Borrowed Creditworthiness: The Guarantee Chain Nobody Maps

When the system can't establish that someone will repay, it doesn't usually refuse outright. It asks for a substitute — a cosigner, a guarantor, collateral, a deposit, a guarantee from another institution. Each of these answers the same question the same way: someone else stands behind it. That's the principal route by which people who can't be assessed on their own record get access at all, which makes it one of the most important mechanisms in consumer and small business finance. And it has a property nobody accounts for: the risk doesn't go anywhere. It lands on parties whose exposure appears in no record until the day it's called.

The substitutes

Substitute	Provided by	Cost to the provider
Cosigner	A relative or friend	Full liability, and their own file
Personal guarantee	A business owner	Household assets
Collateral	The borrower	The asset, if it exists
Security deposit	The borrower, in cash	Liquidity, which they lack
Institutional guarantee	A programme or agency	Borne collectively
Third-party insurance	An insurer, priced	A premium

Per our verification analysis, these are trust substitutes — ways of answering a question the record can't answer.

Two observations from the table. The fourth row is the cruellest substitution: a deposit demands cash from someone whose problem is that they don't have cash, which per our services guide is the standard requirement imposed on exactly the households least able to meet it.

And the first two rows are the ones that move risk onto another household, rather than onto an asset, a premium, or a collective mechanism. Those are the chain this report is about.

The system can't tell whether you'll repay, so it asks who else will. That question has a different answer for different people, and it isn't about reliability.

Transfer, not reduction

The property that gets elided in every discussion of guarantees.

A guarantee does not change the probability that the borrower fails to pay. It changes who bears the consequence. From the lender's position that's a genuine improvement — a second source of repayment is real. From the system's position, the risk has been relocated rather than removed.

Where it lands:

• On another household, in the cosigner case.
• On the same household, in the personal guarantee case — which per our structural analysis removes the containment the entity form existed to provide.
• On the borrower's own assets, in the collateral case.
• On a collective mechanism, in the institutional case — the only one where the risk is genuinely pooled rather than passed.

Which yields the framing: the guarantee chain is a mechanism for moving credit risk out of institutions, which are set up to hold it, and into households, which aren't.

Institutions hold capital against credit risk, diversify across thousands of exposures, price it, and measure it. A cosigner has one exposure, no capital against it, no diversification, and frequently no idea what it's worth. The same risk sits in far worse hands, and the transfer is what made the loan possible.

Why the exposure is invisible

The structural reason, and it's the same one our coverage analysis identified elsewhere.

A guarantee is a contingent obligation and reporting systems are built around current ones. So:

• A guarantor may look uncommitted to a subsequent lender while carrying substantial exposure.
• The guarantor frequently doesn't track it themselves, because nothing arrives monthly to remind them.
• Treatment varies. A cosigned obligation may appear on the cosigner's file and a business guarantee frequently does not, so similar exposures are visible or invisible depending on form rather than on substance.
• It activates on an event they don't control — and per our cosigning guide, frequently without warning, since the guarantor may not learn of missed payments until the obligation reaches them.

That last point produces the sharpest version of the problem. A person can acquire a delinquency on their own file, and a demand for a balance, arising from a decision someone else made and a deterioration they were never told about. Per our timing analysis, by the time the guarantee is called the record damage has already occurred.

And there's an aggregate consequence. Nobody can see the shape of the chain. How much contingent household exposure exists, who holds it, and how it correlates are all unmeasured — not because measuring is hard but because the obligations aren't recorded anywhere.

The correlation problem

The property that makes guarantees weakest exactly when they're needed — and it's the same finding our employer analysis made about payroll-linked credit.

A guarantor is typically close to the borrower: a parent, a sibling, a partner, a friend. Which means their circumstances correlate.

• They may be in the same household — the same income shock hits both.
• They may be in the same local economy, so a regional downturn hits together.
• They may be in the same industry.
• They may already be supporting the borrower informally, so the guarantee is called after other support is exhausted.
• A broad shock hits both at once, which is when guarantees are called across a portfolio simultaneously.

