Expected Loss

Expected loss (EL) is the average credit loss predicted for an exposure or portfolio over a defined horizon. In a common credit-risk model, it combines the probability that default occurs, the amount expected to be exposed at default, and the percentage of that exposure expected to be lost after recoveries.

Expected loss is a model average, not the amount that a single loan is guaranteed to lose. It is also not automatically the same as an accounting allowance under U.S. CECL or IFRS 9.

Key Takeaways

• The simplified formula is PD x EAD x LGD, with PD and LGD expressed as decimals and EAD in currency.
• PD, EAD, and LGD must use compatible default definitions, time horizons, exposure bases, and economic assumptions.
• LGD should reflect recovery timing and costs when the model measures economic loss.
• Portfolio EL is generally the sum of exposure-level or segment-level expected losses, not the EL percentage applied blindly to every balance.
• EL can inform pricing, underwriting, risk appetite, stress testing, provisioning, and regulatory capital analysis, but each use can require different parameter definitions.
• Expected loss covers the average modeled loss; unexpected loss addresses adverse variation around that average.

Expected Loss Formula

For one exposure under a simplified model:

The calculation is only as coherent as its components. A one-year PD should not be combined casually with an LGD from a different default definition or an EAD measured on a different exposure basis.

Worked Example: One Loan

Assume a lender estimates for a one-year horizon:

• PD: 2%
• EAD: $1,000,000
• LGD: 45%

The expected loss is:

The $9,000 is the exposure’s average modeled loss contribution. The actual one-year outcome may be no default and no credit loss, or default with a loss materially above or below $9,000.

The 45% LGD can correspond to a 55% net economic recovery rate when both use the same EAD, recovery cash flows are discounted to the default date, and material workout costs are deducted.

Portfolio Example

Suppose a portfolio contains three segments:

The portfolio EL rate is about 0.86% of EAD, but the segment results show why the average should not be applied indiscriminately. Risk differs by default likelihood, utilization, collateral, seniority, and recovery process.

Economic EL vs. Accounting Expected Credit Loss

An institution can use PD/LGD methods within an accounting process, but that does not make a one-year regulatory or pricing EL equal to the financial-statement allowance. Contractual term, prepayments, discounting, staging, forecast periods, scenario weights, and scope can differ.

Expected Loss vs. Unexpected Loss

Expected loss is the center of the modeled loss distribution. Unexpected loss concerns the possibility that actual losses exceed that average. Pricing and allowances may address expected loss, while capital, limits, stress tests, and diversification analysis address severe but plausible deviations.

A loan spread equal to EL is not necessarily adequate compensation. Funding cost, operating expense, liquidity, capital usage, taxes, concentration, model uncertainty, and required return also matter.

Where Expected Loss Is Used

• Underwriting: comparing risk across borrowers, facilities, and structures.
• Pricing: estimating the average credit-cost component of a rate or spread.
• Limits: aggregating exposure by grade, industry, geography, or product.
• Portfolio management: identifying concentration and migration effects.
• Stress testing: replacing baseline parameters with adverse assumptions.
• Accounting: informing an expected-credit-loss method when consistent with the reporting framework.
• Regulatory capital: applying prescribed or approved IRB parameter rules where permitted.

How to Evaluate an EL Estimate

1. Confirm the default definition and measurement horizon.
2. Identify whether PD is point-in-time, through-the-cycle, or otherwise calibrated.
3. Reconcile EAD with current balance, undrawn commitments, amortization, and expected utilization.
4. Check whether LGD uses discounted net recoveries, includes collection costs, and reflects seniority.
5. Review collateral values, haircuts, lien perfection, guarantees, and time to recovery.
6. Test segmentation, data history, overrides, and low-default portfolio treatment.
7. Compare predicted losses with later defaults and recoveries through back-testing.
8. Stress parameter correlation rather than assuming PD, EAD, and LGD move independently.

Common Mistakes

• Treating EL as the most likely loss on one specific loan.
• Mixing a one-year PD with lifetime EAD or LGD assumptions without adjustment.
• Entering percentages as whole numbers, such as 2 instead of 0.02.
• Using current drawn balance as EAD for a revolving line without considering future draws.
• Setting LGD equal to one minus a nominal gross recovery rate that ignores timing and costs.
• Calling economic EL the CECL or IFRS 9 allowance without reconciling framework rules.
• Adding average EL estimates without considering concentration and correlated stress.
• Assuming a credit spread equal to EL fully compensates for credit risk.

Risks and Limitations

EL is sensitive to sparse default data, changing underwriting, economic cycles, model selection, recovery lags, collateral valuation, and parameter correlation. Average estimates can conceal tail risk and concentration. Model outputs can also create false precision when default observations are limited or definitions are inconsistent.

This page is educational and is not accounting, regulatory, lending, investment, model-validation, or personalized financial advice.

Authoritative Sources

• Basel Framework treatment of IRB expected losses and provisions
• Basel Framework risk components for PD, LGD, and EAD
• Federal Reserve interagency policy statement on allowances for credit losses
• IFRS Foundation overview of IFRS 9 expected credit losses

Related Terms

• Probability of Default: Likelihood component in a simplified expected-loss model.
• Loss Given Default: Conditional loss-severity component.
• Recovery Rate: Recovery-side measure that can complement LGD when definitions match.
• Allowance for Credit Losses: Financial-statement estimate that is not automatically equal to economic EL.
• Credit Risk: Broader risk of loss from failure to meet financial obligations.

FAQs