2. Measurement of Probability of Default from Rating System

2. Measurement of Probability of Default from Rating System

1. Rating Assignment Methodologies

1.1 Definition

Cumulative Default Probability: the probability that a borrower will default over a specified multi-year period.

$$P{D}_t^{\text{cumulated}}=\frac{{\text{Def}}_i}{{\text{Names}}_t}$$

• $\text{Names}$: the number of issuers
• $\text{Def}$: the number of names that have defaulted in the time horizon

Marginal Default Probability: the probability that a borrower will default in any given year.

$$P{D}_k^{\text{marg}}=P{D}_{t+k}^{\text{cumulated}}-P{D}_t^{\text{cumulated}}$$

Forward Probability: is contingent to the survival rate.

$$P{D}_{t,t+k}^{\text{Forw}}=\frac{{\text{Def}}_{t+k}-{\text{Def}}_t}{{\text{Names survived}}_t}$$

Survival Rate: the probability a borrower will not default over a specified multi-year period.

$$S{R}_{t,t+k}^{\text{Forw}}=1-P{D}_{t,t+k}^{\text{Forw}};1-P{D}_t^{\text{cumulated}}=\prod _{i=1}^{t}S{R}_i^{\text{Forw}}$$

Annualized Default Rate(ADR)
If it is necessary to price a credit exposed transaction on a five year time horizon, it is useful to reduce the five-year cumulated default rate to an annual basis for the purposes of calculation.

$$1-P{D}_t^{\text{cumulated}}=\prod _{i=1}^{t}S{R}_i^{\text{Forw}}=(1-AD{R}_t{)}^t=e^{-ADR\times t}$$

1.2 Example

Suppose there are one thousand issuers rated BBB at the beginning of first year, three issuers default in first year, another six issuers default in second year, and another ten issuers default in third year.

Cumulative Default Probability
Cumulative default probability at end of one year: $P{D}_1^{\text{cumulated}}=\frac{{\text{DEF}}_1}{{\text{Names}}_{t=0}}=\frac{3}{1000}=0.3\%$

Cumulative default probability at end of two year: $P{D}_2^{\text{cumulated}}=\frac{{\text{DEF}}_1+{\text{DEF}}_2}{{\text{Names}}_{t=0}}=\frac{3+6}{1000}=0.9\%$

Cumulative default probability at end of three year: $P{D}_3^{\text{cumulated}}=\frac{{\text{DEF}}_1+{\text{DEF}}_2+{\text{DEF}}_3}{{\text{Names}}_{t=0}}=\frac{3+6+10}{1000}=1.9\%$

Marginal Default Rate

Marginal default rate in the first year: $d_1=\frac{3}{1000}=0.3\%$

Marginal default rate in the second year: $d_2=\frac{6}{1000}=0.6\%$

Marginal default rate in the third year: $d_3=\frac{10}{1000}=1\%$

Forward Probability
Forward probability at end of one year: $P{D}_1^{\text{Forw}}=\frac{3}{1000}=0.3\%$

Forward probability at end of two year: $P{D}_2^{\text{Forw}}=\frac{6}{997}=0.6018\%$

Forward probability at end of three year: $P{D}_3^{\text{Forw}}=\frac{10}{991}=1.009\%$

Survival Rate
Cumulated survival rate at the end of one year: $S{R}_1^{\text{cumulated}}=1-P{D}_1^{\text{cumulated}}=1-0.3\%=99.7\%$

Cumulated survival rate at the end of two year: $S{R}_2^{\text{cumulated}}=1-P{D}_2^{\text{cumulated}}=1-0.9\%=99.1\%$

Cumulated survival rate at the end of three year: $S{R}_3^{\text{cumulated}}=1-P{D}_3^{\text{cumulated}}=1-1.9\%=98.1\%$

Forward survival rate at the end of one year: $S{R}_1^{\text{Forw}}=1-P{D}_1^{\text{Forw}}=1-0.3\%=99.7\%$

Forward survivial rate at the end of two year: $S{R}_2^{\text{Forw}}=1-P{D}_2^{\text{Forw}}=1-0.6018\%=99.3982\%$

Forward survivial rate at the end of three year: $S{R}_3^{\text{Forw}}=1-P{D}_3^{\text{Forw}}=1-1.009\%=98.991\%$

Annualized Default Rate (ADR)
Discrete annualized default rate at the end of one year: $AD{R}_1=1-(1-P{D}_1^{\text{cumulated}})=1-(1-0.3\%)=0.3\%$

Discrete annualized default rate at the end of two year: $AD{R}_2=1-\sqrt{1-P{D}_2^{\text{cumulated}}}=1-\sqrt{1-0.9\%}=0.45\%$

Discrete annualized default rate at the end of three year:$AD{R}_3=1-\sqrt[3]{1-P{D}_3^{\text{cumulated}}}=1-\sqrt[3]{1-1.9\%}=0.64\%$

Continuous annualized default rate at the end of one year: $AD{R}_1=-\ln (1-P{D}_1^{\text{cumulated}})=0.3005\%$

Continuous annualized default rate at the end of two year: $AD{R}_2=-\frac{\ln (1-P{D}_2^{\text{cumulated}})}{2}=0.452\%$

Continuous annualized default rate at the end of three year: $AD{R}_3=-\frac{\ln (1-P{D}_3^{\text{cumulated}})}{3}=0.6394\%$

2. Credit Ratings

2.1 Key Features of A Good Rating System

Measurability and verifiability: give correct expectations in terms of default probabilities, adequately and continuously back tested.

Objectivity and homogeneity: generate judgments only based on credit risk considerations, ratings are comparable among portfolios.

Specificity: measure distance from the default event without any regards to other corporate financial features.

2.2 Agencies’ Ratings

Issue and Issuer Ratings
Moody’s releases mainly issues ratings and far less issuers’ ratings.

S&P concentrates on providing a credit quality valuation referred to the issuer.

FITCH adopts an intermediate solution, offering an issuer rating, limited to potential insolvency on publicly listed bonds.

Ratings released by the three international rating agencies are not directly comparable.

Internal Rating and Agencies’ Ratings
Analytical solutions, weights, variables, components, and class granularities are different from one bank to another.

Banks tend to harmonize their valuation tools, favoring a substantial convergence of methods and results.

No proven inferiority or superiority of expert-based approaches versus formal ones.

Methods are good enough in a given period, it is not certain that the same performance will be reached in the future.

Borrower Rating and Probability of Default

Because of law of large numbers, actual frequencies are a good prediction of central probabilities, in the long run.

Rating	Year 1	Year 2	Year 3	Year 4	Year 5	Year 6	Year 7	Year 8	Year 9	Year 10
Aaa
Aa
A
Baa
Ba
B
Caa-C
Inv.
Spec.
All

Investment-grade credits: increase of cumulative PD is more than proportional with the horizon (mean-reversion effect).

Speculative-grade credits: increase of cumulative PD is less than proportional with the horizon (survival effect).

2.3 Migration Matrix