Dealer risk monitoring, explained.

Watching the dealers who write your car loans, told as a story

The life of one dealer relationship, in five steps

1. The dealer is signedLicense, bond, owners, finances and complaints are checked; a dealer agreement sets credit rules, pay and buyback terms.
2. Deals arrive, one at a timeFunding reviews each contract: stipulations, proof of income, book value, signatures. Most fund within days.
3. First payments come dueThirty to 45 days after funding the first payment is due. Missed first payments are the earliest hard signal.
4. The lender reads the dealer's deals togetherDefaults, values, add-ons, income checks, complaints and pends, by dealer, against its own history and its peers.
5. The lender actsWatch, restrict, charge back, demand repurchase or terminate, with the deals that prove it on file.

Step 4 is the subject of this guide. The dealers, people and figures in the example are fictional.

What lenders miss about dealer risk: habits, add-ons and unchecked trim

Indirect lending to power booking: the dealer risk vocabulary

Thirteen terms are defined in the one-page summary.

How an indirect auto contract travels from the finance office to the lender's books

In shortThe dealer takes the application and shops it to several lenders. Each lender answers with a decision, a buy rate and conditions. The dealer writes the contract, may mark up the rate, sells add-ons, and sends the package. The lender funds it, pays the dealer, and from then on owns the risk of everything the dealer typed.

The OCC's examiners' handbook describes dealers as third parties that "take loan applications from their customers and 'shop' them to banks for approval and funding", and notes that dealers "often maintain relationships with several banks to obtain funding for even the least creditworthy consumers".3 That is the core of dealer risk: the dealer chooses which lender gets which deal, and a dealer that is declined by a careful lender can send the same buyer to a less careful one.

How the dealer is paid, and why it matters for risk

Who buys the contracts

Banks and credit unions buy mostly from franchised and large independent dealers; captive finance companies, owned by carmakers, buy mostly from their brand's dealers; finance companies reach further into subprime. Buy-here-pay-here dealers, which lend to their own buyers, held about 2 percent of a $1.6 trillion market and 5 percent of subprime lending in a Federal Reserve analysis of credit bureau data for the third quarter of 2025.9 The rest of this guide is about lenders that buy contracts from dealers they do not own.

Dealer oversight rules in October 2026: what a court struck, what states added, what examiners still ask

In shortThere is no single federal rule for dealer monitoring. Bank and credit union examiners expect a lender to oversee its dealers as third parties and to report on each one. The FTC's dealer rule was vacated and then withdrawn. The CFPB stepped back from auto cases in 2025. States moved in: California's CARS Act took effect on October 1, 2026, and forty attorneys general announced a settlement in September 2026 with a subprime lender over its loans and its dealers' add-ons.

What an examiner expects you to show about each dealer, and where the evidence lives

In shortExaminers want reports by dealer that management actually uses, written criteria for keeping and dropping dealers, and proof that purchased loans meet the same standards as the lender's own. The NCUA's seven red flags are a ready checklist, and five of them can be measured from data most lenders already hold.

The OCC's sample request letter for a retail lending exam asks, in item 20, for "copies of the management reports used to monitor the quality of applicants and credit performance of loans sourced from each third party used". Item 21 asks, for add-on products, for reports on "product penetration, claims rates (approved and denied), reserve method and balances, and profitability".3 Both ask for reports dealer by dealer; a portfolio summary does not answer them.

The NCUA put it more bluntly in 2010: "Credit unions should regularly test for compliance with the contract terms" by "comparing delinquency, loan losses, and rates of return to previous results and budget levels", and it calls static pool analysis "the most effective method of evaluating the performance of a vendor".18

A static pool follows the loans a dealer sent in one month, or one quarter, through their life. A dealer's recent deals can then be compared with its own earlier deals at the same age, and a mixed, aging book cannot hide a change.

The NCUA's seven red flags, and the data that shows each one

A program that produces scores nobody reads is weak evidence. The request letter's present tense ("the management reports used to monitor") asks for the reports management uses today, so a pack assembled for the exam answers it poorly.

