TheronGroup of Consultants

80/20 Golden Rule · Pareto Optimality

Worked example

RSK4804: probability of default from expected loss, and over five years

UNISA's May/June 2024 RSK4804 Question 5 worked in full: what PD does in a credit decision, PD from expected loss, and the cumulative PD of a five-year loan.

UNISA · RSK4804 · May/June 2024 · Question 5

Probability of default (PD) is the number that turns a credit opinion into a decision, and RSK4804 examines it from both directions: what it is for, and how to get it out of the figures you are given. Question 5 of UNISA's May/June 2024 paper, worth 20 marks, does both.

The question is described in our own words below; the answers are TheronGroup’s.


01The question

5.1 (5 marks). Explain, in your own words, how the probability of default helps in credit decisions.

5.2 (5 marks). A bank has an R800 million loan to a civil engineering firm, maturing in one and a half years. The expected loss on the loan is R11,2 million and the recovery rate is assumed to be 23%. Calculate the probability of default.

5.3 (10 marks). You are head of credit at a savings bank, considering a five-year business term loan repaid in a single balloon at the end. Its marginal probabilities of default are 1,2%, 1,4%, 1,5%, 1,7% and 2% for years one to five. How would you find the cumulative probability of default over the five years for pricing purposes?

5.1How PD helps a credit decision

The probability of default is the statistical probability that a borrower will default, and it is the quantity that turns a credit opinion into a number that can be used. It helps a credit decision in five ways.

  1. It quantifies the credit risk. A probability of default lets the risk of one asset class or one obligor be distinguished from another and compared on the same scale, instead of being described in words.
  2. It selects the client. The probability of default is derived from internal and external default data and is tied to the credit grade, so the decision to take the borrower on, and the grade at which to take them on, follows from the estimate.
  3. It prices the credit. The higher the probability of default, the higher the risk premium that must be charged, which is why corporate assets yield more than government securities and why the probability of default is the major driver in pricing a credit default swap or a credit spread option.
  4. It allocates capital. The probability of default feeds the expected loss and the risk-weighted asset calculation under the internal-ratings-based approach, so a higher probability of default means more capital allocated to the exposure and a lower return on that capital unless the pricing carries it.
  5. It monitors credit quality after the decision. Because the estimate is refreshed as the obligor’s position changes, a rising probability of default is the early warning that identifies the high-risk customer for closer monitoring, tighter limits or exit; and for a multi-year exposure the probability of default over the whole term, not the first year alone, is what the decision has to be taken on.

5.2PD from expected loss

Expected loss is the product of the probability of default, the loss given default and the exposure at default. Every parameter but one is given, so the equation is solved for the one that is not.

  • Exposure at default (EAD) = R800 million, the loan outstanding.
  • Recovery rate = 23%, so loss given default (LGD) = 1 − 0,23 = 77%.
  • Expected loss (EL) = R11,2 million.
EL           = PD × LGD × EAD
R11,2 million = PD × 0,77 × R800 million
R11,2 million = PD × R616 million
PD           = R11,2 million ÷ R616 million
             = 0,018182
PD           = 1,82%

Proof: 0,018182 × 0,77 × R800 million = R11,2 million

The probability of default is 1,82%.

The one-and-a-half years. Maturity does not enter the expected loss formula as a multiplier, so 1,82% is the probability of default over the life of the loan. If the question wanted an annual rate, 1,82% spread over the 1,5 years is 1,21% a year. Say which figure you are giving; a candidate who states the basis keeps the marks either way.

5.3The cumulative PD over five years

Marginal probabilities of default are conditional annual rates: each is the probability of defaulting in that year, given survival to the start of it. They cannot simply be added. What compounds is survival, so the cumulative probability of default over the term is one minus the product of the annual survival probabilities.

Cumulative PD = 1 − [(1 − PD1) × (1 − PD2) × (1 − PD3) × (1 − PD4) × (1 − PD5)]
Year Marginal PD Survival probability Cumulative survival Cumulative PD
1 1,2% 1 − 0,012 = 0,988 0,988000 1,20%
2 1,4% 1 − 0,014 = 0,986 0,974168 2,58%
3 1,5% 1 − 0,015 = 0,985 0,959555 4,04%
4 1,7% 1 − 0,017 = 0,983 0,943243 5,68%
5 2,0% 1 − 0,020 = 0,980 0,924378 7,56%
= 1 − [0,988 × 0,986 × 0,985 × 0,983 × 0,980]
= 1 − 0,924378
= 0,075622
Cumulative probability of default over the five years = 7,56%

What it means for the pricing

  1. Simply adding the five marginal rates gives 7,80%, which overstates the risk by 0,24 of a percentage point. The difference is the compounding of survival: a borrower can only default in year five if they have not defaulted already.
  2. The repayment structure is what makes the cumulative figure the right one. A balloon repayment leaves the whole principal outstanding to the end of year five, so the bank is exposed to the full amount through every one of the five years and must price against 7,56%, not against the first-year 1,2%.
  3. Spread evenly, 7,56% over five years is about 1,5% a year (1,56% as a compound annual rate), and that is roughly what the risk premium in the margin has to recover.
  4. The expected loss over the term follows once the loss given default is applied: cumulative PD × LGD × EAD, which is what the spread over the funding cost has to cover before the loan earns anything.
  5. The marginal rates rise year on year, from 1,2% to 2,0%, so the credit is assumed to deteriorate as the term runs. That is the ordinary pattern for a long exposure, and it justifies a term premium on top of the credit premium.
  6. The estimate is only as good as the marginal rates behind it. They come from the rating grade’s historical default experience and should be through-the-cycle rather than point-in-time for a five-year decision.
  7. The cumulative figure also sets the provisioning and capital consequence of the approval, since expected loss and the risk-weighted asset calculation both take the probability of default as an input.
  8. A sensitivity is worth running before approval: if each marginal rate is stressed upwards, the cumulative probability rises faster than the individual rates do, and the pricing must still stand at the stressed level.
  9. Security changes the loss but not the probability. The probability of default is priced through the margin; the loss given default is what collateral reduces, and the two are set separately.
  10. The decision closes on the price: approve the five-year facility at a margin that recovers a 7,56% cumulative default probability at the loss given default the security supports, and decline it at any margin that does not.

Studying RSK4804

RSK4804 is closed book, so the expected-loss formula and the survival logic behind cumulative PD have to be carried in memory and practised until they are automatic. The RSK4804 course has every held sitting worked the same way, from R200 a month.

Get in touch

Tell us your module

Send the module code and your sitting. We reply with exactly what exists for it today, a free extract from that module, and what it costs.

Students

Name your module code and sitting. You get a free extract from your own module — not a generic brochure — before any payment.

Send an enquiry

Free extract

A full worked example and a note extract from the module you are sitting, sent the same week you ask.

Request the extract

Employers and professional bodies

In-house governance, risk and internal-audit training, and bulk licensing for staff in part-time postgraduate study.

Corporate enquiry