What Loss Given Default Actually Measures
Loss given default is severity, expressed as a percentage of exposure. If a borrower owes $500,000 and the lender eventually collects $297,000 after selling collateral and paying workout costs, the loss is $203,000 and LGD is 40.6%. The complement, 59.4%, is the recovery rate. Every recovery study from Moody's to the FDIC's own charge-off data frames results this same way, which makes LGD comparable across loan books, asset classes, and time periods.
LGD is one of the three legs of modern credit risk measurement: probability of default says whether, exposure at default says how much, and loss given default says how bad. Two loans with identical 5% default probabilities are completely different credits if one recovers 70 cents on the dollar and the other recovers 15. Portfolio loss provisions, regulatory capital, and debt pricing all break without a severity estimate sitting under them.
Because severity is about what happens after default, it rewards lenders who watch collateral quality continuously rather than at origination only. Models that flag rising distress early — the same logic behind the Altman Z-Score calculator — buy time to tighten covenants or demand more security while the borrower can still act, and before the haircut a distressed sale requires gets baked into recovery.
The LGD Formula, Step by Step
Exposure at default comes first, and it is not always the current balance. For a term loan it is the outstanding principal plus accrued interest. For a revolving line it is the drawn amount plus a credit conversion factor applied to the undrawn commitment, since borrowers under stress typically draw everything available before missing payments. Lenders tracking utilization with a credit utilization calculator already watch the single strongest behavioral signal of which revolving lines will be fully drawn at default.
The waterfall then runs in order. Collateral of $400,000 at a 20% haircut nets $320,000. Other recoveries — say a $10,000 guaranty installment — lift gross proceeds to $330,000. Workout costs of 10% remove $33,000, leaving $297,000 net. The loss is $500,000 minus $297,000, or $203,000, and $203,000 divided by $500,000 gives LGD of 40.6%. Each intermediate number matters because it shows which lever — haircut, costs, or additional security — moves severity most for a given deal.
Recovery rate is simply one minus LGD: 59.4% here. When net recovery exceeds exposure, the formula clamps the loss at zero rather than producing a negative LGD, because a lender cannot lose less than nothing. Collateral of $700,000 on this same deal would net $513,000 after haircut and costs, cover the full $500,000, and report LGD of 0% with a 100% capped recovery rate — the clean math of genuine over-collateralization.
Haircuts: Why Collateral Never Sells at Appraised Value
A haircut converts market value into liquidation value. Decades of workout data support rough ranges: income-producing real estate typically realizes 70-85% of appraised value, machinery and equipment 40-60%, inventory sold under time pressure 30-50%, and trade receivables 60-90% depending on debtor concentration and aging. The 20% default haircut in this calculator sits in the middle of the secured real estate range, which is the most common collateral class in commercial lending.
The haircut is the most sensitive input in the model. On the default example, LGD moves from 26.2% at a 0% haircut to 40.6% at 20%, 55.0% at 40%, and 69.4% at 60%. Every additional 10 percentage points of haircut on $400,000 of collateral adds $36,000 of loss, or 7.2 points of LGD, holding everything else constant. That is why two lenders holding identical collateral can report materially different LGDs — their severity assumptions, not their borrowers, diverge.
The practical discipline is honest valuation at origination and revaluation on a fixed schedule. Appraisals ordered by the borrower's own broker skew high, and book values for equipment ignore both technological obsolescence and removal cost. Running a formal valuation before relying on a collateral number — the same rigor a business valuation calculator enforces for enterprise value — protects the LGD estimate from optimism at the input stage.
Workout Costs and the Time Value of Recoveries
Recoveries are not free. Foreclosure filings, collection attorneys, auctioneers, brokerage on asset sales, environmental assessments, property taxes, insurance, and maintenance on repossessed assets all come out of gross proceeds before the lender sees a dollar. On $330,000 of gross proceeds, a 10% cost assumption consumes $33,000 — and a contested commercial workout with bankruptcy litigation routinely consumes 25% or more of everything eventually collected.
Cost assumptions matter almost as much as haircuts. Holding the default deal constant, LGD is 37.3% at 5% costs, 40.6% at 10%, 47.2% at 20%, and 53.8% at 30%. The reason costs climb with time is delay itself: commercial real estate workouts commonly run 12-24 months from default to disposition, and OREO holdings can carry taxes, insurance, and security costs for years. A recovery dollar arriving 24 months late is worth less than one arriving in 90 days.
Treating recovery as a probability-weighted outcome rather than a fixed number is the more honest framing. Some files cure completely, some recover through restructuring, and some liquidate at deep discounts — the same expectation logic that drives the expected return calculator for investments. Averaging realized outcomes across your own defaulted loans, not just modeling the single scenario you hope for, is how a defensible LGD assumption gets built.