Six dealer performance measures, worked through on one Houston store's 148 contracts

In shortSix measures cover most of what goes wrong at a dealer: early payment default, value against book, add-on penetration, failed income checks, complaints and funding pends. Each means little alone and a lot against two baselines: the dealer's own history and dealers like it. Volume matters too: a dealer that doubles its deals while its measures worsen is the one to read first.

Industry definitions vary. In a 2025 survey of 204 lenders, the charge-back criteria in use were early payment default at 90 days (33 percent of lenders), first payment default (26 percent) and early payment default at 120 days (22 percent).5

One dealer against its peers

In this guide's fictional example, Brazos Bend Auto Plaza, an independent used-car dealer in Houston, wrote 148 contracts for $3,612,400 in the 90 days to October 12, 2026, an average of $24,408. Here are its six measures against the bank's other independent dealers over the same 90 days.

Brazos Bend's early default rates against its peers, worked

Fictional example. The ratio of 6.1 percent to 1.2 percent is about five to one, the scale of the "as much as 500 percent" that one fraud analytics firm reports for systematic dealer risk.2

Volume tells you which dealer to read first

A dealer whose measures worsen while its volume holds steady is a problem of quality. A dealer whose measures worsen while its volume jumps is a problem of time: every month of delay adds deals. At September's pace, each 30 days of waiting adds about 61 contracts and, at the average of $24,408, about $1.5 million of new loans from a dealer whose first payments are already being missed.

Why the slowest measure sets the pace

Illustrative: a first payment due about 30 to 45 days after funding and counted as missed some weeks later. The funding-time measures are the early warning; defaults confirm.

Five ways a dealer inflates a deal, and the trace each leaves across its contracts

In shortMost dealer fraud changes either what the lender is told about the buyer or what it is told about the car. Lenders surveyed in 2025 put 60 percent of dealer-facilitated fraud on consumer data and 40 percent on vehicle data. Each scheme is hard to see in one file and easier to see in a dealer's files together, because dealers repeat what works.

The NCUA lists "dealer-created down payments through dealer incentives, inflated or fraudulent trade-in or purchase price" among its red flags.18 The FTC alleged in 2018 that an Arizona and New Mexico dealer group inflated buyers' income and down payments on applications sent to lenders, and that those buyers defaulted at a higher rate.27 In March 2026 a Tampa dealer was sentenced to 54 months for sending lenders falsified applications from straw purchasers for cars that did not exist, at least six vehicles and at least $372,000.28

The auto finance fraud lenders report most

A fraud analytics firm's 2025 report put income and employment misrepresentation at 43 percent, $3.9 billion, of $9.2 billion of estimated auto lending fraud exposure, and found that up to 70 percent of early payment defaults carried evidence of fraud on the application.2 Its March 2026 brief counted more than 400 dealerships operating in "extreme risk conditions" and reported that one in five paystubs submitted with auto loans was fake.29

The firm's 2026 report puts exposure at $10.4 billion for 2025, up 13 percent in a year, with income and employment fraud at 45 percent of it30, and in October 2026 the National Insurance Crime Bureau reported the same $10.4 billion.31 These are vendor and industry estimates without public datasets; they show direction more than precise size.

The vehicle side is checked least

Dealer markup and fair lending, from the 2013 consent orders to the 2026 Regulation B rule

In shortWhen a dealer can charge any rate between the buy rate and the lender's cap, buyers with the same credit can pay different rates. Studies found minority borrowers paid more. The CFPB's 2013 to 2016 orders paid minority borrowers back and, from 2015, cut markup caps; Congress voided the CFPB's guidance in 2018. In 2026 the CFPB said ECOA does not reach disparate impact, and the OCC and FDIC no longer examine for it. Intentional or unequal treatment is still illegal, and California caps the dealer's share by law.

