Loss Given Default, or LGD, is the share of a loan or credit exposure a lender actually loses when a borrower defaults, expressed as a percentage of the amount at risk. If a bank lends $1 million and recovers $600,000 after default, the LGD is 40%. The core formula is straightforward: LGD equals one minus the recovery rate. Banks use the figure to price loans, set loss provisions, and calculate regulatory capital under the Basel framework.
What the Formula Actually Captures
The recovery rate looks simple on paper. It’s everything the lender gets back, through collateral liquidation, restructured payments, or the sale of the defaulted debt, divided by what was owed. In practice, the numerator has to be built carefully, because “what came back” isn’t the same as “what the lender ended up with.”
Three categories of cost eat into the recovery.
Direct recovery costs are the out-of-pocket expenses of the workout: legal fees for foreclosure or litigation, court filings, appraisals, and the allocated salaries of internal workout teams. A property that sells for $400,000 but costs $30,000 in legal and liquidation fees delivers $370,000 in net recovery, not $400,000.
Indirect costs are harder to see but still real. Capital tied up in a non-performing loan can’t be deployed elsewhere, and management attention gets diverted to the workout. These costs resist easy quantification, but they belong in a true economic loss figure.
Time value of money matters because recoveries rarely arrive in one lump sum. A workout can stretch several years, and $500,000 collected three years after default isn’t economically worth $500,000 at the default date. Workout LGD calculations discount future recoveries back to the default date. The chosen discount rate meaningfully affects the result. Three approaches are common: the original contractual loan rate (the yield the lender expected to earn), the risk-free rate (which produces a lower LGD but arguably understates the true cost), and a risk-adjusted rate estimated from market data. No single method is universally required.
The Denominator: Exposure at Default
The bottom of the LGD ratio is Exposure at Default (EAD), the amount at risk when default occurs. Getting EAD wrong distorts everything downstream.
Term Loans
For a fully drawn term loan, EAD is the outstanding principal plus accrued but unpaid interest. If you lent $2 million and the borrower has paid down $400,000, the EAD at default is roughly $1.6 million plus accrued interest.
Revolving Facilities
Revolving credit lines are harder because a distressed borrower can, and often does, draw more of the available commitment right before default. EAD for a revolver depends on a Credit Conversion Factor (CCF), which estimates how much of the undrawn commitment gets tapped before default.
Under the Basel standardized approach, prescribed CCFs range from 10% for commitments the bank can cancel at any time without notice, up to 100% for firm commitments like standby letters of credit and forward purchases. General commitments that don’t qualify for a lower factor carry a 40% CCF; trade-related contingencies and note issuance facilities receive 50%.1Bank for International Settlements. CRE20 – Standardised Approach: Individual Exposures
The formula is: amount already drawn, plus undrawn commitment multiplied by the CCF. A borrower with a $100,000 line who has drawn $30,000, where the remaining $70,000 carries a 40% CCF, has an EAD of $58,000.
Derivatives
Derivative exposures don’t have a loan balance in the ordinary sense. Their value moves with the market, so EAD must reflect both current replacement cost and the possibility the position gets worse before the counterparty defaults. Under the Standardized Approach for Counterparty Credit Risk (SA-CCR), EAD equals 1.4 times the sum of replacement cost and a potential future exposure add-on.2Bank for International Settlements. CRE52 – Standardised Approach to Counterparty Credit Risk
How Banks Estimate LGD
No lender waits for a single default to happen and calls the resulting number its LGD. Estimation is systematic, and the method depends on data availability, exposure type, and whether the number is feeding regulatory capital or internal risk management.
Workout LGD
The most granular approach. The bank tracks every cash flow recovered from its own defaulted loans, subtracts direct and indirect costs, discounts the net recoveries to the default date, and expresses the loss as a fraction of EAD. It uses only internal data, so it’s specific to that institution’s portfolio and workout processes.
The catch is time. Workouts commonly stretch five to seven years, which means a significant share of the default portfolio is still in progress on any given estimation date. Banks use statistical techniques to project final recoveries for incomplete cases. The finished workouts you can measure today also reflect economic conditions from years ago, which may not describe the current environment.
Market LGD
When defaulted debt trades on secondary markets, the price offers a real-time consensus estimate of recovery. A defaulted bond trading at $35 per $100 of face value implies a recovery near 35% and an LGD around 60% once accrued interest and trading costs are considered. The approach is fast and forward-looking. Its limitation is scope: only debt that actually trades has an observable price. Publicly issued corporate bonds and syndicated loans qualify; bilateral loans to mid-market borrowers do not. Market prices also absorb sentiment and liquidity conditions, which can drift from fundamental recovery value during stress.
Statistical Modeling
Rather than lean on portfolio-wide averages, statistical models predict LGD for a specific exposure from its characteristics. Typical inputs include seniority, collateral type and appraised value, borrower industry, loan-to-value ratio, and macroeconomic variables like unemployment or property price indices. The model produces a predicted LGD for each individual facility, which supports finer risk differentiation than a single portfolio average.
What Actually Drives the Number
Three factors explain most of the variation in realized LGD.
Seniority
Where a debt sits in the repayment hierarchy is the strongest single predictor of recovery. Senior secured creditors are paid first; whatever remains flows to senior unsecured creditors, then to subordinated holders. Historical data on defaulted corporate bonds shows the pattern clearly. Over a multi-decade study period, senior secured bonds recovered an average of roughly 50 cents on the dollar (about 50% LGD), senior unsecured bonds recovered about 33 cents (67% LGD), and subordinated bonds recovered around 27 cents (73% LGD).3Moody’s Investors Service. Recovery Rates on Defaulted Corporate Bonds and Preferred Stocks Standard deviations run 20 to 27 percentage points, so any single default can land far from the average, but the ranking is consistent.
