Back

IC Report on Subprime Auto ABS

From prospectus upload to full IC package in minutes

SE

Shariff Elkordy

March 2026

I uploaded the prospectus and loan tape for an Exeter Automobile Receivables Trust deal (EART 20XX-X) and ran the full underwriting workflow. Within minutes, Graam's agent produced a complete IC report.

The agent parsed the deal structure, extracted a 10-tranche sequential-pay waterfall with an excess turbo mechanism, and validated it against the flows engine. It then stratified the 59,907-loan tape by FICO, rate, term, new/used split, geography, and vehicle make — and queried EDGAR shelf performance to build empirical loss curves from matched Exeter deals.

What made this interesting: the matched comps were Exeter subprime deals (avg FICO 598, avg rate 21.8%), while EART 20XX-X is a near-prime pool (FICO 696, rate 9.0%). The agent flagged the mismatch and noted that the base curves likely overstate losses by 40–60% — a conservative bias that actually strengthens the case for the senior tranches.

Result: all seven offered tranches (A1 through E) cleared the waterfall with no shortfalls under the base scenario. The E tranche showed a negative yield at par, confirming its equity-like first-loss character.

Here's the full IC-style report generated from that workflow.

Prepared for: Investment Committee

Asset Class: Prime/Near-Prime Auto ABS

Deal Reference: @exeter_auto

1. Executive Summary

Exeter Automobile Receivables Trust 20XX-X is a sequential-pay auto loan ABS backed by a pool of 59,907 retail automobile contracts with a collateral balance of $1,670M (tape) and a deal issuance balance of $1,134.5M. The pool carries a WA FICO of 696 and a WAC of 21.1% (deal) / 9.0% (tape WA rate), with a WA original term of 75 months.

Key Pool Metrics

MetricValue
Issuance Balance$1,134,491,333
Collateral Balance (Tape)$1,670,000,000
Loan Count59,907
WA FICO696
WAC (Deal / Tape)21.1% / 9.0%
WA Original Term75 months
Closing Date

The capital structure comprises 10 tranches in a sequential-pay waterfall with an excess turbo mechanism. Under base assumptions (CDR avg 13.85%, CPR avg 9.53%, severity avg 62.5%), all offered tranches from A1 through E pay in full with no shortfalls.

The E tranche (6.69% coupon) shows a negative yield at par under base assumptions, suggesting it is priced to absorb first-loss risk and should be evaluated at a discount.

2. Deal Structure

2a. Capital Stack

The deal is structured as a sequential-pay transaction. Principal flows A1 → A2 → A3 → B → C → D → E. Losses are allocated in reverse seniority (CERTIFICATE → E → D → C → B → A3 → A2 → A1).

TrancheBalanceCouponType% of Deal
A1$94,200,0003.90%Senior8.3%
A2$241,000,0004.08%Senior21.2%
A3$226,420,0004.03%Senior20.0%
B$118,510,0004.22%Mezzanine10.5%
C$123,000,0004.40%Mezzanine10.8%
D$162,310,0005.00%Subordinate14.3%
E$103,340,0006.69%Subordinate9.1%
N$30,000,0006.45%Certificate2.6%
RESERVE$11,232,587N/AReserve Fund1.0%
CERTIFICATE$24,478,746N/AFirst Loss2.2%
OC Validation: Collateral balance (tape): $1,670M. Sum of all tranche balances: $1,134.5M. Overcollateralization: $535.5M (~47.2% of deal balance). The substantial OC reflects the difference between the full tape and the deal issuance balance — consistent with Exeter's typical structure where a portion of the pool is retained as the seller's interest.

2b. Key Structural Features

  • Payment Priority: Sequential (principal flows A1 → A2 → A3 → B → C → D → E)
  • Interest Priority: Sequential (B → C → D → E)
  • Excess Turbo: Excess spread directed sequentially to accelerate principal (A1 → E)
  • Excess Release: Residual to CERTIFICATE
  • Reserve Fund: $11.2M (~1.0% of deal balance)
  • Clean-Up Call: Optional redemption when pool balance falls below 5% of original collateral value
  • WAL Test: 0 of 36 tests failed (RMSE: 0.0348 years; Max Error: 0.0634 years) — PASSED

3. Collateral Analysis

3a. Pool Overview

CharacteristicDetail
Loan Count59,907
Pool Balance$1,670,000,000
Average Loan Size$27,875
WA FICO696
WAC (Tape)9.0%
WA Original Term72 months
Delinquent at Cutoff6.8% (3,510 loans, $113M)

3b. FICO Distribution

FICO Bucket% of Balance
720+47.3%
660–71945.8%
620–6592.3%
<6204.6% (1,915 loans)
No Score3.2% (3,046 loans)

93.1% of the pool is FICO 660+. Sub-620 exposure limited at 4.6%.