What studies show about dealer markup and minority borrowers

How the caps came down

The CFPB's March 2013 bulletin told indirect lenders that discretionary markup could expose them under ECOA. In December 2013 the CFPB and the Justice Department ordered Ally to pay $80 million to more than 235,000 minority borrowers and an $18 million penalty.36 Honda (2015) followed with the caps above.35 In May 2018 Congress disapproved the bulletin under the Congressional Review Act, and the CFPB's page now says it "has no force or effect".37,38

Disparate impact, disparate treatment and state markup caps in 2026

Why the lender ends up paying for the add-ons its dealers pack into contracts

In shortThe products are sold at the dealer but financed by the lender, which collects interest on them and, in many states, owes refunds when a loan ends early. Supervisors and attorneys general have treated packed or useless add-ons as the lender's problem as well as the dealer's, and the 2024 to 2026 cases show the price.

In the dealer-group cases the details repeat. The Connecticut dealer allegedly put total loss protection on about 90 percent of sales and told buyers some products were mandatory.22 The Illinois complaint alleged the group "regularly failed to actually install or apply the add-on products" it charged for.21 A lender buying those contracts sees the trace as a penetration rate far above its other dealers, the signal in section 05.

Auto credit in 2026: record subprime delinquency, rising fraud estimates and more dealer buybacks

In shortAmericans owed $1.71 trillion on auto loans in mid-2026. Subprime delinquency in securitized loans hit a record early in the year. Lenders lean more on dealer buybacks: about two-thirds require dealers to repurchase fraud or early default loans, but recover only a share of what goes wrong.

How lenders recover from dealers

The same survey found lenders estimating that only 45 percent of fraud-related defaults are correctly flagged as fraud inside their own organizations, and a median of 30 percent of fraudulent indirect loans charged back to dealers.5

A March 2026 survey of more than 35 finance companies by a fraud analytics firm found 60 percent requiring dealer buybacks of fraud and early default loans and 15 percent never requiring them.29 A buyback demand is only as good as the evidence behind it and the dealer's ability to pay; section 10 covers both.

From signing a dealer to ending the relationship: each decision and the evidence it needs

In shortDecide the exits before the entry: the agreement should say what triggers a watch, a restriction, a charge-back, a repurchase and a termination. Then use the least action that stops the pattern, backed by sampled deals proven at the source. Restricting new deals does nothing for contracts already funded; recovery on those depends on the agreement's repurchase and reserve clauses.

Onboarding: what to check before the first deal

The escalation ladder

The NCUA expects "clear policies governing the selection of vendors, as well as conditions requiring removal of vendors" and "an exit strategy in the event of an unexpected or unplanned increase in the program risk profile".18 A termination also moves the dealer's marginal deals to some other lender, which is one more reason lenders document the decision carefully.

Scoring dealers by machine every night, and where that falls under the 2026 model and third-party guidance

In shortSoftware can join funding, fraud, servicing and complaint data by dealer every night, compare each dealer with its history and peers, and look for details that repeat across its deals. A statistical dealer score is a model under the April 2026 guidance; rules and arithmetic are not. The lender keeps responsibility for every third party it relies on, dealer or vendor, and a person should decide every action against a dealer.

The work suits software because the hard part is joining. The signals sit with different teams: funding sees pends, fraud sees failed income checks, collections sees missed payments, compliance sees complaints. Reading them together by dealer, every night, is arithmetic at scale. Looking across a dealer's deals for repeated details (the same down payment, employer phone, stated income band or reference) is pattern-matching that no person reviewing one file at a time can do.

What nightly dealer scoring does not prove

• A pattern is a reason to look. Proof is a deal checked at the source, with a document on file, and only that supports a repurchase demand.
• Defaults still need payments to come due, so earlier warning has to come from the funding-time signals.
• Recoveries from dealers should be matched to what borrowers paid; CFPB examiners found servicers keeping interest on options they had already recovered for.32
• The relationship manager, the credit head and, for a termination, the head of auto finance still decide.

Five teams, one dealer and a quarterly calendar: how dealer reviews run today

In shortFive teams each hold one piece of a dealer's story, and credit risk joins them in a monthly or quarterly dealer review built on loss and delinquency reports. Because early defaults need time to show and the review has a calendar, a dealer whose pattern changes in late summer often reaches the committee in the new year.

