The Intermediation Chain: How Many Parties Touch a Consumer Dollar

A consumer takes a $4,000 loan from a brand they've heard of. That brand may not have underwritten it, does not hold it, will not service it, and will not collect on it if it goes wrong. Six firms are involved before the money moves and a seventh appears if it doesn't get repaid. Each takes a margin, and no single margin looks unreasonable. The chains exist for real reasons — different activities need different capital, capabilities, and permissions, and specialization genuinely lowers costs. But length has a cost the pricing analysis misses entirely: every handoff is a point where responsibility can be declined and where the information needed to fix a problem stops travelling.

Tracing two transactions

A $100 card purchase. The participants, and roughly what each does:

Party	Function	Takes
Issuer	Extends credit, bears fraud and credit risk	The largest share of acceptance cost
Card network	Rules, routing, settlement	Assessments and switch fees
Acquirer	Holds the merchant relationship, bears merchant risk	A markup
Processor	Authorization and settlement mechanics	Per-transaction and monthly fees
Gateway	Connects the merchant's systems	A monthly and per-transaction fee
Sales organization	Sold and supports the account	A share of the markup

The merchant sees one effective rate — the figure in our effective rate guide — and six firms sit behind it.

A $4,000 consumer loan. Frequently more parties, and they change over the life of the loan:

• The consumer-facing brand — marketing, application, and the customer relationship.
• A sponsor bank, which may originate the loan and hold the lending authority.
• A capital provider, which funds the balance and bears the credit risk.
• A servicer, which takes payments, sends statements, and makes workout decisions.
• A sub-servicer or specialty servicer if the loan deteriorates.
• A collection agency if it defaults.
• A debt buyer if it's sold.
• Data furnishers and bureaus throughout, per our furnisher guide.

Two features worth naming immediately. The consumer's counterparty changes over time without the consumer choosing it. And the brand on the product may be involved in only the first step — the arrangement our servicing analysis describes, generalized across the whole lifecycle.

The customer chose a brand. The brand's involvement ended at origination, and everything that happens afterward is decided by firms the customer never selected and cannot reach.

Why chains form

The case for intermediation is genuine and worth stating properly, because an analysis that treats every layer as extraction will get the remedies wrong.

Different activities need different things. Originating requires marketing and underwriting. Holding requires capital and balance sheet capacity. Servicing requires operational scale and systems. Collecting requires yet another operation with its own compliance burden. These are four different businesses, and being good at one implies nothing about the others.

Specialization lowers unit cost. A specialist servicer handling millions of accounts spreads fixed costs across them — the arithmetic in our selection analysis operating in the industry's favour. A small originator servicing its own book pays much more per account.

Capital is cheaper when separated from operations. A firm good at acquiring customers may be a poor place to hold credit risk, and a capital provider may want exposure without operational burden.

Permissions are hard to obtain. Lending authority, money transmission, and collection licensing each require regulatory approval that takes years. Partnering with a firm that holds a permission is faster than acquiring one, and for many products it's the only route.

So the chain isn't a failure. It's what a market looks like when the activities genuinely decompose and the pieces have different optimal owners. Our argument is about what length costs, not about whether specialization should exist.

Why they're longest where costs are highest

The empirical regularity that makes the analysis matter, and the mechanism behind it.

Chains are longest in the products serving the most constrained consumers. A prime credit card from a large bank is close to vertically integrated — the bank originates, holds, services, and collects. A small-dollar loan to a thin-file borrower may involve a brand, a sponsor bank, a capital partner, a servicer, and eventually a buyer.

Why the correlation runs that way:

• The permissions are harder. Higher-rate lending requires authority that fewer firms hold, so the sponsor arrangement is necessary rather than chosen.
• The capital is specialized. Funding higher-loss assets requires investors with a specific appetite, who are rarely the same firms doing origination.
• Servicing is more intensive, which favours specialists.
• Defaults are more frequent, so the collection and sale links activate more often.
• The economics are thinner per account, so no single firm can afford to build every capability.

The consequence is one that compounds every other disadvantage this desk has documented. The consumers paying the most, with the least margin for error, transact through the most intermediated structures — which means they face the longest chain at exactly the point where problems are most likely and their capacity to pursue a resolution is lowest. That's the same compounding as our poverty premium analysis, appearing in organizational structure rather than in price.

The accountability problem

The cost that pricing analysis misses entirely, and the one we'd argue is largest.

Every link is a place where a problem can be routed to someone else. The customer with a servicing error contacts the brand, which directs them to the servicer, which says the data came from the originator, which says the account was sold. Each statement may be accurate.

What degrades along a chain:

• Information. Records get transferred in summary form. The original agreement, the payment history, the notes explaining an anomaly — each handoff loses some, which is why the dispute investigations in our dispute guide so frequently fail at transferred accounts.
• Authority. The party the customer can reach frequently cannot decide anything.
• Incentive to resolve. A firm paid per action has no stake in a resolution that ends the account.
• Institutional memory. The reason an account looks unusual is known to a firm no longer involved.

The specific failure worth naming, because it's the one that costs consumers most: the party with the customer relationship is frequently not the party making decisions about them. A consumer chose the brand. The brand's involvement ended at origination. Every decision that matters afterward — payment application, credit reporting, hardship treatment, collection intensity — belongs to firms the consumer never selected.