3c. Rate Distribution

Rate Bucket% of Balance
0–5%12.4%
5–10%65.3%
10–15%18.6%
>15%3.7%

65.3% of loans carry rates between 5–10%, consistent with near-prime origination.

3d. Term Distribution

Term Bucket% of Loans
≤36 months0.1%
37–60 months8.8%
61–72 months15.3%
>72 months75.8%

3e. New vs. Used

Type% of BalanceAvg Loan
New41.1%$37,500
Used58.9%$23,641

3f. Geographic Concentration

State% of Balance
Texas13.4%
Florida8.6%
California6.9%
Top 3 Combined28.9%

No single state exceeds 15%. Geographic diversification is adequate.

3g. Vehicle Make Concentration

Make% of Balance
Chevrolet16.1%
Ford13.9%
RAM8.9%
Jeep8.1%
Tesla4.6% ($77M)
Other (20+ makes)48.4%

3h. Key Risk Factors

  • Extended-Term Exposure (75.8% >72 months): The dominant risk factor. Longer-term loans carry higher cumulative default probability and slower amortization, increasing loss severity on defaults.
  • Used Vehicle Concentration (58.9%): Used car values have been volatile post-COVID; a normalization in prices increases severity on defaults due to faster depreciation.
  • Delinquency at Cutoff (6.8%): Elevated vs. typical prime pools. Delinquent loans (avg FICO 671, avg balance $32,239) represent a higher-risk segment that could accelerate early-period losses.
  • Tesla EV Exposure (4.6%): EV residual values face structural uncertainty from rapid model refreshes, charging infrastructure changes, and brand-specific risks.
  • No-Score Borrowers (3.2%): 3,046 loans with no FICO carry a 14.0% WA rate — the highest-rate bucket in the pool.

4. Historical Performance (Empirical Benchmarks)

4a. Matched Deal Universe

The empirical engine matched 3 comparable Exeter deals from EDGAR shelf data. However, the matched deals are materially different from the subject pool.

DealVintageWA FICOWA RateSeasoningCNL @ 12MSimilarity
Exeter Deal A60021.5%12 mo3.5%52%
Exeter Deal B59721.8%16 mo3.5%51%
Exeter Deal C59622.2%18 mo4.2%50%
Comp mismatch: The matched deals (avg FICO 598, avg rate 21.8%) are Exeter subprime transactions. EART 20XX-X (FICO 696, rate 9.0%) is a near-prime/prime pool. Loss curves derived from these comps will likely overstate expected losses for the subject pool.

4b. Empirical CNL Milestones (3-Deal Average)

MonthCNLCDR (Ann.)SeverityVPR
12M3.68%11.17%54.43%11.55%
18M7.50%14.37%59.41%8.69%
24M+

Data available through month 19 only. Longer-dated milestones require older vintage deals not yet in the database.

Key Observations

  • CDR ramps from near-zero in months 1–3 to a peak of ~11.96% at month 11, then stabilizes in the 11–14% range through month 18
  • This seasoning pattern is consistent with typical subprime auto defaults — but EART 20XX-X's near-prime collateral should perform materially better
  • Severity rises from ~40% early to ~60% at 18 months, reflecting used vehicle depreciation and recovery costs
  • Analysts should consider applying a 40–60% haircut to CDR for a true base case given the FICO mismatch

5. Loss & Prepay Forecast

5a. Base Scenario Curve Assumptions

CurveAverageRangeSource
CDR (annualized)13.85%0.00%–15.97%19 months empirical + extrapolation
CPR (voluntary)9.53%0.00%–12.23%3 matched deals
Severity62.52%40.33%–100.00%18 months empirical + extrapolation
These base curves are calibrated to Exeter subprime deals (~100 FICO points below this pool). The actual performance of EART 20XX-X's near-prime collateral (FICO 696, rate 9%) would be expected to be materially better. The base scenario represents a conservative/stressed view for this specific pool.

5b. Projected Loss Trajectory

MonthProjected CDRProjected SeverityImplied Monthly Loss
12M~11.17%~54.4%~0.51%
18M~14.37%~59.4%~0.71%
24M+~15.97% (peak)~62.5%~0.83%

Terminal CNL estimated at 18–25% of original balance under base curves given the comp mismatch. Actual losses for this near-prime pool likely 40–60% lower.

6. Breakeven Loss Analysis

6a. Credit Enhancement by Tranche

TrancheBalanceSub Below ($M)CE % of Deal
A1$94.2M$1,040.3M91.7%
A2$241.0M$799.3M70.5%
A3$226.4M$572.9M50.5%
B$118.5M$454.4M40.1%
C$123.0M$331.4M29.2%
D$162.3M$169.1M14.9%
E$103.3M$65.7M5.8%
CERTIFICATE$24.5M$00.0%

Waterfall result (base scenario): All 7 offered tranches (A1 through E) paid in full with no shortfalls under the conservative base assumptions. Given that the base curves are calibrated to subprime comps (~100 FICO points below this pool), the actual cushion for EART 20XX-X is likely significantly wider than implied.

