Loss Severity
Definition
Loss severity is the share of a defaulted exposure that is not recovered after collateral sales, guarantees, workouts, or other recoveries. It is the complement of recovery rate.
Why it matters
Default rate tells you how often loans fail; loss severity tells you how bad failures are. A real estate lender can survive elevated defaults if collateral recoveries are strong, but high default frequency plus high severity quickly erodes equity and investor trust.
Common misconceptions
- •Default rate and loss severity are different dimensions.
- •Low LTV does not guarantee low severity if collateral marks are stale or foreclosure costs are high.
- •Severity should be measured after legal costs, taxes, repairs, and sale expenses.
- •A portfolio-average severity can conceal large differences by lien position, collateral type, vintage, jurisdiction, and resolution strategy.
Technical details
Simple formula
Loss severity = (exposure at default - net recovery) / exposure at default.
Example: $250,000 defaulted loan, $190,000 net recovery after sale and costs. Loss severity = 24%.
Recovery rate is 76% in the same example.
Exposure at default
The denominator should state whether it includes principal only or also accrued interest, protective advances, fees, and undrawn commitments funded before resolution.
Changing the exposure definition can materially change reported severity even when recovery cash is identical.
Gross vs net severity
Gross recovery from a collateral sale should be reduced by senior liens, taxes, legal costs, repairs, insurance, servicing, and selling expenses.
Investor-level severity may differ again after platform fees and the note waterfall allocate proceeds.
Portfolio expected loss
A common approximation is expected loss = probability of default x loss severity. A 6% default rate and 35% severity imply roughly 2.1% expected principal loss before timing effects.
Correlation and changing collateral values can make stress-period severity higher precisely when defaults also rise.
Cohort and recovery analysis
Compare severity by origination vintage, collateral type, seniority, state, servicer, and workout path rather than relying only on a blended lifetime average.
Include unresolved defaults separately because excluding open workouts can make completed-case severity appear artificially favorable.
Lien-position waterfall
Apply net collateral proceeds first to taxes, preservation expenses, and legally senior claims, then through first-lien, second-lien, mezzanine, and equity positions. The same asset sale can produce low severity for senior debt and total loss for junior debt.
Report attachment point and exposure at each level rather than one property-wide recovery percentage.
Open-case bias
Completed workouts often resolve faster and more favorably than difficult cases still open. Calculating severity only on closed defaults can therefore overstate recovery quality.
Include unresolved exposure with scenario-weighted estimates, show results both with and without open cases, and update estimates as bids and legal milestones replace appraisals.
How it shows up in deals
Loss Severity usually appears in private credit offering documents, collateral schedules, note purchase agreements, servicing reports, investor update memos, or workout summaries. The label alone is not enough; the investor has to know whether it controls cash timing, collateral eligibility, reserve release, default treatment, loss recognition, or recovery priority.
Example context: in a marketplace-credit note, the term may determine which loans can enter the pool, when cash is trapped, or when a servicer must move a borrower from performing to delinquent status. In a real-estate credit note, the same concept may affect draw approvals, collateral release, foreclosure timing, or property-level recovery assumptions. In a small-business credit pool, it may determine whether renewed contracts are treated as clean payoffs or as refinanced exposure.
The practical test is: if this definition changed by 10%-20%, would investor cash flows change? If the answer is yes, the term belongs in the actual underwriting model, not just the glossary. Investors should map it to the waterfall, reserve account, loan tape, reporting package, and manager discretion rights before relying on the sponsor's summary.
Diligence questions
Definition source: identify the controlling definition in the PPM, offering circular, note indenture, servicing agreement, collateral eligibility schedule, or monthly report. Sponsors sometimes use a clean marketing definition while the legal documents contain exceptions, cure periods, manager discretion, or alternate calculations that matter more under stress.
Calculation owner: confirm who calculates the metric or status, how frequently it is updated, and what data supports it. A monthly servicer calculation based on borrower-reported data is not the same as a daily controlled-account calculation tied to cash receipts. If a third-party administrator, trustee, or backup servicer receives the data, confirm whether it independently verifies anything or merely republishes sponsor files.
Cash impact: determine whether the term affects payment priority, eligibility, borrowing base availability, concentration limits, delinquency migration, default triggers, reserve releases, overcollateralization tests, investor distributions, or early amortization. Terms that change cash priority deserve more scrutiny than terms used only for descriptive reporting.
Stress behavior: ask what happens when the metric deteriorates. Does cash trap immediately, is there a 10-30 day cure period, can the manager waive the breach, can new collateral be substituted, does the reserve step up, or does the deal merely disclose the issue? Protective terms are only useful if the remedy activates before collateral value has already leaked away.
Documentation to review
Core documents: review the PPM or offering circular, subscription documents, note purchase agreement, indenture, collateral schedule, servicing agreement, waterfall model, tax disclosures, investor reporting package, and any historical performance exhibits. If the deal references a separate credit policy or servicing standard, request that document too; many important definitions live outside the glossy memo.
Collateral evidence: for loan pools, inspect a representative loan tape with origination date, borrower type, balance, rate, maturity, collateral value, delinquency status, charge-off status, recovery status, and concentration fields. For asset-backed notes, review appraisal files, custody records, insurance certificates, account-control agreements, title documents, and collateral release conditions.
Structural evidence: confirm whether there is a reserve account, lockbox, overcollateralization test, borrowing-base certificate, servicer report, backup-servicer agreement, trustee report, and amendment threshold. A structure with clear monthly reporting and hard cash controls is materially different from a structure where the issuer calculates everything and remits only after discretionary expenses.
Definition reconciliation: compare the sponsor's definition with industry usage and with adjacent terms in the same documents. If a sponsor defines 'default' only after 120 days but stops reporting loans as current after 30 days, the difference can shift performance optics. If 'fair value' can be based on manager marks without recent transactions, reported NAV may lag economic loss.
Reporting and risk signals
Good reporting separates beginning exposure, new originations or purchases, principal collections, interest or fee collections, realized losses, recoveries, servicing fees, reserve activity, fair-value marks, amendments, extensions, and ending exposure. The strongest packages tie each status label to cash: what came in, what was written down, what was reserved, and what remains at risk.
Watch-list signals include delayed reports, one-time manual adjustments, definition changes, rising extensions, higher renewal or refinancing activity, large unexplained cures, servicer commentary that emphasizes gross collections without net loss data, and performance that improves while actual cash distributions do not. These are not automatic red flags, but they are reasons to ask for the bridge from reported status to investor cash.
Numeric sensitivity matters. Example: a pool with 20% gross yield, 5% annual defaults, 40% recoveries, and 3% servicing/platform cost may look attractive. If defaults rise to 12%, recoveries fall to 20%, and collections lag by 90 days, net investor yield can compress sharply or turn negative even though the headline coupon or factor-rate economics appear high.
Investor action: build a simple downside bridge. Start with expected gross cash, subtract fees, subtract losses net of recoveries, delay collections by one or two reporting periods, then test whether reserves and overcollateralization still cover promised distributions. If the term cannot be mapped into that bridge, it may be descriptive rather than protective.
