Credit Risk Fundamentals › Loss Given Default
Quantitative FinanceAll chapters Credit Risk Fundamentals › Loss Given DefaultOne section · Sections in this chapter5 / 15 ↓
Reading tools Definitions & inline links
Choose your expertise level to adjust how many terms are explained. Beginners see more tooltips, experts see fewer to maintain reading flow. Hover over underlined terms for instant definitions.
Article links
Make inline references clickable
When default occurs, creditors rarely lose their entire exposure. Recovery comes through liquidation of assets, restructuring agreements, or bankruptcy proceedings. Bankruptcy processes aim to maximize creditor value, often through reorganization or liquidation. Loss Given Default (LGD) is the percentage of exposure not recovered after default:
where:
- : Loss Given Default, representing the percentage of exposure lost
- : percentage of exposure recovered through liquidation or restructuring
Estimating LGD is equivalent to estimating recovery rates. Both perspectives are useful. When thinking about what drives losses, the LGD framing is natural. When analyzing bankruptcy outcomes and workout processes, the recovery rate perspective may be more intuitive. The challenge lies in predicting, at the time of loan origination, what recovery rate will ultimately be realized years later under uncertain circumstances.
Factors Affecting Recovery Rates
Recovery rates vary significantly based on several factors, and understanding these drivers is essential for accurate LGD estimation. Let's look at the main factors.
Seniority in the capital structure determines the priority of claims in bankruptcy. The capital structure is essentially a queue: when assets are liquidated, senior secured debt holders are paid first, followed by senior unsecured, subordinated, and finally equity holders. This priority ordering is legally enforced through bankruptcy proceedings and significantly affects expected recoveries. A senior secured lender may recover most or all of their claim, while junior creditors may receive only pennies on the dollar.
# Historical average recovery rates by seniority
# Based on Moody's default and recovery studies
seniority_data = {
"Instrument": [
"Senior Secured Bank Loans",
"Senior Secured Bonds",
"Senior Unsecured Bonds",
"Senior Subordinated Bonds",
"Junior Subordinated Bonds",
],
"Average_Recovery": [0.72, 0.54, 0.40, 0.31, 0.24],
"Std_Dev": [0.25, 0.26, 0.26, 0.24, 0.22],
}
seniority_df = pd.DataFrame(seniority_data)
seniority_df["LGD"] = 1 - seniority_df["Average_Recovery"]Recovery Rates and LGD by Seniority ================================================================= Senior Secured Bank Loans | Recovery: 72.0% | LGD: 28.0% Senior Secured Bonds | Recovery: 54.0% | LGD: 46.0% Senior Unsecured Bonds | Recovery: 40.0% | LGD: 60.0% Senior Subordinated Bonds | Recovery: 31.0% | LGD: 69.0% Junior Subordinated Bonds | Recovery: 24.0% | LGD: 76.0%
The data shows a clear hierarchy of claims that reflects the legal priority structure: senior secured loans recover roughly 72% of their value, while junior subordinated bonds recover only 24%. This spread illustrates how capital structure priority directly impacts credit losses. The difference of nearly 50 percentage points in recovery rates translates directly into LGD estimates and, consequently, into dramatically different expected losses for lenders at different positions in the capital structure. This is why seniority is one of the most important factors in loan pricing and why senior secured lending commands lower spreads than unsecured or subordinated lending.
Collateral and security significantly affect recovery. Secured loans backed by specific assets (equipment, real estate, receivables) typically achieve higher recoveries than unsecured obligations because the lender has a direct claim on identifiable assets that can be seized and sold. The quality and liquidity of collateral matters enormously: marketable securities or prime real estate can be sold quickly at near-market prices, while specialized industrial equipment may fetch lower prices in distress as potential buyers are limited. You must carefully evaluate the current value of collateral and its likely liquidation value under stressed conditions.
Industry and economic conditions at the time of default influence recovery rates in ways that create unfortunate correlations. During systemic crises, when many firms default simultaneously, asset values are depressed and recoveries tend to be lower. If numerous steel companies default at once, the market for used steel equipment becomes glutted, driving down liquidation values. Industries with tangible assets (utilities, telecommunications) generally see better recoveries than those with primarily intangible assets (technology, services), where the value resides in intellectual property, brand names, and human capital that may evaporate in bankruptcy.
# Simulate recovery rate distribution (bimodal, as observed empirically)
np.random.seed(123)
n_defaults = 1000
# Recovery rates tend to be bimodality, either very low or relatively high
low_recovery = np.random.beta(2, 8, size=int(n_defaults * 0.4)) * 0.5
high_recovery = 0.5 + np.random.beta(5, 3, size=int(n_defaults * 0.6)) * 0.5
recovery_rates = np.concatenate([low_recovery, high_recovery])
The distribution of recovery rates is non-normal, often exhibiting bimodality; defaults either result in near-total loss or substantial recovery, with fewer outcomes in between. This pattern emerges because some defaults involve fundamentally unviable businesses where asset values collapse, while others represent temporary liquidity problems in otherwise sound enterprises that can be successfully restructured. This characteristic makes LGD estimation challenging and suggests that using a single point estimate like 40% oversimplifies the true risk. The wide standard deviation around mean recovery rates implies substantial uncertainty in any individual LGD estimate, which must be accounted for in risk management.
Downturn LGD
Regulators require banks to consider "downturn LGD" (recovery rates during stressed economic conditions) for capital calculations. Historical evidence shows that recoveries decline precisely when default rates spike, creating an unfortunate correlation that amplifies credit losses during crises. This phenomenon represents one of the most challenging aspects of credit risk: the parameters that determine losses are not independent but move together in adverse ways during stressed periods.
# Illustrate correlation between default rates and recovery rates
years = np.arange(2000, 2024)
# Simulated annual default rates (elevated during recessions)
default_rates = np.array(
[
0.012,
0.015,
0.025,
0.018,
0.010,
0.008,
0.007,
0.008,
0.015,
0.045,
0.032,
0.022,
0.015,
0.012,
0.010,
0.012,
0.015,
0.012,
0.010,
0.025,
0.018,
0.015,
0.020,
0.018,
]
)
# Recovery rates inversely correlated with default rates
recovery_rates_ts = (
0.50
- 1.5 * (default_rates - default_rates.mean())
+ 0.02 * np.random.randn(len(years))
)
recovery_rates_ts = np.clip(recovery_rates_ts, 0.25, 0.65)
The inverse relationship between default rates and recovery rates creates a double hit for creditors during downturns. More borrowers default, and each default results in larger losses. This correlation arises from common underlying factors: economic recessions depress both corporate profitability (increasing defaults) and asset values (reducing recoveries). Understanding and modeling this correlation is essential for stress testing and capital planning, as it means that average-cycle estimates of LGD may significantly understate losses during the periods that matter most.