Collateral
Collateral only helps when the lender’s claim is legally enforceable and the asset holds its value through the workout. A properly filed security interest in commercial real estate offers meaningful downside protection. An improperly documented lien on depreciating equipment often does not. Real estate and financial assets tend to retain value and have liquid resale markets. Specialized industrial equipment, perishable inventory, and intangibles like intellectual property are harder to liquidate and typically produce lower recoveries. Frequent, conservative appraisals matter because LGD models built on stale valuations underestimate losses.
Macroeconomic Conditions
LGD is procyclical. When the economy weakens, default rates rise at the same time recovery values fall. Property prices drop, buyers for distressed assets disappear, and workout timelines stretch as courts and servicers get busier. That correlation is a core reason regulators require “downturn LGD” rather than through-the-cycle averages.
LGD in the Basel Capital Rules
Internationally active banks calculate regulatory capital under the Basel framework, and LGD is a required input. Banks using the Internal Ratings-Based (IRB) approach choose between the Foundation approach, where regulators set the LGD values, and the Advanced approach, where banks estimate LGD from their own data subject to supervisory approval.4Bank for International Settlements. CRE36 – IRB Approach: Minimum Requirements to Use IRB Approach
Foundation IRB
Under the Foundation approach, banks don’t model LGD themselves. The framework prescribes fixed values. Senior unsecured claims on most corporates receive a 40% LGD; senior claims on banks, sovereigns, and other financial institutions carry 45%. Subordinated claims receive 75%.5Bank for International Settlements. CRE32 – IRB Approach: Risk Components
Eligible collateral lowers the number. Exposures fully secured by eligible financial collateral can receive an LGD as low as 0%. Real estate and receivables collateral bring the LGD down to 20% for the secured portion; other physical collateral gets 25%.5Bank for International Settlements. CRE32 – IRB Approach: Risk Components These prescribed values act as regulatory benchmarks and align reasonably with long-run historical recovery data on corporate debt.
Advanced IRB
Banks approved for the Advanced approach estimate their own LGD per exposure using internal models and historical workout data. Institutions gain risk sensitivity but must demonstrate robust data, validated models, and strong governance. Each estimate must reflect a minimum observation period and rest on the bank’s own loss experience.4Bank for International Settlements. CRE36 – IRB Approach: Minimum Requirements to Use IRB Approach
Downturn LGD
Regulators don’t accept fair-weather loss estimates for capital purposes. LGD used for capital must reflect economic downturn conditions where relevant, and it cannot fall below the long-run default-weighted average loss rate. Where recoveries deteriorate in stress periods, banks must incorporate that cyclicality.4Bank for International Settlements. CRE36 – IRB Approach: Minimum Requirements to Use IRB Approach European regulators have further specified that downturn LGD estimates, including a margin of conservatism, should sit at least 15 percentage points above the long-run average LGD, capped at 105%.6European Banking Authority. Guidelines for the Estimation of LGD Appropriate for an Economic Downturn The point of a capital buffer is to absorb losses when conditions are worst, not when they’re average.
When Default Is Triggered
LGD tracking begins at the moment of default, so the definition of default matters. Under Basel, a borrower is in default when any material credit obligation is more than 90 days past due. Default is also recognized when the bank places the loan on non-accrual, takes a material write-down, agrees to a distressed restructuring, or concludes the borrower is unlikely to repay in full without the bank resorting to actions like seizing collateral.1Bank for International Settlements. CRE20 – Standardised Approach: Individual Exposures A bankruptcy filing counts regardless of how current the payments are.
Where LGD Sits in the Expected Loss Formula
LGD is one of three inputs in the Expected Loss (EL) formula banks use to estimate the average credit loss a portfolio will produce over a given horizon: EL = PD × LGD × EAD, where PD is the probability of default and EAD is the exposure at default.7Bank for International Settlements. CRE35 – IRB Approach: Treatment of Expected Losses and Provisions
Each piece answers a different question. PD asks how likely default is. EAD asks how much is at stake. LGD asks how bad it is when default happens. A loan with a 2% PD, $1 million EAD, and 40% LGD carries an expected loss of $8,000. That figure drives pricing, provisioning, and portfolio-level capital allocation. Expected Loss covers the average outcome; regulatory capital exists to absorb the unexpected losses beyond it, which is why the Basel framework runs the same PD, LGD, and EAD inputs through a risk-weight function calibrated to tail risk.
LGD Under CECL
The Current Expected Credit Losses standard, FASB Topic 326, changed how U.S. financial institutions use LGD for accounting. Under the prior incurred-loss model, banks recognized losses only after a trigger event. CECL requires institutions to estimate expected credit losses over the full remaining life of each financial asset, pulling loss recognition forward into current provisions.8National Credit Union Administration. CECL Accounting Standards
That lifetime horizon differs from the Basel IRB framework, which typically uses a one-year default window. Under CECL, LGD estimates must cover the entire contractual term and incorporate a reasonable and supportable forward-looking forecast of economic conditions; beyond that forecast horizon, the institution reverts to historical loss experience. LGD figures feeding accounting provisions therefore tend to be more forward-looking and scenario-dependent than those used purely for capital calculation.