It also creates a supervisory gap. Different participants answer to different regulators, and some answer to none directly — a servicer's conduct may be governed through its contract with the owner rather than by direct supervision. That means a problem can be simultaneously real, harmful, and outside anyone's clear remit until an enforcement action establishes otherwise.

Misaligned incentives along the chain

Each participant optimizes its own economics, and the sum is not what a single firm would choose.

Originators paid on volume have weak incentive to care about performance beyond whatever retention or repurchase obligation they carry. The quality question belongs to whoever holds the risk, and they weren't in the room at origination.

Servicers paid per action or per account face the problem our forbearance analysis identifies as the sharpest version of that report's argument: a fee-based servicer has no stake in recovery, so the patient workout that maximizes the owner's return is unpaid work for the party who decides whether to do it. The economics favour forbearance and the decision-maker doesn't share in them.

Collection agencies on contingency maximize collections per unit effort, which the frontier logic in our intensity analysis shows can diverge from what the creditor would choose — and the reputational cost of aggressive treatment lands on the brand rather than the agency.

Debt buyers have no relationship to preserve at all, which changes the calculus entirely.

The pattern: as the chain lengthens, the share of participants with a long-term interest in the customer falls toward zero. By the final links, nobody involved will ever transact with this person again, and every incentive that a continuing relationship would have supplied has been designed out.

Chains and regulatory boundaries

A connection worth drawing explicitly, because it links this report to the definitional analysis.

Chains form partly to place activities under favourable regulatory treatment. A structure where one firm holds a charter, another holds the economics, and a third holds the customer relationship distributes activities across regulatory categories — and which firm is deemed to be doing what determines which rules apply.

That's the same mechanism our definitional analysis describes, operating on organizational form rather than product form. Both are consequences of regulating by category: if rules attach to what a thing is called or who is deemed to be doing it, then structuring around the labels is available and will be used.

Two things follow.

Some chains are longer than the economics alone would produce. Where a structure exists to distribute regulatory exposure rather than to allocate activities efficiently, the extra links are cost without benefit.

Distinguishing the two is genuinely hard. A sponsor bank arrangement can be a legitimate allocation of a permission to the firm that holds it, or a structure whose purpose is the treatment it produces — and the same arrangement can look like either depending on facts that aren't visible from outside. We'd resist the assumption that length implies avoidance, and equally resist the assumption that it doesn't.

The strongest objections

"Longer chains can be cheaper, so the framing is wrong." Conceded, and it's why this report avoids claiming that length raises price. Specialization can reduce each function's cost by more than the margins add — a specialist servicer at scale genuinely may cost less than an originator servicing its own small book. Our claim is about diffusion, not price, and diffusion rises with length regardless of what happens to cost.

"Vertical integration has its own problems." True and worth weighting. An integrated firm can be worse — no external party checks its servicing, no counterparty scrutinizes its underwriting, and a customer with a complaint faces a single entity with no incentive to move. The integrated alternative isn't obviously better on accountability; it's better on locatability. You know who to hold responsible even if they won't respond.

"You haven't quantified the total extracted." Correct, and deliberately. Credible estimates would require assumptions about each participant's cost base that aren't publicly available, and a number built on those assumptions would carry more authority than it deserves. The structural argument doesn't depend on the total — it depends on the count of handoffs, which is observable.

Testable implications

1. Chain length should correlate inversely with borrower credit quality, holding product type constant.
2. Dispute resolution times should rise with the number of parties that have held the account.
3. Accounts that have transferred should show higher dispute rates than comparable accounts that haven't, because information degrades at handoffs.
4. Forbearance rates should be lower where servicing is separated from ownership, per the incentive argument above — the sharpest test, and one a comparison of self-serviced and third-party-serviced books answers.
5. Complaint volumes should concentrate at the firms with the least authority, because customers contact the brand they recognize.
6. Products with identical economics should show different chain lengths across regulatory regimes, which would evidence the structuring channel.

The third and fourth are both computable from data lenders already hold, and the fourth would settle a question our forbearance analysis could only argue from theory.

What would improve it

Not shortening chains, which would sacrifice real efficiencies. What we'd argue for instead:

• A single named accountable party disclosed to the customer, obliged to resolve or to route with the problem rather than the customer.
• Complete record transfer at each handoff, including notes and history rather than balances and statuses. Most information degradation is a data standard problem, not an inevitability.
• Servicer compensation aligned with recovery rather than with actions, which addresses the largest documented misalignment.
• Disclosure of the chain at origination — who holds the loan, who services it, and what may change.
• Warm handoffs on transfer, so a customer mid-arrangement isn't reset to zero.
• Supervisory reach that follows the activity rather than the entity, so that a function performed by a contracted third party carries the same obligations as one performed in-house.

The first is the cheapest and would do the most. Nearly all of the consumer-facing harm from long chains is a routing problem — the customer has a legitimate issue and no way to reach the party that can fix it. That's solvable without touching the economics of specialization at all.

Frequently asked questions

---

This report presents an analytical framework and the authors' interpretation; it is not legal or financial advice. Participant roles, contractual allocations of responsibility, and supervisory authority vary substantially by product and arrangement; no aggregate figure for total intermediation cost is asserted here, as credible estimation would require cost data that is not publicly available.