Probability of Default and Expected Loss
Expected loss multiplies likelihood by severity: EL = PD x LGD x EAD. On the default inputs, a 5% probability of default combined with 40.6% LGD on $500,000 produces an expected loss of $10,150 per year, 2.03% of EAD. That number is the pricing floor for the credit — the spread and fees charged have to clear it plus the cost of the capital regulators require the lender to hold against the position.
The leverage in the formula is striking. Doubling PD from 5% to 10% doubles expected loss to $20,300; at 2% PD it falls to $4,060, and at 20% PD it reaches $40,600, all with severity unchanged. This is why underwriting effort concentrates on marginal PD improvements for large exposures: a 2-point PD reduction on this deal is worth $4,060 per year, every year the loan is outstanding. Portfolio-level EL simply sums this product across every credit.
Estimating PD has its own toolkit. For corporates, coverage ratios are among the strongest early signals — a borrower whose earnings cannot service interest is default risk regardless of collateral quality, which is why analysts pair severity work with an interest coverage ratio calculator check before finalizing the credit assessment.
LGD by Loan Type and Seniority
Long-run default studies land in consistent bands. Senior secured bank loans recover enough to post LGDs of roughly 30-45%; senior unsecured corporate bonds run 55-75%; subordinated and mezzanine debt recovers little, with LGDs of 75% or more. Seniority dominates: the same enterprise value can produce a 35% LGD for the first lien and a 90% LGD for the second, because the senior position collects first in any liquidation.
Consumer credit sits at the harsh end. Credit card debt is unsecured, and charge-off recoveries after collection and sale of the account commonly imply LGDs of 70-90%. Underwriting on the consumer side leans on payment capacity instead — a debt to income calculator is the retail equivalent of a severity screen, while commercial lenders pair collateral analysis with a DSCR calculator to confirm the property actually covers its debt service before the collateral ever matters.
Collateral quality shifts bands within a class. On a $500,000 exposure secured only by $150,000 of receivables at a 40% haircut plus $10,000 of other recoveries, net recovery is $90,000 and LGD reaches 82% — unsecured in substance despite the paperwork. The calculator makes this visible instantly: a 'secured' loan whose modelled LGD exceeds 70% should be priced, provisioned, and monitored as unsecured.
Using LGD in Loan Pricing and Covenants
Risk-based pricing starts from expected loss. A business loan calculator can structure the payment schedule, but the LGD output decides whether the rate clears the risk: on the default deal, 2.03% annual expected loss means a 3% all-in spread leaves barely 1% for capital and overhead. Deals whose EL exceeds their spread are losing money on day one, whatever the coupon looks like.
Severity estimates also drive structure. If the modelled LGD is 60% against a target of 40%, the levers are concrete: lower the advance rate, demand a guaranty worth real dollars in the other-recoveries field, or shorten the facility so collateral gets refreshed sooner. During workouts, restructured terms and an updated payoff schedule — modelled with a loan payoff calculator — often recover more than a forced liquidation, which is exactly what the workout-cost input is measuring.
Covenants are the monitoring layer that keeps the LGD estimate honest between reviews. Borrowing-base certificates, periodic reappraisals, and coverage covenants catch collateral erosion before default, when the lender still has options. A credit that triggers a covenant at 85% collateral coverage gives the workout team a 40% LGD problem; the same credit ignored until missed payments gives them a 70% one.
Basel III, IRB Models, and Downturn LGD
Under the Basel internal-ratings-based approach, LGD is a regulatory input, not just an internal metric. Foundation IRB banks apply supervisory values — 45% for senior unsecured corporate claims, with lower prescribed LGDs for certain financially collateralized exposures — while advanced IRB banks estimate LGD from their own default history, subject to regulator review. The estimate flows straight into risk-weighted assets, so a 10-point LGD change moves required capital on every loan in the portfolio.
Basel III requires LGD estimates calibrated to downturn conditions, not average times. Recoveries weaken exactly when defaults cluster: fire-sale discounts deepen, buyers disappear, and legal timelines stretch. In the 2008-2009 cycle, commercial real estate recoveries fell 15-20 points versus mid-cycle averages. Stressing this calculator's haircut and cost inputs — 20% haircut becomes 35%, 10% costs becomes 25% — approximates a downturn scenario, and on the default example that stress pushes LGD from 40.6% to 59.5%.
Liquidity is the buffer that separates manageable stress from forced liquidation, for borrowers and lenders alike. A borrower holding adequate cash reserves can survive a bad quarter without triggering collateral sales at the worst possible moment — the resilience an analyst checks with a cash ratio calculator. Lenders who model LGD with honest downturn inputs, monitor collateral on a schedule, and price to the stressed number keep severity a measured risk instead of a surprise.