Performance metrics

Four metrics dominate ABF collateral analysis: CDR, CPR, CNL, and severity. Add roll rate analysis and you have the complete toolkit for measuring, projecting, and benchmarking loan pool performance.

This page covers how to calculate these metrics, interpret them in context, and use them for forward projections.

CDR: constant default rate

CDR is the annualized rate at which loans default. It’s the standard measure of credit performance for monthly-pay assets and the primary input to cash flow models.

Calculation

Step 1: Calculate monthly default rate (MDR)

MDR = Defaults This Month / Beginning UPB

Step 2: Annualize to CDR

CDR = 1 - (1 - MDR)^12

The annualization assumes the same monthly rate persists for 12 months. It’s a standardized way to express default rates regardless of actual observation period.

Worked example

Component	Value
Beginning UPB	$50,000,000
Defaults this month	$250,000

Calculation:

MDR = $250,000 / $50,000,000 = 0.50%

CDR = 1 - (1 - 0.0050)^12
    = 1 - (0.9950)^12
    = 1 - 0.9418
    = 5.82%

Definition nuances

What counts as “default”? This varies by originator and deal:

90+ days past due (common for consumer)
120+ days past due (common for mortgage)
Charge-off (accounting-triggered)
Bankruptcy filing
Repossession (secured assets)

Clarify the definition before comparing CDR across pools or deals. A pool using 90+ DPD will show higher CDR than one using charge-off, even with identical performance.

Dollars vs. count: Standard practice uses dollar-weighted CDR (defaults in $ / UPB in $). Count-weighted CDR (loans defaulting / loan count) is sometimes used but understates risk if larger loans default more frequently.

CDR benchmarks by asset class

Asset class	Prime CDR	Near-prime CDR	Subprime CDR
Auto loans	1-2%	4-6%	8-15%
Consumer unsecured	3-5%	6-10%	10-18%
Credit cards	4-6%	7-12%	12-20%
Equipment	1-3%	3-5%	4-8%

A CDR of 6% on a prime consumer portfolio would be alarming. The same 6% on a subprime consumer portfolio might be in line with expectations and pricing.

CDR trends

Calculate CDR monthly and track trends:

Month	Beginning UPB	Defaults	MDR	CDR
Jan	$50.0M	$250K	0.50%	5.82%
Feb	$52.0M	$285K	0.55%	6.39%
Mar	$54.5M	$327K	0.60%	6.96%

Rising CDR indicates either credit deterioration or portfolio seasoning (younger loans reaching peak default period). Distinguish the two by running vintage-level CDR.

CPR: constant prepayment rate

CPR is the annualized rate at which loans prepay—paying off early beyond scheduled amortization. It affects WAL, yield, and reinvestment assumptions.

Calculation

Step 1: Calculate single monthly mortality (SMM)

SMM = Prepayments This Month / (Beginning UPB - Defaults)

Note: Defaults are excluded from the denominator because defaulted loans can’t prepay.

Step 2: Annualize to CPR

CPR = 1 - (1 - SMM)^12

Worked example

Component	Value
Beginning UPB	$50,000,000
Defaults	$250,000
Prepayments	$750,000

Calculation:

SMM = $750,000 / ($50,000,000 - $250,000)
    = $750,000 / $49,750,000
    = 1.51%

CPR = 1 - (1 - 0.0151)^12
    = 1 - (0.9849)^12
    = 1 - 0.8345
    = 16.55%

Why CPR matters

WAL impact: Higher CPR shortens weighted average life. A 60-month pool at 0% CPR has ~30-month WAL. At 20% CPR, WAL drops to ~22 months.

Yield impact: Prepayments return principal but not future interest. High CPR reduces total interest income over the pool’s life.

Reinvestment risk: In revolving facilities, prepaid principal must be reinvested. If new originations have lower rates or worse credit, CPR becomes an adverse selection mechanism.

CPR drivers

Rate environment: In a falling rate environment, borrowers refinance, and CPR spikes. This is most pronounced for mortgage but affects all consumer products with fixed rates above market.

Seasoning: Very new loans rarely prepay (prepayment penalty, transaction costs not recouped). CPR typically ramps up from months 6-12, stabilizes months 12-36, then declines as remaining loans are less refinance-sensitive.

Credit quality: Counterintuitively, better credit borrowers prepay faster—they have more refinancing options. Subprime borrowers tend to have lower CPR.

Economic conditions: Strong economy = more payoffs from home sales, business sales, wealth events. Weak economy = lower CPR as borrowers stay put.

CPR benchmarks

Asset class	Typical CPR range
Auto loans	12-20%
Consumer unsecured	15-25%
Equipment	8-15%
Mortgage	5-30% (highly rate-sensitive)

CPR varies significantly with rate environment. The ranges above assume moderate rate conditions.

