The Consumer Credit Waterfall: Auto Loans to Pension Funds

Nine levels of credit intermediation. From AI‑originated sub‑subprime loans through ABS syndication, CLO packaging, and insurance wrapping to pension fund ETFs. Risk is not reduced through securitisation — it is redistributed and obscured.

As of2026‑04‑04 Levels9‑tier intermediation chain DataFRED · SEC · CFPB · IEA TypeSystem Map
Elevated Risk Public‑source
Total consumer credit outstanding (FRED TOTALNS, Jan 2026)
$5.10tn
Confirmed
Revolving credit at near‑record highs (FRED REVOLSL, Jan 2026)
$1.33tn
Confirmed
Credit card delinquency rate (FRED TERMCBCCALLNS, Nov 2025)
20.97%
Confirmed
ABS issuance 2025 (+22.8% YoY)
$358bn
Confirmed

Executive Summary

Provenance badges

A single consumer loan flows through nine levels of intermediation before reaching a pension fund ETF. At each level, fees are extracted, complexity is added, and opacity increases — but the underlying credit risk is never eliminated. The consumer who cannot afford a $767/month car payment is the same consumer whose debt sits in a pension fund ETF nine levels later. Modeled

Total consumer credit outstanding reached $5.10 trillion in January 2026, with revolving credit at $1.33 trillion near record highs. Credit card delinquency stands at 20.97% — historically elevated. ABS issuance surged 22.8% to $358 billion in 2025, meaning the volume of newly securitised consumer debt entering the chain is accelerating. Confirmed

All FRED data pre‑dates the Iran–Hormuz conflict (February 2026 onwards). Q2–Q3 2026 FRED releases will reflect the oil shock impact. Current readings are the baseline before the catalyst, not the impact. Confirmed

The Crassus Principle
The system's exterior (AAA ratings, ETF performance, aggregate delinquency metrics) remains stable while the interior (subprime cohort deterioration, negative equity records, AI lending defaults, PE withdrawal queues) deteriorates. The gap between exterior and interior widens before it closes. When it closes, it closes violently — as it did in September 2008. Modeled

Nine‑Level Waterfall Topology

Derived

Network Map — Consumer Origination to Pension Fund

Full chain
Level 1AI OriginationPagaya · Upstart · LendingClub
Level 2BNPL LayeringAffirm · Klarna · Afterpay
Level 3Auto LoansALLY · COF Auto · Santander
Level 4Credit CardsCOF · SYF · C
Level 5Bank WarehouseWFC #1 · JPM · BAC
Level 6PE / CLOBX · KKR · ARES
Level 7BDC DistributionARCC · FSK (junk)
Level 8Insurance WrapAGO · PE‑backed insurers
Level 9Pension Fund ETFsJAAA · JABS · Target‑Date
Signal
Origination risk is repriced at every level — but never eliminated. A $767/month auto loan originated by an AI algorithm, layered with BNPL, packaged by Wells Fargo, tranched by Blackstone, distributed through ARCC, and guaranteed by a PE‑backed insurer ultimately sits in a pension fund ETF held by a teacher in Manchester saving for retirement. Modeled

FRED Data — Consumer Credit Baseline

Confirmed

Table 1 — Confirmed FRED Series (Crassus Terminal)

FRED
SeriesNameLatest ValueDateTrendBadge
TOTALNSTotal Consumer Credit Outstanding$5,099.49BJan 2026Rising +$57.2B/5moConfirmed
REVOLSLRevolving Credit Outstanding$1,328.99BJan 2026Rising +$18.4B/5moConfirmed
NONREVSLNon‑Revolving Credit Outstanding$3,785.69BJan 2026Rising +$24.8B/5moConfirmed
TERMCBCCALLNSCredit Card Delinquency Rate20.97%Nov 2025Falling −0.50/5qConfirmed
DRALACBSAll Consumer Loan Delinquency Rate1.48%Oct 2025Falling −0.05/5qConfirmed
DRSFRMACBSSingle Family Mortgage Delinquency1.78%Oct 2025Rising +0.01/5qConfirmed
DRCCLACBSCharge‑Off Rate Credit Card Loans2.94%Oct 2025Falling −0.14/5qConfirmed
Critical timing note
All FRED readings pre‑date the Iran–Hormuz conflict (February 2026 onwards). The apparent decline in credit card delinquency from 21.76% is driven by charge‑offs moving the worst accounts from "delinquent" to "loss" — not by borrower recovery. Q2–Q3 2026 FRED releases will reflect the post‑war deterioration. Current readings are the baseline before the catalyst. Confirmed

Level 1 — Sub‑Subprime Origination (The AI Lending Layer)

ConfirmedModeled

Pagaya · Upstart · LendingClub

Level 1

AI lending expands the borrower pool without improving creditworthiness. Pagaya originated $10.5 billion in 2025, up 9% YoY. AI‑originated default rates run 5–10% vs traditional bank rates of 2–3%. ML prediction accuracy (AUC) reaches 0.85–0.95+ per NBER studies — accurate enough to be profitable in normal conditions. Confirmed

