Skip to content
// WORKED EXAMPLETHE ARITHMETIC A DECISION RESTS ON

Expected Loss: a Worked Example

PD × LGD × EAD — the number a risk margin is built from.

BY MICHAEL SHANG · SIXTEEN YEARS IN COMMERCIAL BANKING · UPDATED

// THE SHORT ANSWER

Expected loss is the average loss a lender expects on a loan: the probability the borrower defaults in a period, multiplied by the share of the exposure lost if it does, multiplied by the exposure at that moment. On a US$10m loan with a 2% chance of default and a 40% loss given default, that is US$80,000 a year — the floor under the risk margin and the figure a provision is held against. This page works the three factors, shows why a well-secured loan to a weaker borrower can cost less than an unsecured loan to a stronger one, and where the number decides a file.

BASIS  Practitioner judgment from sixteen years in commercial banking, set out in How Bankers Think (Highbank Press, 2026). Every borrower on this site is a composite constructed for teaching, not a client. The market is mid-market commercial lending as practised in New Zealand and Australia; where a convention differs elsewhere, the page says so.

// MAKE THE CALL

Issue 08 — Basalt Ridge Developments. A new-to-bank developer, a claimed 28% ROE and two other banks circling. Your bank has a growth target. Your call first, then the senior banker's.

Read the file →

// 01 — THE FORMULA, IN PLAIN TERMS

Expected loss is what a lender expects to lose on a loan on average — not in the bad case, on average, across many loans like it:

EL = PD × LGD × EAD

Probability of default is the chance the borrower stops paying within the period, usually a year, and it comes from the borrower's risk grade. Loss given default is the share of what is owed that the lender does not get back once that has happened, after the security is sold and the costs are paid; it comes from the security and what it would fetch. Exposure at default is what will be owed at that moment: the whole of a term loan, and for a revolving facility the drawn balance plus some share of the undrawn limit, because a borrower in trouble draws down. The glossary entry defines the term the way a credit desk uses it; this page runs the numbers.

// 02 — THE EXAMPLE

A US$10m term loan, fully drawn. The borrower's grade gives a 2% chance of default in the coming year. The loan is secured over plant and receivables the bank expects would fetch 60% of what is owed in a forced sale, after costs — a loss given default of 40%.

EL = 2% × 40% × US$10m = US$80,000 a year
= 0.80% of the exposure

That US$80,000 is not a forecast of this loan's loss; this loan will either lose nothing or lose about US$4m. It is the average across a hundred loans like it, two of which fail in a year and lose US$4m each. It is also the first thing the margin on this loan has to pay for.

Now two borrowers side by side, asking for the same US$10m:

Borrower A · weaker, well secured: PD 3.0% × LGD 25% = 0.75% → US$75,000
Borrower B · stronger, unsecured: PD 1.5% × LGD 65% = 0.98% → US$97,500

The weaker borrower costs the bank less to lend to, because the bank keeps more when it fails. That is the sum behind a pricing decision that looks wrong from outside — a lower margin for the riskier business — and it is why a credit desk reads the security as closely as the borrower.

// 03 — FROM THE NUMBER TO A MARGIN

The margin on a loan is the bank's price for carrying it, and expected loss is its floor. Above the floor sit two more layers: a return on the capital the bank holds against the losses that are larger than expected — the bad year, not the average one — and the cost of running the facility. On the first loan above, as an illustration:

Expected loss 0.80%
Return on capital held 1.00%
Cost of running the facility 0.40%
Margin before any relationship pricing ≈ 2.20% over the cost of funds

A relationship manager who offers 1.5% has not been generous; they have priced the loan below its first default. The argument in the room is never about the expected loss, which is arithmetic, but about the grade and the security that go into it — which is where pricing is actually decided.

WORK IT · THE NUMBER → · RISK-BASED PRICING → · HOW BANKS PRICE BUSINESS LOANS →

// 04 — WHERE IT DECIDES A FILE

A new-to-bank borrower has no grade the bank trusts yet, so its probability of default is the open question in the file — which is why Issue 08's developer, with six projects behind him and none of them with this bank, is offered a smaller first deal rather than a bigger margin: the bank cannot price what it cannot yet read. A lender reading a bank's own accounts meets the same three factors from the other side, in the provision it holds against the loans already going bad.

ISSUE 08 · THE COURTSHIP → · PROVISION COVERAGE → · RISK-WEIGHTED ASSETS →

// STANDARDS THIS SITS BESIDE

The recognised ground this page sits beside. Each was opened and checked when cited; none is a source for the framework itself.

  1. Basel Committee on Banking Supervision, Basel Framework, CRE30 — IRB approach: overview, paragraphs 30.1–30.2 — The risk components are the probability of default, the loss given default and the exposure at default; capital covers unexpected losses and expected losses are treated separately — the three factors in section 01.
  2. Basel Committee on Banking Supervision, Basel Framework, CRE32 — IRB approach: risk components, paragraphs 32.3 and 32.5 — A borrower's probability of default is the one-year figure attached to its internal grade, and a loss-given-default estimate is made for each exposure — where the inputs in section 02 come from.
  3. Basel Committee on Banking Supervision, An Explanatory Note on the Basel II IRB Risk Weight Functions (July 2005) — section 2 — Banks treat expected losses as a cost of doing business, managed through pricing and provisions, while capital is the buffer against losses above that level — the floor and the layer above it in section 03.

// QUESTIONS PEOPLE ASK

Is expected loss the same as a loan loss provision?
They are built from the same three factors and used for different things. Expected loss is the lender's own pricing and risk number, run over the life of the loan or a year of it. The provision in the accounts is what the accounting standard requires to be held against the loan today — in the expected credit loss model, twelve months of expected loss while the loan is performing normally and lifetime expected loss once its risk has risen significantly. The arithmetic is shared; the horizon and the purpose are not.
What is a good probability of default?
There is no number to aim at. A probability of default is read off a risk grade, and the grade is read off the borrower: its cash flow against its debt, its industry, its history, the people running it. A lender does not turn away a 3% borrower; it prices the 3% and asks what it is secured by. The question a credit desk asks is not whether the probability is good but whether the margin and the security are right for it.
Why does security change loss given default and not probability of default?
Because they measure different events. Probability of default is about whether the borrower stops paying, which depends on the business, not on what the bank holds. Loss given default is about what the bank recovers once that has happened, which is exactly what security decides. A lender who takes more security has not made the borrower safer; it has made its own loss smaller if the borrower fails — which is why a secured loan to a weaker borrower can carry a lower expected loss than an unsecured loan to a stronger one.
How does expected loss turn into the margin on a loan?
It is the floor. The margin has to recover the expected loss first, then a return on the capital the bank holds against losses larger than expected, then the cost of running the facility, on top of the cost of the money itself. Expected loss of 0.8% of the exposure means a margin below 0.8% loses money before the first default, on average, whatever the relationship is worth. Everything above that is a decision; the floor is arithmetic.

THE IN-TRAY · ONE MINUTE

Make the call on one file.

A short file on the ground this page covers. Every option opens with how far it sits from a senior banker’s read.

THE NUMBER · THE LOAN · PRICE

A US$10m loan. The bank estimates a 2% chance the borrower defaults within a year, and expects to lose 40% of what is owed if it does.

What is the expected annual loss?

1 FILES · PICK A CALL ON ANY OF THEM

5 MORE ON THE LOAN · PRICE →WHERE DO YOU READ FROM? THE PLACEMENT →