Key Points
- LGD represents the percentage of exposure the lender will not recover after a borrower defaults, net of collateral and collection proceeds.
- Most entities estimate LGD between 20% and 60% for unsecured commercial exposures, depending on jurisdiction and seniority.
- An overstated recovery assumption in LGD directly understates the expected credit loss provision on the balance sheet.
- The auditor must test whether the entity's LGD reflects current conditions, not just historical averages from benign economic periods.
What is Loss Given Default (LGD)?
On about half the IFRS 9 files we've reviewed, the LGD input hasn't changed in two or three years. The team just rolls it forward from last year's model, maybe with a token tweak to the forward-looking overlay so the WP doesn't look stale. That's a problem, because LGD is doing real work inside the expected credit loss (ECL) calculation and a wrong recovery assumption flows straight to the provision line.
IFRS 9.5 .5.17 requires an entity to measure ECL as an unbiased, probability-weighted amount determined by evaluating a range of possible outcomes. For most entities that model ECL on a component basis, LGD sits alongside probability of default (PD) and exposure at default (EAD). The formula is ECL = PD x LGD x EAD, applied over either a 12-month or lifetime horizon depending on staging.
LGD is not a single number plucked from a table. IFRS 9 .B5.5.52 requires forward-looking information to be incorporated, which means the entity must adjust historical loss-given-default rates for current conditions and reasonable forecasts. A bank that recovered 70 cents on the euro from defaulted commercial loans between 2015 and 2019 can't assume 70% recovery still holds if the collateral pool has shifted from real estate to unsecured receivables. The auditor's job under ISA 540.13 (a) is to evaluate whether the entity's method for building the LGD estimate is appropriate given the nature of the financial instruments in the portfolio.
Worked example: Henriksen Shipping A/S
Client: Danish maritime logistics company, FY2025, revenue EUR 140M, IFRS reporter, portfolio of trade receivables and loan receivables from charterer advances.
Step 1 — Segment the portfolio by recovery characteristics
The finance team at Henriksen splits its credit exposures into two pools. Pool A consists of trade receivables from investment-grade shipping counterparties (EUR 22M gross). Pool B consists of charterer advance loans secured by vessel liens (EUR 9M gross).
Step 2 — Estimate LGD per pool
For Pool A, Henriksen uses five years of internal loss data showing a 58% average loss rate on defaulted unsecured receivables. The entity adjusts this to 62% to reflect weakening freight rates in the forward-looking overlay. For Pool B, historical recoveries from vessel liens produced a 15% loss rate. The entity adjusts to 20% because second-hand vessel values have declined 12% year on year.
Step 3 — Calculate the ECL contribution from LGD
Assuming Pool A has a 12-month PD of 3.5% and EAD of EUR 22M, the ECL contribution is: 3.5% x 62% x EUR 22M = EUR 477,400. For Pool B (PD 2.0%, EAD EUR 9M): 2.0% x 20% x EUR 9M = EUR 36,000. Total ECL from both pools: EUR 513,400.
Step 4 — Auditor challenge on the forward-looking adjustment
The engagement team tests whether the 4-percentage-point uplift on Pool A is sufficient. Their own sensitivity analysis, using a stressed freight scenario, produces an LGD of 68%. The EUR 1.3M difference in ECL at the stressed rate exceeds performance materiality (PM) of EUR 420,000. That's genuinely uncomfortable. The estimate is technically supportable at the entity's rate, but the auditor's independent point estimate sits well outside PM.
Conclusion: Henriksen's LGD estimates are supportable at EUR 513,400 total ECL, with the forward-looking adjustment documented by reference to observable market data and the team's sensitivity analysis recorded to demonstrate the estimate falls within a reasonable range.
Why it matters in practice
- Teams frequently accept management's historical average loss rate as the LGD without testing the forward-looking overlay. We've seen this become a "just roll it forward" exercise where last year's recovery rate gets copied into this year's model with no real challenge. IFRS 9 .B5.5.52 requires reasonable and supportable forward-looking information to be incorporated. An LGD built purely on backward-looking data fails that requirement, and ISA 540.13 (b) requires the auditor to evaluate whether the data on which the estimate is based is relevant and reliable.
- Collateral valuations embedded in LGD estimates are often stale. If the entity last appraised the collateral pool 18 months ago, the LGD reflects outdated recovery assumptions. ISA 540.15 requires the auditor to design procedures that respond to the assessed risk of material misstatement, which includes testing whether collateral values are current. At its worst this becomes SALY with a methodology shield: the model documentation looks thorough, the inputs haven't moved, and nobody asks why the 2023 recovery rate still appears in a 2025 file.
Related terms
Related tools
Related reading
Frequently asked questions
How do I audit loss given default under IFRS 9?
Test the components of the LGD estimate separately: historical loss data, forward-looking adjustments, collateral recovery assumptions, and any model overlays. ISA 540.13(a) requires you to evaluate whether the entity's method for estimating LGD is appropriate. Run an independent sensitivity analysis on the forward-looking overlay and compare management's recovery rates to external benchmarks or peer data where available.
Does LGD change when a loan moves from Stage 1 to Stage 2?
Not necessarily. LGD measures what you lose given default, and that percentage depends on collateral and seniority, not on the staging trigger. What changes between stages is the horizon: Stage 1 uses 12-month PD while Stage 2 uses lifetime PD. IFRS 9.5.5.3 requires lifetime ECL once a significant increase in credit risk occurs, but the LGD component may remain unchanged if recovery expectations have not shifted.
When should I challenge management's LGD assumptions?
Challenge whenever the entity's LGD relies on a single data source, uses historical averages without adjustment for current conditions, or ignores collateral deterioration. IFRS 9.B5.5.49 requires that estimates reflect conditions at the reporting date. If market data (property indices, secondary market prices) contradicts the entity's assumptions, ISA 540.18 gives you grounds to request revision or record a misstatement.