Under stress ($100+ oil), the marginal borrowers approved by algorithms are the first to default because they were marginal from origination. Algorithmic bias concerns persist: ML models use proxy variables that can penalise minorities. Modeled

Crassus Principle Applied
Beautiful exterior: "AI‑driven financial inclusion, democratising credit access." Rotten interior: lending to consumers who will default at 5–10% rates. Modeled

Level 2 — BNPL Layering (The Hidden Debt Layer)

Confirmed

Affirm · Klarna · Afterpay/Block

Level 2

BNPL creates a hidden debt layer invisible to traditional credit scoring. Global BNPL market projected at $509.2 billion for 2026, with US Pay‑in‑4 originations at $63–70 billion for 2025. Confirmed

69% of BNPL borrowers carry revolving credit card debt. BNPL users have 8% higher probability of first‑year mortgage default. BNPL does not consistently appear on standard credit reports — a consumer appears creditworthy while carrying thousands in BNPL obligations. Confirmed

Cross‑reference: Crassus BNPL Systemic Risk Report Related

Level 3 — Auto Loans (The Visible Crack)

Confirmed

ALLY Financial · Capital One Auto · Santander Consumer

Level 3
MetricValueBadge
Average new car payment$767–772/monthConfirmed
Average used car payment$537–570/monthConfirmed
New car buyers at $1,000+/month1 in 6Confirmed
New car trade‑ins with negative equity28.1%Confirmed
Average negative equity rolled into new loan$6,905 (record)Confirmed
Subprime auto 60+ day delinquencies6.43%Confirmed
Non‑revolving credit outstanding (FRED NONREVSL)$3.79tnConfirmed

Auto loans are the visible crack because they are fixed obligations. Unlike credit cards (can reduce spending) or BNPL (can stop buying), auto loans cannot be reduced. When oil rises above $100, the cost of owning the car (fuel + insurance) stacks on top of financing it. 28.1% negative equity means borrowers cannot sell their way out — they owe more than the car is worth. Modeled

Level 4 — Credit Cards (The Pressure Valve)

Confirmed

Capital One (post‑Discover) · Synchrony · Citigroup

Level 4
MetricValueBadge
Capital One domestic card net charge‑off rate (Q4 2025)4.93%Confirmed
Capital One 30+ day delinquency3.99%Confirmed
Synchrony delinquencies4.2–4.9%Confirmed
Synchrony charge‑offs5.1–6.8%Confirmed
Average credit card APR22.8–25%Confirmed
Subprime utilisation70%+Confirmed
FRED credit card delinquency (TERMCBCCALLNS)20.97%Confirmed
FRED charge‑off rate (DRCCLACBS)2.94%Confirmed

Credit cards are the pressure valve absorbing overflow from auto and BNPL stress. When fuel and car payments squeeze the consumer, credit cards fund groceries, utilities, and essentials. The 20.97% delinquency rate is historically elevated. The apparent decline from 21.76% is driven by charge‑offs moving the worst accounts from "delinquent" to "loss" — not by borrower recovery. Capital One absorbed Discover's combined card portfolio in May 2025, doubling exposure at cycle peak. Confirmed

Cross‑reference: Crassus Consumer Credit Report Related

Level 5 — Bank Warehouse & Syndication (The Packaging Layer)

Confirmed

Wells Fargo (#1 ABS syndicator) · JPMorgan · Bank of America

Level 5

ABS issuance reached $358 billion in 2025, up 22.8% YoY. Aggregate ABS spreads sit at approximately 48 bps — tight but widening in subprime tranches. Wells Fargo dominated 2024–2025 ABS league tables as the #1 syndicator. Confirmed

Wells Fargo has touched more consumer ABS than any other institution. The bank earns fees on packaging and distribution but retains reputational and contractual exposure through representations and warranties. When underlying loans deteriorate, buyers examine syndication documentation for misrepresentation — exactly as in 2008. The FIRREA precedent: Bank of America and JPMorgan paid tens of billions in fines for 2008‑era syndication practices. Confirmed

Cross‑reference: Crassus Consumer ABS Report Related

Level 6 — PE / CLO Packaging (The Leverage Layer)

Confirmed

Blackstone · KKR · ARES Management

Level 6
MetricValueBadge
US CLO new issuance 2025$209bnConfirmed
CLO refinancing/resets 2025$337bnConfirmed
Total CLO activity 2025$546bnDerived
CLO equity median historical IRR~12%Confirmed
CLO BB tranche historical default0.04%Confirmed
Blackstone BCRED fund size$83bnConfirmed

$546 billion in total CLO activity. Leverage amplifies returns in good times and losses in bad times. When consumer defaults rise, equity tranches absorb first losses through suspended distributions. PE firms holding these tranches face NAV declines, fundraising pressure, withdrawal requests, and forced asset sales. KKR's FSK has been downgraded to junk by Moody's. Confirmed

Level 7 — BDC Distribution (The Canary Layer)

Confirmed

ARCC (Ares Capital) · FSK (FS KKR Capital)

