Introduction
Loss given default (LGD) measures the portion of an exposure a lender expects to lose if a borrower defaults, after taking into account all recoveries (collateral proceeds, guarantors, collections, etc.). LGD can be expressed as a dollar amount or as a percentage of exposure at default (EAD). It is a core input in credit-risk models and regulatory frameworks (e.g., Basel) and, together with probability of default (PD) and exposure at default (EAD), determines expected loss (EL).
Key takeaways
– LGD = expected loss on an exposure after recoveries, expressed as $ or % of EAD.
– Expected loss = PD × LGD × EAD.
– Two common formulations:
• LGD($) = EAD × (1 − Recovery Rate)
• LGD(%) = 1 − (Net Recovery Proceeds / Outstanding Debt)
– LGD varies by product, collateral, seniority, jurisdiction, and economic cycle; regulators typically require “downturn” LGDs for capital calculations.
– Usage Given Default is another name for Exposure at Default (EAD).
How LGD fits in the risk framework
– PD (Probability of Default): likelihood the borrower will default in a defined horizon.
– EAD (Exposure at Default): amount outstanding or likely to be outstanding at the moment of default.
– LGD: fraction of that exposure that will not be recovered.
– Expected Loss (EL) = PD × LGD × EAD — used for provisioning, pricing and capital calculation.
Two simple LGD formulas
– Dollar form: LGD($) = EAD × (1 − Recovery Rate)
• Recovery Rate = (net recoveries) / EAD
– Percentage form (collateral-focused): LGD(%) = 1 − (Potential Sale Proceeds / Outstanding Debt)
• Potential sale proceeds should be net of disposition costs and realistic haircuts.
Practical, step-by-step approach to calculating LGD
1. Define the scope and segmentation
• Split portfolios by product type, collateral type, seniority (senior unsecured, subordinated), jurisdiction, vintage, and borrower sector. LGD varies across these segments.
2. Determine EAD methodology
• For term loans use the outstanding balance at default.
• For revolvers/commitments estimate likely drawdown at default (credit conversion factor approaches).
• Document and consistently apply the definition used.
3. Gather recovery data
• Collect historical recovery data: gross recoveries (collateral sales, guarantor payments), recoveries from collections, fees/penalties.
• Track timing of recoveries (months to recovery) and net recoveries (after legal, repossession and disposition costs).
4. Net recoveries and costs
• Subtract recovery-related costs: legal fees, enforcement costs, auction costs, repair/storage, taxes, and any other direct costs to obtain proceeds.
• Consider tax effects, if material.
5. Calculate the recovery rate
• Recovery rate = (present value of net recoveries) / EAD
• Decide on discounting: recoveries received far in the future should be discounted to default date. If using undiscounted recoveries, document the approach and be consistent.
6. Compute LGD
• LGD($) = EAD × (1 − Recovery Rate)
• LGD(%) = 1 − (Net Recovery Proceeds / Outstanding Debt)
7. Adjust for cure rates and write-offs
• Distinguish between cured accounts (those that return to performing status) and final default losses.
• For capital or worst-case analyses, use post-cure outcomes appropriately (e.g., exclude cures if modeling realized losses vs. final outcomes).
8. Apply downturn adjustments (regulatory and stress)
• For regulatory capital (Basel), use downturn LGD that reflects stressed macroeconomic conditions.
• Upward adjustments or add-on factors can be applied to point-in-time LGDs to produce through-the-cycle or downturn LGDs.
9. Validate and backtest
• Backtest model LGDs versus realized losses by vintage and scenario. Use statistical tests and expert judgment to refine segmentation and assumptions.
10. Governance and documentation
• Maintain a documented methodology, data lineage, assumptions (discount rates, haircuts), and periodic re-calibration processes. Ensure auditability and governance.
Worked example (condominium mortgage)
– Outstanding balance at default (EAD): $300,000
– Expected foreclosure sale price (gross proceeds): $200,000
– Disposition costs and legal fees: $10,000 → net recovery = $190,000
– Recovery rate = 190,000 / 300,000 = 63.33%
– LGD(%) = 1 − 0.6333 = 36.67%
– LGD($) = 300,000 × 0.3667 = $110,000
Notes: If you use the simpler formula without netting disposition costs, the LGD will be understated. If recoveries occur over many months, discounting to the default date will increase LGD.
LGD vs EAD — quick distinction
– EAD is the total amount exposed when default occurs (conservative, gross of recoveries).
– LGD measures what portion of that exposure is ultimately lost after recoveries. EAD is typically larger or equal to the eventual loss before recoveries.
Can LGD be zero?
– Theoretically yes: if the lender expects full recovery of the exposure (100% recovery rate), LGD = 0. In practice a zero LGD is rare because recoveries are uncertain and costs/legal delays usually leave some loss.
What is Usage Given Default?
– Usage Given Default (UGD) is another term for Exposure at Default (EAD) — the expected gross exposure when default happens, including drawn balances and likely draws on undrawn commitments.
Regulatory context — Basel model and capital
– LGD is a primary input in the Basel framework (e.g., Basel II/III) used to calculate expected loss and regulatory capital. Supervisors often require conservatism (e.g., downturn LGDs) and may prescribe minimum LGD values or floors for certain exposure types.
Common practical tips and pitfalls
– Segment exposures finely: LGD differs substantially between secured vs unsecured, senior vs subordinated, and across asset classes.
– Use vintage analysis: track recoveries by default vintage to capture lifecycle effects.
– Net recoveries, not gross: always subtract direct recovery costs.
– Include recovery timing: long recovery horizons materially increase loss when discounted.
– Apply conservative downturn adjustments for capital planning and stress testing.
– Watch for data gaps: insufficient historical recoveries can bias estimates; use proxy data or expert overlays but document them.
– Keep LGD governance robust: periodic validation, model risk controls, and regulatory alignment.
When to use which LGD
– Accounting/provisioning: use point-in-time expected loss measures consistent with accounting standards (e.g., IFRS 9 / CECL).
– Pricing/underwriting: use expected LGD reflecting current conditions and collateral value.
– Capital and stress testing: use downturn or stressed LGD values.
References and further reading
– Investopedia: Loss Given Default (LGD). (Source material summarized and paraphrased).
– Bank for International Settlements: “Basel II: Revised international capital framework.”
– Consumer Financial Protection Bureau: Differentiating between secured and unsecured credit (context on secured debt). /
Editor’s note: The following topics are reserved for upcoming updates and will be expanded with detailed examples and datasets.