Who watches a dealer: risk analyst, fraud, relationship manager and compliance

Five teams' data meet once in a quarterly dealer review

Why a quarterly dealer review spends most of its time waiting

None of those days is anyone's fault. Early defaults take time to appear, analysts join reports by hand, and the committee meets on a calendar. The cost is volume: a dealer that writes 60 contracts a month keeps writing them while the review waits.

Static-pool spreadsheets, fraud scores or nightly dealer monitoring

Follow Brazos Bend Auto Plaza from a changed pattern to a co-signed restriction

In shortThis is the example from the top of the page, inside the product. Seven agents read every dealer's deals overnight, score each dealer and look across each dealer's deals for repeats. Dana Okafor, the dealer risk analyst, opens the alert, drills from the score to the 17 contracts behind the pattern, reviews a pack of twelve deals checked at the source, and signs a restriction. Russell Teague, Head of Auto Credit Risk, co-signs.

1

The dealer network, scored before the analyst arrives

At 5:12 am on Monday, October 12, the scoring agent has scored all 412 dealers. Dana opens a map: each dealer is a dot sized by contracts funded in 90 days and colored by risk. One red dot in Houston stands out, Brazos Bend Auto Plaza at 91, with an alert.

The side panel shows the score crossing the 80 line, first payments missed at 6.1 percent against 1.2 percent for peers, add-ons on 87 percent of deals against 38, and three patterns found across its deals.

2

Seventeen deals with exactly $2,000 down

One click on the first pattern opens every Brazos Bend contract as a dot placed by its cash down. Seventeen stack in a red tower at exactly $2,000. At other dealers no exact figure repeats more than three times.

The seventeen share two employers, one employer phone line and the job title "route driver", and state incomes from $5,600 to $6,125 a month. $439,720 is financed, four have missed the first payment, and two more wait to fund.

3

A score of 91, explained signal by signal

Dana runs "Explain this score". Four agents re-read the 148 contracts, follow every first payment and join complaints and fraud referrals. The score breaks down as employment and income 24 of 25 points, early payment default 23 of 25, value against book 17 of 20, add-ons 14 of 15, complaints 8 of 10 and pends 5 of 5.

Beside it, volume: about 24 contracts a month until June, then 31, 46 and 61. The alert line sits at 80; the score crossed it on Friday, October 2.

4

Twelve deals checked at the source; four proven false

The review pack agent samples twelve deals across the signals and checks each where it can be proven: the employer called at an independently found number and searched on the state register, the VIN decoded against the build sheet, the product menu read. Four paystubs are falsified, five deals carry options the car does not have, three product menus are unsigned.

The four contracts with falsified employment, $97,640 in all, are eligible for repurchase under the sample dealer agreement. The pack also estimates about $812,000 of net loss over 12 months on the dealer's 263 open contracts, against $214,000 for a dealer its size.

5

Restricted, with two signatures

The action agent recommends the least action that stops the pattern: income verified by the bank on every new deal, the advance capped at 115 percent of book on VIN-decoded options only, service contract and GAP each on a signed product menu, the repurchase demand, and a review in 30 days.

Dana signs with the meaning "I have reviewed the pack and the deals behind it". A restriction needs a second signature, so it goes to Russell Teague, who co-signs. Only then does the notice go to the dealer. The alert came about five and a half weeks after the pattern changed on August 24; the signed restriction came ten days after the alert, seven weeks after the change.

The solution's design targets are about five weeks from a dealer's pattern change to a signed action, and every active dealer scored every night on six signals. They are targets, to be measured on your own dealers. In this example the alert fired at five and a half weeks and the signature followed at seven.

Twelve more screens, from a six-month replay to the bank's own dealer policy

In shortOne dealer is the story. These screens show the rest: other ways to color the network, the 26-week replay, every dealer ranked, the alerts queue, the deals behind a pattern and a single deal, the employers shared across dealers, the paystub evidence, the decision and the conditions, the action log and its results, the linked dealer, and the settings that hold the bank's policy.

Select any screen to open it larger. The bank, dealers, employers, people and figures are fictional; the dealer agreement and its clause numbers are sample text written for these screens.

Seven agents read, score and draft. An analyst decides, and a credit head co-signs.