7. Waterfall & Tranche Analytics

7a. Bond Analytics (Base Scenario, Price = 100)

TrancheBalanceCouponWAL (yrs)YieldType
A1$94,200,0003.90%0.145.14%Senior
A2$241,000,0004.08%0.584.44%Senior
A3$226,420,0004.03%1.174.22%Senior
B$118,510,0004.22%1.684.38%Mezzanine
C$123,000,0004.40%2.104.54%Mezzanine
D$162,310,0005.00%2.765.14%Subordinate
E$103,340,0006.69%3.56-2.27%Subordinate
WAL Test: All 36 WAL tests passed (RMSE: 0.0348 years; Max Error: 0.0634 years vs. 0.1-year threshold). Unlike the NQM deal, the auto ABS model calibrates well — the excess turbo mechanism and sequential structure produce predictable paydown patterns.

Yield Observations

  • A1 (WAL 0.14 years): Essentially money-market equivalent. 5.14% yield at par is attractive — implies the excess turbo is accelerating paydown significantly
  • A2/A3 (WAL 0.58–1.17 years): Short-duration senior paper at 4.22–4.44%. Straightforward
  • B/C (WAL 1.68–2.10 years): Modest yield pickup at 4.38–4.54%. Adequate subordination buffer
  • D (WAL 2.76 years): 5.14% yield — reasonable compensation for subordination risk at 14.9% CE
  • E (WAL 3.56 years, yield -2.27% at par): The negative yield confirms this tranche should not be purchased at par. The 6.69% coupon is insufficient to compensate for projected losses under the conservative base. Requires a significant discount (est. 15–25 points) to achieve a positive return

8. Purchase Recommendation

TrancheCouponWALYield @ ParRecommendation
A13.90%0.14y5.14%Monitor — very short WAL; attractive yield but limited duration
A24.08%0.58y4.44%Neutral — fair value for short senior paper
A34.03%1.17y4.22%Neutral — adequate CE; reasonable yield
B4.22%1.68y4.38%Neutral — modest spread over seniors
C4.40%2.10y4.54%Neutral — adequate subordination buffer
D5.00%2.76y5.14%Cautious — limited CE; comp mismatch in loss curves
E6.69%3.56y-2.27%Do Not Buy at Par — requires discount pricing

Preferred entry points: A1–A3 offer the best risk-adjusted profile given the high CE levels (50–92%), short WALs, and the likelihood that actual pool performance will be materially better than the subprime-calibrated base curves. The E tranche would need to be purchased at an estimated 15–25 point discount to achieve a positive risk-adjusted return.

9. Key Risks

9a. Credit Risks

RiskSeverityDetail
Comp MismatchMEDIUMMatched deals are subprime (FICO 598, rate 22%) vs. near-prime pool (FICO 696, rate 9%). Creates conservative bias — protects seniors but limits precision
Delinquency at CutoffMEDIUM6.8% delinquent at cutoff; elevated vs. prime. Avg FICO 671, avg balance $32K — adverse selection risk
Extended-Term ExposureHIGH75.8% of loans >72 months. Higher cumulative default probability and slower equity build-up
Sub-620 FICO TailMEDIUM4.6% of balance with FICO <620 carries disproportionate loss risk at avg loan $40K

9b. Collateral & Structural Risks

RiskDetail
Used Vehicle Depreciation58.9% used vehicles; Manheim index has declined from 2021–2022 peaks. Further normalization increases severity
Tesla / EV Residual4.6% Tesla exposure ($77M) faces structural uncertainty from model refreshes and brand-specific risks
Geographic ConcentrationTX (13.4%) + FL (8.6%) = 22% in two hurricane-exposed states
E Tranche PricingNegative yield at par; structurally first-loss. Should be treated as equity-like
Clean-Up Call (5%)Optional redemption may extend subordinate tranche life if not exercised — extension risk for D and E
Excess Turbo DependencySenior paydown speed depends on excess spread. WAC compression from high-rate prepays slows turbo and extends WALs

9c. Macro Sensitivity

RiskDetail
UnemploymentAuto defaults highly correlated with unemployment. A 1–2% increase could push CDR 20–30% above base
Used Car Price DeclineFurther Manheim normalization increases loss severity on defaulted collateral
Rate EnvironmentAt WAC 9.0% (tape), refinancing incentive is limited. CPR unlikely to accelerate unless rates decline significantly

Report generated from deal model @exeter_auto. All projected values are model outputs based on empirical curves from 3 matched Exeter deals. Empirical data available through month 19 only; 24M+ milestones are extrapolated. Comp mismatch (FICO 598 vs. 696; rate 21.8% vs. 9.0%) should be considered when interpreting loss projections.

IC Report on Subprime Auto ABS — Graam