Recovery Rate Modeling: LGD, Seniority, and Cycles

Key Takeaways

• Historical average recoveries: senior secured loans ~75%, senior unsecured bonds ~38%, subordinated bonds ~15% (Moody's data).
• Recovery rates and default rates are inversely correlated; holding LGD fixed in stressed simulations understates tail credit losses by 30–50%.
• Debt cushion, the share of capital structure sitting below a tranche, is a key determinant of that tranche's expected recovery.
• Jurisdiction matters: US Chapter 11 favors going-concern value preservation; many European regimes lean toward liquidation with lower recoveries.

What It Is

The recovery rate is expressed as a percentage of par, either at default (trading-price recovery) or at emergence from bankruptcy (ultimate recovery). Loss given default (LGD) is the complement: LGD = 1 - Recovery.

Moody's and S&P publish annual default and recovery studies that segment by seniority, security, and industry. Historical averages from Moody's Ultimate Recovery Database show senior secured bonds recover roughly 65 percent on average (median 67 percent), senior unsecured bonds around 38 percent, and junior subordinated bonds near 15 percent. Senior secured bank loans recover about 75 percent, notably higher than senior unsecured bonds.

The Intuition

Recovery is driven by what creditors can actually lay claim to. A senior secured lender with a lien on accounts receivable, inventory, or hard assets gets paid first and often in full if the assets hold value. An unsecured bondholder stands behind the secured lenders, banks, and tax authorities, so they collect from whatever is left. Subordinated holders collect only after the unsecured layer is made whole, which rarely happens.

The second driver is the macro environment. Recovery rates fall sharply in recessions because the collateral itself loses value at the same time defaults spike. Moody's research shows the inverse correlation: in years with high default rates, average recovery rates are low, and vice versa. This correlation is exactly what simple LGD-equals-constant models miss.

How It Works

A basic recovery model estimates expected LGD as a function of three inputs: seniority and security, debt cushion, and industry or macro regime.

E[LGD] = f(Seniority, DebtCushion, Industry, DefaultRate)

Debt cushion is the share of the capital structure that sits below the tranche being modeled. A senior unsecured bond with 40 percent of the company's debt sitting as subordinated paper below it has a larger cushion than one with zero junior debt, so its expected recovery is higher.

Industry matters because asset tangibility varies. Airlines have pledgeable aircraft; software companies have little beyond intellectual property. Historical industry-level recovery tables are widely published.

More sophisticated models, following Altman and colleagues at the NYU Salomon Center, fit a log-linear relationship between aggregate default rates and recovery rates. When the default rate doubles, average recovery can fall 10 to 20 percentage points. Large banks run stochastic LGD Monte Carlo on top of correlated default simulations to capture this joint behavior.

Worked Example

A mid-market industrial firm issues three classes of debt totaling 800 million: a 300 million senior secured term loan, a 400 million senior unsecured bond, and a 100 million subordinated note. The firm enters Chapter 11 after a cyclical downturn.

The restructuring values the going concern at 500 million. The senior secured lenders have first claim and receive the full 300 million, for a 100 percent recovery. The senior unsecured holders have a claim of 400 million against the remaining 200 million of value, for a 50 percent recovery. The subordinated holders receive nothing, a 0 percent recovery.

A bank holding 20 million of the unsecured bond and carrying a pre-default 40 percent LGD assumption would have projected losses of 8 million. Actual realized loss is 10 million. The model underestimated LGD by 25 percent because it ignored debt cushion and the cyclical recovery drag. Scale that gap across a 10 billion high-yield book, and unmodeled recovery risk translates into hundreds of millions of unexpected loss in a bad year.

Common Mistakes

• Using a single constant LGD. A 40 percent flat assumption understates losses in recessions and overstates them in expansions. Basel-style "downturn LGD" floors attempt to fix this, but simple regression against macro factors works better for internal risk.
• Ignoring the default-recovery correlation. Holding recovery fixed while simulating correlated defaults makes tail losses look artificially smooth. Portfolio credit VaR can be 30 to 50 percent too low under this assumption.
• Using market-price recovery as final recovery. Post-default bond prices reflect what the market expects ultimately, but in distressed auctions and restructurings the actual outcome can differ by 20 points or more in either direction.
• Double-counting collateral value. Two tranches that both claim the same collateral cannot both recover 100 percent. The debt cushion and subordination schedule must be internally consistent.
• Ignoring jurisdiction. US Chapter 11 tends to preserve going-concern value; many European insolvency regimes favor liquidation. Cross-border portfolios need country-specific recovery curves.

Frequently Asked Questions

Why do recovery rates fall when default rates are high? High default rates typically coincide with recessions when asset values decline broadly. Collateral that backs senior secured loans is worth less, distressed buyer appetite for defaulted bonds diminishes, and the supply of defaulted assets overwhelms demand. The inverse correlation between default rates and recovery rates is well-documented in Moody's historical research and means that the worst credit years produce a double blow: more defaults and lower amounts recovered per default.

What is the difference between trading-price recovery and ultimate recovery? Trading-price recovery is the market price of the defaulted bond shortly after the credit event, which reflects the market's expectation of ultimate recovery discounted for time and uncertainty. Ultimate recovery is the actual cash or securities received by creditors at the completion of restructuring or bankruptcy, which can take months or years. Trading-price recovery is available quickly and is used in CDS settlement auctions; ultimate recovery is the more accurate economic measure but requires waiting for the process to conclude.

How does the debt cushion below a senior unsecured bond affect its recovery? The more subordinated capital sits below a senior unsecured bond in the capital structure, the more buffer absorbs losses before the senior unsecured tranche is touched. A senior unsecured bond with 40% of capital structure in subordinated instruments below it enters default with significant cushion. A senior unsecured bond with nothing subordinated below it immediately competes for recovery with all other senior creditors. Debt cushion is one of the most reliable predictors of senior unsecured recovery rates.

How do CDS settlement auctions determine recovery rates? When a CDS credit event occurs, ISDA facilitates a standard auction process to determine a single settlement price for all affected contracts. Participating dealers submit initial bids and offers for the deliverable bonds, and the process converges on a final price that reflects the market's collective assessment of recovery. All CDS contracts on that reference entity cash-settle at par minus the auction recovery price. The auction price becomes the de facto market measure of trading-price recovery.

Why is a flat 40% LGD assumption problematic for internal bank credit risk models? A constant 40% LGD does not vary with the credit cycle, which means it overstates losses in benign periods and understates them in stress. The empirical inverse correlation between default rates and recovery rates means that in a severe recession, when a bank's default rate doubles or triples, its recovery rate simultaneously falls. A model that holds LGD fixed at 40% will significantly understate credit losses in tail scenarios, potentially misleading capital allocation decisions.

Sources

1. Moody's Investors Service. Corporate Default and Recovery Rates, 1920-2006. https://www.moodys.com/sites/products/DefaultResearch/2006400000429618.pdf
2. Moody's Investors Service. Ultimate Recovery Database Summary. https://www.moodys.com/sites/products/defaultresearch/2006600000428092.pdf
3. Moody's Investors Service. Determinants of Recovery Rates on Defaulted Bonds. https://www.moodys.com/sites/products/defaultresearch/2003000000444168.pdf
4. Altman, E. NYU Stern Salomon Center. Default and Recovery Research archive. https://pages.stern.nyu.edu/~ealtman/Zscores.pdf

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.