CNL: cumulative net loss

CNL is total losses to date as a percentage of original pool balance. Unlike CDR (a rate), CNL is a running total that only increases over time.

Calculation

CNL = Cumulative Net Losses / Original Pool Balance

Where:
Net Losses = Gross Losses (Principal + Accrued Interest) - Recoveries

Worked example

Component	Value
Original pool balance	$100,000,000
Cumulative gross losses	$6,500,000
Cumulative recoveries	$1,200,000
Cumulative net losses	$5,300,000

Calculation:

CNL = $5,300,000 / $100,000,000 = 5.30%

CDR vs. CNL

Metric	Type	Best for
CDR	Rate (annualized)	Comparing periods, cash flow modeling, triggers
CNL	Cumulative total	Static pool tracking, terminal loss, credit enhancement sizing

A pool can have declining CDR but increasing CNL (defaults slowing but still occurring). A pool can have stable CDR but rapidly increasing CNL (early seasoning, defaults just starting).

Using CNL

Terminal CNL: The final CNL when a pool is fully liquidated. This is your actual realized loss—the number that matters for credit enhancement adequacy.

Projected terminal CNL: Using seasoning adjustments from static pool analysis, project where current CNL will end. Compare to original assumptions.

Trigger analysis: Many deals have CNL triggers (e.g., if CNL exceeds 5%, revolving period ends). Track CNL against trigger levels monthly.

Loss severity

Severity is the percentage of principal lost when a loan defaults. It’s the complement of recovery rate.

Calculation

Severity = (Principal at Default + Accrued Interest + Expenses - Recoveries) / Principal at Default

Or simply:

Severity = 1 - Recovery Rate

Worked example

Component	Value
Principal at default	$25,000
Accrued interest	$500
Collection expenses	$1,200
Gross claim	$26,700
Collateral proceeds	$12,000
Deficiency judgment	$2,000
Total recoveries	$14,000

Calculation:

Net loss = $26,700 - $14,000 = $12,700

Severity = $12,700 / $25,000 = 50.8%

Severity above 100%

Severity can exceed 100% when accrued interest and collection expenses exceed recoveries. For unsecured loans with extended delinquency before charge-off:

Component	Value
Principal at default	$10,000
Accrued interest (6 months)	$600
Collection costs	$1,500
Gross claim	$12,100
Recoveries	$800
Net loss	$11,300
Severity	113%

This is common for unsecured consumer loans and credit cards.

Severity benchmarks

Asset class	Typical severity
Prime auto	35-50%
Subprime auto	50-70%
Consumer unsecured	80-100%+
Credit cards	85-100%+
Equipment (titled)	40-60%
Equipment (soft collateral)	70-90%

Unsecured assets have high severity because there’s no collateral to liquidate. Secured assets have lower severity, but liquidation takes time and costs money.

Recovery timing

Severity calculations depend on when you cut off recovery efforts. A loan charged off today might have 95% severity at month 3, but recoveries over the next 24 months bring it down to 75%.

Standard practice:

Calculate severity at fixed intervals (6, 12, 24 months post-default)
Report “liquidated severity” only for fully resolved accounts
Use historical recovery curves to project ultimate severity for recent defaults

Roll rate analysis

Roll rate analysis tracks how loans move between delinquency states. It answers: if a loan is 30 DPD this month, what’s the probability it cures, stays 30, rolls to 60, or defaults?

Building a transition matrix

Track month-over-month state changes for all loans:

From \ To	Current	30 DPD	60 DPD	90 DPD	Default	Paid Off
Current	94.0%	3.5%	0.0%	0.0%	0.0%	2.5%
30 DPD	45.0%	30.0%	20.0%	0.0%	0.0%	5.0%
60 DPD	25.0%	10.0%	25.0%	35.0%	0.0%	5.0%
90 DPD	10.0%	5.0%	5.0%	20.0%	55.0%	5.0%

Each row sums to 100%. Each cell is the probability of transitioning from the row state to the column state in one month.

Reading the matrix

Row 1 (Current loans):

94% stay current next month
3.5% roll to 30 DPD (first delinquency)
2.5% pay off entirely
0% skip directly to 60+ DPD (not allowed in this model)

Row 2 (30 DPD loans):

45% cure back to current (good!)
30% stay at 30 DPD (make partial payment)
20% roll to 60 DPD (worsen)
5% pay off (settle or payoff in full)

Row 4 (90 DPD loans):

Only 10% cure to current
55% default next month
The rest either make partial payments or pay off

What roll rates reveal

Cure rates (movement left in the matrix) indicate servicer effectiveness and borrower quality. High cure rates from 30 DPD (>40%) suggest temporary payment difficulties, not fundamental credit problems.

