Basel IV: What influence do PD/LGD input floors have on RWAs?

Background

The now-completed second stage of the Basel III framework (“Basel IV”) proposes a number of reforms that will make banks more resilient and improve the consistency and comparability of the calculation of risk-weighted assets (RWAs) across banks. Given the backdrop of the previous global financial crisis, regulators took it as their central task to create a new framework that increases credibility of and mitigates risks within the international banking system.

This article takes a careful look at the new Basel IV “input floors” for the probability of default (PD input floors) and the loss given default (LGD input floors) in IRBA models. These are key dimensions which are used to limit excessive RWA variability resulting from banks’ use of internal risk-based approaches (IRBA).

What are input floors and why are they used?

The idea of input floors is based on the nature of the historical statistical data used in the models. The lower a parameter value is, e.g. at a very low probability of default of 0.03%, the greater the number of observations a bank needs to statistically model and validate that parameter. If the number of observations is too low, it means that there is a significant risk in which the probability and extent of the associated risks for the bank are underestimated.

It was precisely in their analyses of the causes of the financial crisis that regulatory authorities found that banks do not always have sufficient historical default observations. Among the many aspects of the overall framework of Basel III, supervisors believe that input floors are an important aspect to increase the robustness and risk sensitivity of the IRBA models used in RWA calculations. The Basel IV framework therefore proposes to increase the starting point for the risk components PD and LGD.

What is changing and what is the impact?

The Basel IV framework largely sets the PD input floor for exposure classes at 5 basis points (0.05%). This is an increase from the input floor of 3 basis points (0.03%) set in the Basel II framework. It also sets LGD input floors at values ranging from 25% to 50% for the unsecured part of a credit exposure, and from 0% to 15% for the secured part of the credit exposure.

Influence of PD input floors

In order to gauge the expected impact of the Basel IV reforms, the European Banking Authority (EBA) conducted a survey of the leading European financial institutions. In terms of the impact of PD input floors, the study found that the greatest impact on RWA will be on corporate loans, particularly loans to financial institutions and large corporate customers (Figure 1). This may seem counterintuitive at first glance, as these risks were considered fundamentally low under Basel II. However, given the small sample size of the historical data behind each of these asset classes and combined with high exposure allocation, this result seems plausible and consistent with the reforms proposed by regulators.

Note: Based on a sample of 48 banksSources: EBA 2018-Q2 QIS data and EBA calculations

Impact of LGD input floors

In the same study, the results suggest that LGD input floors for IRBA positions will be even more significant. In particular, the impact of the proposed LGD input floors on RWA will be particularly noticeable for exposures to specialist banks, corporate SMEs, small and medium-sized enterprises and most categories of retail exposures (Figure 2). Since experience shows that LGD modelling is generally based on poorer data quality than PD modelling, the generally higher impact of LGD input floors on RWA results seems feasible.

Note: Results are subject to the F-IRB approach under the revised framework so that there is no difference in impact for these portfolios due to the LGD input floor. Specialist credit exposures under slotting are not subject to the PD/LGD parameters
Sources: EBA 2018-Q2 QIS data and EBA calculations

Outlook

The EBA study has shown that the IRBA input floors for PD and LGD will have a significant impact on RWA. The participating banks have clearly indicated increased RWA values based on the PD and LGD input floors. In this respect, it appears that the regulatory authorities will achieve their goal of increasing RWA results.

However, this situation may lead banks to limit the impact of regulation by restructuring their portfolios. For example, the expected increase in input floors could potentially increase the risks among banks if they decide to include larger risks in their books to achieve better returns but similar capital requirements based on the new input floors.

Ultimately, we will all see the decisions of individual banks when Basel IV comes into force on January 1, 2022. But already now we can help financial institutions to make a strategic decision on the partial application of CRSA/IRBA, to redevelop internal models, to improve early warning systems and to optimize the separation power and the result of RWA by portfolio restructuring.

Many thanks to the co-author Christopher Beck.

Capgemini Invent addresses the CxO data strategy and supports its clients in data-driven value creation.

With Inventive Finance, Risk & Compliance (Inventive FRC) we master the challenges of the finance, risk, and compliance function in the financial sector. This blog article focuses on credit risk.

Grigori Tymchenko

Expert in Financial Risk Modeling, IFRS9, Validation and Management

I have 12 years experience in Financial Risk modelling and validation, compliance, data management as well as development of early warning systems based on open data.
I delivered numerous projects on Basel IRB, IFRS9, market, operational, business and model risk topics at financial institutions in and outside of the EU.
Recent work includes:
Basel IV readiness assessment and designing the transition roadmap for G-SIB in Europe to achieve regulatory compliance.
Development of early warning systems and KYC checks for credit risk and market compliance monitoring based on structured and unstructured public information: social media data, news, blogs etc. for internationally operating large banks in Germany to improve separation power of existing rating/scoring systems.
Risk & regulatory management content, credit risk model development and validation incl. backtesting and ongoing monitoring (PD/LGD/EaD/CCF retail/non-retail under ECB/EBA-GL, esp. DoD, MoC, TRIM), automation of reporting in R, SAS as well as closure of regulatory findings, action point control, model tiering and model risk management for large banks in Europe.
IFRS9 gap analysis incl. road map design, implementation planning and budgeting, followed by complete implementation projects esp. PD/LGD/EaD modelling for IFRS9 purpose in R, SQL as well as model validation in R, SAS, SQL, external and internal data management, software selection at large European banks.
Before joining Data, Risk and Regulatory of Capgemini Invent, I have been working for several years at KPMG and Deutsche Bank in risk analytics and regulatory area.
I am fluent in German, English, Russian and Ukrainian. I like travelling a lot, as well as going for biking and surfing.