In shortEach agent has one job and limits it does not cross. None of them restricts or terminates a dealer, contacts a dealer or a borrower, or changes a funding decision. A pattern is never called fraud: only a deal checked at the source, with a document on file, counts as proven. The analyst signs a watch; a restriction needs the Head of Auto Credit Risk; a termination also needs the Head of Auto Finance.

Seven agents, from the nightly deal reader to the action recommender

Who signs a dealer watch, co-signs a restriction and co-signs a termination

Signatures record the name, the action, its meaning and the time. Agent steps and human decisions are kept apart on the audit trail; settings changes are versioned; borrower details appear only inside a review pack; dealer compliance has read access for fair lending review.

Dana Okafor's dealer risk team and the two heads who co-sign

Funded contracts and payments in, ranked dealers and signed actions out, with your dealer policy in between

In shortIt reads data your teams already hold, from the loan system, the funding desk, fraud checks, servicing, complaints and book values, joins it by dealer every night, and leaves behind a ranked, explained list of dealers, alerts when a pattern changes, review packs with proven sample deals, and a log of signed actions and their results. Your systems of record stay where they are.

Origination, funding desk, complaints and vehicle data, joined by dealer

For each, check whether it has an API, a file feed or only screens, and who owns it.

Scores explained by deals, patterns as reasons to look, people signing

What a problem dealer costs you for every month it keeps sending deals

In shortThe cost of a bad dealer is the deals you keep buying while you wait to see it. Published figures frame the size of the problem; your own numbers decide the case. Start with how long your last three dealer actions took from the first sign to a signature, and what the dealer sent you in between.

Deals bought while waiting, excess losses and uncollected buybacks

A fraud ring moving on, a manager's credibility, borrowers in packed deals

The return on acting against a problem dealer three months sooner

In shortThe return comes mostly from deals you stop buying sooner from the few dealers that go bad, then from buybacks you can prove and analyst hours you get back. Put in your own numbers. Every default below is illustrative and marked so.

Six gains from nightly dealer scoring, from fewer bad deals to provable buybacks

Losses avoided by acting on problem dealers months sooner

A lender buying from a few hundred independent dealers with several problem stores a year sits on the middle bar or above. A lender buying mostly from franchised dealers with low early default rates may find the hours and the exam evidence matter more than the losses, and should say so before buying anything.

Nine questions that show whether your team is ready to monitor dealers nightly

In shortPick the answer closest to your team today. Eight questions score 0 to 2; the last is for reflection. The meter adds up your score as you go, and each question says why it matters and what a strong answer sounds like.

When a lender is ready to score its dealers every night

What to ask a dealer monitoring or dealer scoring vendor: seventeen questions in six groups

In shortAsk every vendor the same questions, us included. The answers show whether the tool reads your own data or only outside data, whether a score can be traced to deals, whether a pattern is separated from proof, and how you would test it on your own dealers.

We found no published independent test of any dealer monitoring vendor's accuracy, as of October 2026. A back-test on dealers you have already acted on is the fairest comparison, and the same test measures your current review.

A twelve-week back-test and parallel run you can take to the dealer committee

In shortMeasure your last actions, back-test on a year of your own contracts to see when each problem dealer would have surfaced, then score nightly beside your current review. Decide with three numbers: weeks gained, alerts that led nowhere, and analyst hours. None of this commits you to a vendor.

1. Time your last three dealer actions, from the first rise in early defaults to the signature, and count the contracts bought in between. That is your baseline.
2. Pull twelve months of contracts with pends, checks, payments, complaints and dealer numbers, and an approval path to share them.
3. Back-test: score the year week by week and record when each dealer you acted on would have crossed your alert line.
4. Run in parallel for five to six weeks on live data, with your analysts deciding every action as today.
5. Decide with three numbers: weeks gained per problem dealer, alerts that led nowhere, and analyst hours per review.

Weeks gained, alerts that led nowhere, missed dealers and proof rate

Questions lenders ask about dealer risk monitoring, answered

In shortThe questions auto lenders and buyers search for most, each answered on its own.