WARR (Weighted Average Recovery Rate)
Definition
WARR measures the expected recovery rate of a CLO's collateral pool in default scenarios, calculated as the weighted average of individual loan recovery assumptions based on seniority and collateral type. First-lien senior secured loans typically assume 70% recovery, second-lien loans 30% recovery, unsecured loans 20% recovery. WARR is calculated by multiplying each loan's par amount by its recovery rate assumption, summing across the portfolio, and dividing by total par. A CLO with 95% first-lien, 5% second-lien collateral would have WARR ≈ 68%. Rating agencies use WARR alongside WARF (default probability) to model expected portfolio losses and determine tranche ratings.
Why it matters
WARR directly impacts CLO tranche ratings and required subordination levels. Higher WARR (more senior secured collateral) reduces expected loss severity, allowing thinner subordination for equivalent ratings—a CLO with 70% WARR might achieve AAA at 30% subordination, while 60% WARR requires 35% subordination for the same rating. This creates structural incentive for CLO managers to favor first-lien loans over second-lien despite higher second-lien yields, as recovery assumptions dominate loss calculations. During 2008-2009, realized recoveries fell to 50-60% (vs 70% assumptions), causing rating agency model recalibrations and massive downgrades. Understanding WARR assumptions versus realized recovery experience is critical for stress testing CLO tranches—if actual recoveries underperform assumptions by 10 points, BBB tranches can face losses while models predicted safety.
Common misconceptions
- •WARR assumptions are not guarantees—they're statistical averages. Individual loan recoveries range from 0-100%, and correlation matters more than average during stress.
- •Higher WARR doesn't always mean lower risk. A portfolio concentrated in cyclical industries might have high WARR assumptions but correlated default/recovery risk.
- •Recovery rates aren't static—they vary with default timing, collateral quality, bankruptcy duration, and market liquidity. 70% assumption can become 40% reality in dislocated markets.
Technical details
Recovery rate assumptions by seniority
Moody's recovery assumptions: First-lien senior secured = 65-70%, Second-lien senior secured = 30-35%, Senior unsecured = 40-45%, Subordinated unsecured = 20-25%. S&P uses similar but slightly higher assumptions (70%/35%/45%/25%).
Collateral-specific adjustments: Asset-backed first lien (equipment, inventory) = 75%, Cash-flow first lien (no hard assets) = 65%, Covenant-lite first lien = haircut of 5 points (concern about lender control loss). These adjustments reflect variation in collateral quality and lender protections.
Geographic considerations: US loans assume 70% recovery based on historical experience. European loans assume 60-65% due to longer workout timelines and less creditor-friendly bankruptcy regimes. Asian loans assume 50-60% given less developed restructuring markets.
Example WARR calculation: $400M first-lien (70% recovery) + $50M second-lien (35% recovery) + $50M unsecured (25% recovery). WARR = ($400M × 70% + $50M × 35% + $50M × 25%) / $500M = ($280M + $17.5M + $12.5M) / $500M = 62%.
WARR constraints in CLO structures
Minimum WARR covenant: Most CLOs require WARR ≥ 60-65% to maintain ratings. Managers cannot trade into second-lien or unsecured positions if it would violate minimum WARR. This creates binding constraint on portfolio composition—managers may want second-lien exposure for yield but cannot exceed WARR limits.
WARR interaction with CCC buckets: CCC-rated loans typically receive 50-60% recovery assumption (lower than first-lien average). High CCC concentration reduces WARR and may trigger WARR covenant violations even if loans are first-lien. Creates compound effect where credit deterioration impacts both default probability (WARF) and recovery (WARR).
Dynamic WARR management: As portfolio migrates from BB to CCC, WARR deteriorates even without trading. Managers must actively trade up in recovery (sell unsecured, buy first-lien) to maintain WARR compliance. This forced trading can occur at disadvantageous prices during stress when WARR pressure peaks.
WARR vs subordination trade-off: Rating agencies allow looser WARF if compensated by higher WARR, and vice versa. Example: 65% WARR portfolio might tolerate 2900 WARF for BBB rating, while 70% WARR portfolio tolerates 3100 WARF. Managers optimize this trade-off based on market opportunities.
Historical recovery experience vs assumptions
Pre-2008 experience: Actual first-lien recoveries averaged 70-75%, validating agency assumptions. Second-lien averaged 30-35%. Models appeared accurate during benign credit conditions.
2008-2009 stress: First-lien recoveries collapsed to 55-65%, second-lien to 15-25%. Dislocated markets, fire sales, and minimal M&A activity depressed outcomes. Senior secured bonds fared even worse (40-50% recoveries) as buyers scarce.
2020 experience: First-lien recoveries held 65-75% due to swift market recovery, government support, and SPV buying. However, exit multiples compressed—companies that might have recovered 80% in 2019 achieved 65% in 2020 due to sector distress.
Key insight: Recovery assumptions reflect long-run averages but fail to capture tail risk. 10th percentile recovery outcomes (relevant for senior tranche stress) are 20-30 points lower than mean assumptions. This explains why AAA tranches faced losses despite 'safe' positions in waterfall.
Implications for tranching and rating
Subordination sizing: Rating agencies size subordination to withstand default scenarios with WARR-adjusted losses. Example: 30% cumulative default rate × (100% - 70% WARR) = 9% expected loss. AAA tranche sized at 60% of capital structure can absorb 40 points of subordination eating 9% losses plus cushion.
WARR deterioration scenarios: If WARR falls from 70% to 60%, same 30% default rate produces 12% expected loss (vs 9%), requiring 33% more subordination for equivalent protection. Rating agencies may downgrade tranches or require deleveraging to restore subordination.
Monte Carlo modeling: Agencies run thousands of scenarios varying both default rates (WARF-driven) and recovery rates (WARR-driven). Correlation between defaults and recoveries (both worsen in stress) creates tail risk. Models assume 0.3-0.5 correlation but 2008 showed 0.7+ correlation in practice.
Portfolio composition incentives: WARR methodology creates preference for first-lien regardless of relative value. If second-lien loans offer L+700 vs first-lien L+400, the 300 bps pickup may not compensate for subordination pressure from WARR deterioration. This perpetuates CLO focus on first-lien despite cyclical pricing.