Credit risk measurement: avoiding unintended results Part 1

RMA Journal, The, April, 2004 by Peter O. Davis

This article--the first in a series--provides an overview of credit risk measurement terminology and a basic risk measurement framework. The series focuses on credit risk measurement concepts, how they are implemented, and potential inconsistencies between theory and application. Subsequent articles will present common implementation options for a given concept and examine how each affects the credit risk measurement result.

The basic concepts of credit risk measurement--default probability, recovery rate, exposure at default, and unexpected loss--are easy enough to describe. But even for people who agree on the concepts, it's not so easy to implement an approach that is fully consistent with the starting concept. Small differences in how credit risk is measured can result in big swings in estimates of credit risk--with potentially far-reaching effects on risk assessments and business decisions.

Trend Toward Credit Risk Quantification

As credit risk modeling methodologies have improved over time, banks have incorporated models into risk-grading, pricing, portfolio management, and decision processes. As the role of credit risk models has grown in significance, it is important to understand the different options for measuring individual credit risk components and relating them for a complete measure of credit risk.

Consumer lenders, in particular, rely heavily on borrower scorecards to lower underwriting costs and improve portfolio management. Quantification of commercial credit risk has moved forward at a slower pace, with a significant acceleration in the past few years. The relative infrequency of defaults and limited historical data have constrained model development for commercial credit risk.

Although vendors offer default models for larger (typically public) firms, quantifying the risk of small business credits remains a challenge for banks where there is a wild between scorecard-driven approaches to retail credits and formally graded large facilities.

Basel II as a driver. The introduction of dual rating systems that would meet the criteria proposed in Basel II has accelerated the interest in credit models to support risk-grading frameworks. Credit models have been introduced to explicitly measure borrower default risk and the risk of loss given default on individual loan facilities. Basel II has been a catalyst for banks to develop and or strengthen measures of loss given default. Significant improvements in LGD modeling approaches can be expected over the next few years as more data becomes available and institutions validate recently implemented modeling/grading methodologies.

The development of robust economic capital frameworks has allowed banks to measure credit risk concentrations and the marginal contribution of new credit exposures. As institutions have become more comfortable with their economic capital frameworks, risk-adjusted performance measurement has become a key driver for assessing customer profitability, measuring business line performance and setting senior management compensation. Basel II also will be a catalyst for the introduction and enhancement of economic capital frameworks. To qualify for the advanced approaches under Basel II, banks must be able to measure economic capital and use this information to support business decisions.

Credit Risk Measurement Conceptual Framework

The general framework for measuring credit risk is simple. Credit risk can be divided into two components: expected loss and unexpected loss. As formalized in Pillar I of Basel II, expected losses are calculated as:

Exposure x Probability of Default (PD) x Loss Given Default (LGD) x Exposure at Default (EAD) = Expected Loss (EL)

Historically, banks have combined the risk of PD and LGD into a single measure. Rather than having dual ratings systems (one rating for PD and the other for LGD), most banks had a one-dimensional rating system. Borrower and facility were considered together in assigning a rating, which was the product of the probability of default (PD) and the fraction of the loan's value likely to be lost in the event of default (LGD). Typically, strong collateral supporting a facility could notch up an obligor risk rating. These rating systems were commonly used to rank-order risk, rather than to link ratings to explicit measures of credit loss.

The last term in the formula for expected loss--exposure at default (EAD)--is necessary for commitments and revolving credits, which may be undrawn, fully drawn, or anywhere in between. (Depending on how an institution manages its lines, a borrower may not be able to draw as it nears default or the borrower may fully draw down on the line.)

If exposure-weighted default rates are used, then EAD is already implicitly captured in the default risk component. (An exposure-weighted default rate measures the default amount as percent of the outstanding portfolio balance.) If incidence-weighted default rates are used (as required for Pillar I internal ratings calculations under Basel II), then outstanding exposure at the time of default must be measured separately in an EAD calculation. (An incidence-weighted default rate measures the number of defaults as a percent of the number of active borrowers.)