Level 7

BDCs hold the riskiest credit with the least buffer. They must distribute 90%+ of income — no retained earnings cushion. FSK non‑accruals reached 3.4% at fair value and 5.5% at amortised cost. Sector dividend coverage has narrowed to 1.05–1.15×. In the 2020 crisis, BDC price‑to‑book ratios fell to 0.63×. Confirmed

BDCs are the canary — they hold the lowest tranches, report most frequently, and have the least buffer to absorb losses. Non‑traded BDC withdrawal queues replicate the BCRED dynamic: investors want out but liquidity gates prevent exit. Capital is trapped in deteriorating assets — exactly the dynamic that preceded 2008 money market fund breaks. Modeled

Level 8 — Insurance Wrapping (The Guarantee Layer)

ConfirmedModeled

Assured Guaranty · Apollo‑backed · KKR‑backed insurers

Level 8

Critical structural change from 2008: In 2008, monolines (AMBAC, FGIC) held the guarantee risk. In 2026, monolines have retreated to municipals. AGO structured finance exposure sits at $5.1 billion (~7.8% of $65.3B total). The consumer ABS insurance exposure has migrated to life insurers and PE‑backed insurers. Confirmed

US life insurers hold 14.5% of bond portfolios in ABS/structured securities. PE‑backed insurers carry up to 31% structured exposure — skewed toward AAA/AA tranches. This is more dangerous than 2008 because life insurers hold policyholder money, not just capital markets positions. If structured tranches deteriorate, life insurance policyholders are indirectly exposed. Confirmed

Cross‑reference: Crassus Insurance & Reinsurance Report Related

Level 9 — Pension Fund ETFs (The Final Destination)

Confirmed

JAAA · JABS · Target‑Date Funds · Public Pension Allocations

Level 9

Public pension funds allocate 2–5% of $30+ trillion to structured credit. Underfunded pensions drift to BBB/BB tranches chasing 7%+ return targets. CLO ETFs (JAAA, JABS) are booming, holding mostly AAA tranches. Target‑Date Funds are being pushed to include structured credit sleeves in 2026 — the Level 9 exposure is actively expanding. Confirmed

The waterfall ends here. The teacher and the car buyer have never met. They are connected through nine levels of intermediation. When the car buyer defaults, the teacher's pension suffers — eventually. The time lag between origination default and pension impact is 12–18 months, creating false confidence at each level. Modeled

The Stress Test — $100+ Oil (The Catalyst)

ConfirmedModeled

Sequential Cascade Under Oil Stress

Active
LevelUnder $100+ Oil StressBadge
L1 — AI OriginationAlgorithms continue approving — models don't price oil shocks in real‑timeModeled
L2 — BNPLBNPL payments compete with petrol costs — BNPL loses firstModeled
L3 — AutoBorrowers cannot afford the car AND the fuel — 28.1% already underwaterModeled
L4 — Credit CardsAbsorb overflow — utilisation surges toward limitsModeled
L5 — ABSDefault rates rise — collateral performance deterioratesModeled
L6 — CLO/PEEquity tranche distributions suspended — PE fund NAVs declineModeled
L7 — BDCNon‑accruals accelerate — dividend coverage breachedModeled
L8 — InsuranceGuarantees tested — PE‑backed insurer portfolios stressedModeled
L9 — PensionsETF NAVs decline — retirement savings erodedModeled

$100+ oil is not an energy story. It is a credit story. The oil price is the input. The waterfall is the transmission mechanism. The pension fund is the output. Nine levels of intermediation add fees, complexity, and opacity — but they do not add resilience. Modeled

Catalyst Confirmation

Confirmed
  • Oil at $111+ (April 2026) Confirmed
  • IEA: 12 million barrels/day offline Confirmed
  • IEA: April disruptions twice as bad as March Confirmed
  • Malaysia restructuring civil service to save fuel Confirmed
  • Lufthansa CEO preparing for jet fuel shortages Confirmed

Prediction Markets

Polymarket
  • $110 oil April: 100% (settled) Confirmed
  • $120 oil April: 67% Confirmed
  • $130 oil April: 37% Confirmed
  • Ceasefire April 15: 14% (86% NO) Confirmed
  • Ceasefire May 31: 46% (54% NO) Confirmed

Historical Parallel — 2006–2008

Modeled

Sequential Failure Pattern

24‑month timeline

The waterfall does not break at one level simultaneously. It breaks sequentially from bottom (sub‑subprime origination) to top (pension funds). The time lag between levels creates false confidence at each stage. The consensus at each stage: "it won't spread to the next level." Modeled

DateEventConsensus
2006Subprime mortgage origination deteriorates"Contained"
Early 2007CDO equity tranches fail"Isolated"
Mid 2007Bear Stearns hedge funds collapse"Specific to structured credit"
Early 2008Monoline downgrades begin"Manageable"
Sep 2008AIG fails, Lehman collapses, pension funds crash"Systemic"
Current Position Assessment
FRED data, options chain positioning, corporate earnings, and prediction markets collectively suggest we are in the early stages of origination stress (Levels 1–4), with institutional positioning indicating expectation of progression to Levels 5–7 through the April–May earnings window. Modeled