Roll rates (movement right) indicate default pipeline. If the 60-to-90 roll rate spikes from 35% to 50%, expect higher defaults in 60-90 days.

Default probability from each state:

Using the matrix above and following loans through multiple transitions:

Starting state	Ultimate default probability
Current	~3%
30 DPD	~15%
60 DPD	~40%
90 DPD	~70%

These are steady-state probabilities assuming the transition matrix is stable.

Forward default projection

Use current delinquency and roll rates to project near-term defaults.

Current state:

State	UPB
30 DPD	$2.0M
60 DPD	$1.0M
90 DPD	$0.5M

Month 1 projection:

From 90 DPD: $0.5M × 55% = $275K expected defaults

From 60 DPD: $1.0M × 35% = $350K rolls to 90 DPD

From 30 DPD: $2.0M × 20% = $400K rolls to 60 DPD

Month 2 projection:

90 DPD after month 1: $350K (new from 60) + $0.5M × 20% (stayed 90) = $450K

Expected defaults month 2: $450K × 55% = $248K

Bottoms-up projection:

Rolling forward 3-6 months using the transition matrix gives you a bottoms-up default projection based on current delinquency inventory—more grounded than simply applying historical CDR.

Monitoring roll rate changes

Track transition probabilities monthly:

Date	30→Current	60→90	90→Default
Jan	45%	35%	55%
Feb	43%	37%	58%
Mar	40%	42%	62%

Deterioration is evident: cure rates declining, roll-forward rates increasing, default rates rising. This is a leading indicator of increasing losses before they show up in CDR.

Connecting metrics

These metrics work together in analysis:

Loss projection formula

Expected Loss = Default Balance × Severity

Where:
Default Balance = UPB × CDR (annualized) or use roll-rate projection

Expected Annual Loss Rate = CDR × Severity

Example:

Pool UPB: $50M
CDR: 6%
Severity: 65%

Expected annual defaults: $50M × 6% = $3.0M
Expected annual net losses: $3.0M × 65% = $1.95M
Expected CNL (annualized): $1.95M / $50M = 3.9%

WAL calculation using CPR

WAL = Σ (Principal Payment_t × t) / Total Principal

Where:
Principal Payment_t = Scheduled amortization + Prepayments + Defaults

Higher CPR accelerates principal payments, reducing WAL. Higher CDR also shortens WAL (defaulted loans return principal, albeit with losses).

Break-even analysis

What CDR can the excess spread absorb?

Break-even CDR = Excess Spread / Severity

Where:
Excess Spread = WAC - Cost of Funds - Servicing - Losses (existing)

Example:

WAC: 10.0%
Cost of funds: 5.5%
Servicing: 0.5%
Current loss rate: 2.0%
Remaining excess: 2.0%
Severity: 65%

Break-even CDR = 2.0% / 65% = 3.1% additional CDR absorbable
Current CDR + Break-even = 3.1% + existing CDR before loss coverage exhausted

Benchmarking

Metrics without context are just numbers. Benchmark against:

Internal benchmarks

Historical performance of the same pool
Other pools from the same originator
Budget/forecast assumptions

External benchmarks

Public ABS performance:

EDGAR ABS-EE filings (SEC-required loan-level data)
10-D distribution reports (monthly performance summaries)

Rating agency indices:

S&P Auto ABS Performance Indices
Moody’s Consumer ABS Index
Fitch Prime Auto Index

Industry surveys:

TransUnion/Equifax delinquency trends
Federal Reserve consumer credit data

Adjustment factors

Don’t benchmark blindly. Adjust for:

Factor	Adjustment
Credit quality	Near-prime vs. prime: +100-200 bps CDR
Vintage	2023 vs. 2019 economic conditions
Product type	Secured vs. unsecured
Pool structure	Public ABS (tight eligibility) vs. warehouse (broader)

Build a comps table showing your pool characteristics alongside comparable benchmarks, with adjustment rationale.

Key takeaways

CDR measures the rate of default. Annualized monthly defaults, standard for period-over-period comparison and cash flow modeling.
CPR measures prepayment speed. Affects WAL, yield, and reinvestment. Highly sensitive to rate environment.
CNL measures cumulative loss. Running total against original balance. The ultimate measure of credit performance.
Severity determines loss given default. Varies dramatically by collateral. Can exceed 100% for unsecured.
Roll rates reveal the default pipeline. Monitor transitions between delinquency states to project near-term losses.
Benchmark everything. Metrics without context are meaningless. Find comparable pools and adjust for differences.