default probabilities to the risk factors (while for the base model, given the realizations of the risk factors, defaults were uncorrelated).…”

“…However, the existing quantum algorithm was designed to work with the already mentioned canonical framework Basel II [1] based on an ASFR (asymptotic single-factor risk) model [12], which assumes that a borrower will default if the value of its assets were to fall below the value of its debts. Though this model helps to reserve an EC amount which suits every default scenario, it is intended to be a standard tool for CRA and therefore it is deliberately conservative [13], rather than an optimal solution, especially for complex Credit Risk Portfolios. As an example, Intesa Sanpaolo (Italy's largest bank by total assets [14]) uses a custom model that considers several risk factors instead of just one, since this refinement allows to reserve a more precise amount to cover potential losses [15].…”

Developing an Impairment Loss Given Default Model Using Weighted Logistic Regression Illustrated on a Secured Retail Bank Portfolio

“…The LGD can be modelled through either the direct or the indirect approach. When using the direct approach, the LGD is equal to one minus the recovery rate (De Jongh et al 2017). The indirect approach uses two components that are modelled separately, namely the probability component and the loss severity component.…”