Jan 12

The 2007 FRM. How to get a jump start: Part 3 (Credit Module)

by David Harper, CFA, FRM, CIPM

FRM | Risk |

BlueCreditCard

In this post, I will point you toward a jump start on the Credit Risk module of the Financial Risk Manager (FRM) exam. At Bionic Turtle, we start with the following building blocks:

CreditBuildingBlocks

From left to right, the credit risk building blocks are: external and internal credit ratings; probability of default (PD) and loss given default (LGD); credit risk portfolio models; and credit derivatives.

In regard to credit ratings, the assigned text last year was De Servigny?s Measuring and Managing Credit Risk. This text is likely to repeat this year. Chapters 2 through 4 introduce ratings, credit scoring and loss given default (LGD). Notice the philosophical difference between Standard & Poors (tends to reflect a view on the probability of default) and Moody's (i.e., tends to reflect a view on the expected loss; i.e., probability of default multiplied by LGD).

Among the readings, six different credit risk portfolio models are reviewed:

 

CreditRiskPortfolioModels2

You don't need to become an expert in each model. Instead, notice that the models are sliced along key feature-dimensions or "classes": underlying variables (e.g., spreads, macroeconomic variables); analytical versus simulation-based; distributional assumptions (lognormal, binomial, etc); correlation assumptions; and model type. The texts encourage you to study the models along these key dimensions.

For example, in regard to model type, the difference is between reduced-form and structural models. This corresponds to a classic econometric distinction between exogenous and endogenous variables. An exogenous variable is external to the model or system; it represents an unexplained input. An endogenous variable is explained by the model; it represents modeled output.

ExogenousModel2

Under this framework, reduced-form models are exogenous in the sense that they take debt prices or credit spreads (as Meissner says, they "abstract from the company's specific data") as input into a model that calculated probability default (PD). On the other hand, a structural model simulates the assets and liability of the company (going forward in time), such that they "endogenize" the bankruptcy process.

Comments

  1. Be the first to leave a comment!

Leave a Comment