Loss Given Default (I)

“What matters isn’t what a person has or doesn’t have; it is what he or she is afraid of losing.” Nassim Nicholas Taleb, Skin in the Game

In an introductory article I already referred the concepts of LGD (1-RR) and Recovery Rates (1-LGD). In this we go deeper in the models for LGD.

Please note that LGD stands for Loss Given Default, meaning that the loss is conditional on the default event. So, to clearly quantify the LGD, the first step is to have a well-defined default concept. If you use different definitions of default, then there will be inconsistencies between the default rates, and the loss or LGD values.

Usually, a bank will distinguish among different types of defaults. An operational default is due to technical issues on the obligor side. For example, an obligor is accidentally late when making the payment. A technical default is a default due to an internal information system issue. For example, the payment was made on time, but on the wrong account.

For the purpose LGD modelling we should focus on ‘real’ defaults, the ones due to financial problems or insolvency. Operational or technical defaults related to technical issues or internal information system issues should not be considered.

In case of default, different outcomes can take place:

1. Cure: a defaulter will pay back all outstanding debt and return to a performing/non defaulter status.
2. Restructuring: the bank and the defaulter work out a recovery or repayment plan, as for example an extension of the loan maturity.
3. Liquidation/repossession/foreclosure: the bank takes full possession of the collateral asset, if available, and sells it by starting up a bankruptcy procedure.

Definition of LGD

The loss given default can now be defined as the ratio of the loss on an exposure due to the default of an obligor to the amount outstanding at default.

The loss concept referred in the LGD is an economic loss and not an accounting loss, therefore all costs and benefits, need to be properly considered when defining the LGD. Examples of costs are the the costs for realizing the collateral value, administrative costs of the recovery process or legal costs. Examples of benefits are interest on arrears, penalties for delays or other commissions.

LGD can be measured using various methods:

1. Workout method: used for both corporate and retail exposures
2. Market approach: used for corporate exposures
3. Implied historical LGD approach: used for retail exposures, and
4. Implied market approach: used for corporate exposures

The most popular method for defining LGD and the ones we will be focusing on is the workout method. The idea is to ‘work out’ the collection process of a defaulted exposure by considering both incoming/outgoing cash flows as well as both direct/indirect cash flows discounted back to the moment of default using a discount factor (relax, we will come back to this later). Here is an example with number for the ones already lost in theoretical definitions:

We’ll divide the modelling process in two steps to be considered in two separate future articles:

1. Computing the observed LGD. Where we will be dealing with:

• Full business cycle data completion.
• Definition of workout period.
• ?Handling of incomplete workouts.
• The definition of a discount rate.
• Handling of LGDs outside the normal range.

2. Modelling LGD

In a step stage, the bank is interested in modelling the determinants of LGD and provide LGD forecasts. In this section we will therefore deal with several variants of regression-type models that can be applied by a bank for modelling and forecasting LGD.?

Source: Credit Risk Analytics: Measurement Techniques, Applications, and Examples in SAS, 2016