So the arrangement is weakest precisely when it matters. An individual guarantee is worth most in an idiosyncratic failure — the borrower specifically has trouble — and worth least in a correlated one, which is when failures cluster.

Which has an implication lenders should price and mostly don't: a portfolio supported by guarantees has more correlated tail risk than the individual assessments suggest, because the second source of repayment fails alongside the first. Assessing each guarantee on its own merits misses that they'll be called together.

Who can borrow creditworthiness

The distributional consequence, and it's the one that matters most.

Access via guarantee is distributed by who has someone willing and able to stand behind them — which is a different criterion from creditworthiness and one that tracks existing advantage closely.

Consider two identical applicants — same income, same reliability, same thin file:

• One has a parent who owns property. They get the loan, the apartment, the business line.
• One doesn't. They don't.

Nothing about their own reliability differs. The mechanism that determined the outcome was family wealth.

Which makes the guarantee chain a transmission mechanism for inherited advantage that operates through access rather than through transfer. No money changes hands — the parent simply signs — and the effect on the two applicants' trajectories is substantial and compounding, since the one who got the loan now builds the record that makes the next one unnecessary.

And it's self-reinforcing in the way our coverage analysis describes. The applicant who couldn't find a guarantor stays thin-file, so they need a guarantor next time too. The applicant who found one stops needing them.

This is also, uncomfortably, an argument for the mechanism. Without it the second applicant gets nothing at all, so guarantees expand access even while distributing it by the wrong criterion. The objection isn't that guarantees exist — it's that they're carrying a load that a functioning assessment system wouldn't hand them.

The small business case

Where the chain is most universal and least examined.

Per our guarantee guide, personal guarantees are close to standard in small business lending. Which means:

• The entity's limited liability is routinely waived for its largest obligations, so per our structural analysis most small businesses are households with extra paperwork.
• A business failure becomes a household failure, which per our failure analysis raises the personal cost of attempting a business substantially.
• Which suppresses formation among people who can't absorb that downside — a selection effect on who starts businesses, operating through the financing structure.
• The guarantee is weakest where correlation is strongest, since the owner's personal position depends on the business.

That last point is the one lenders should take most seriously. A guarantee from the owner of the failing business is a claim on someone whose income just stopped — which is the correlation problem in its purest form, and per our small business analysis it's why recovery on guaranteed small business obligations depends on the owner's separate assets rather than on the guarantee itself.

So the guarantee does two things: it improves recovery modestly, and it transfers the full downside of business failure onto the owner's household, which changes who is willing to start one. The second effect is larger than the first and nobody prices it.

What would replace it

The constructive question: if guarantees exist because the record can't answer the question, what would let the record answer it?

• Broader payment history. Per our coverage analysis, someone with eleven years of rent and utility payments is assessable — the data simply isn't collected. Recording it would eliminate the need for a guarantor for a substantial population.
• Cash flow assessment, per our affordability analysis, which works on people with no credit history at all.
• Lower cost of assessment, per our pricing analysis — since much of the barrier is the cost of establishing facts rather than the facts being bad.
• Structural risk reduction rather than transfer — the payroll channel, which reduces loss by changing collection rather than by finding someone to chase.
• Priced insurance instead of unpriced household guarantees, which at least puts a number on the exposure and puts it somewhere designed to hold it.

The last deserves more attention than it gets. An insurer holding this risk prices it, capitalizes it, and diversifies it — everything a cosigner doesn't. A guarantee is an insurance contract written by someone with no premium, no capital, and no portfolio, and the only reason it's the default is that it costs the lender nothing to ask for.

And the honest caveat: priced insurance costs money the borrower doesn't have, which is why the free version persists. The comparison isn't guarantees versus insurance — it's guarantees versus not lending, which is why they endure.

The strongest objections

"Guarantors consent." True and it matters. A cosigner agrees, and treating them as a victim is condescending. The response is about comprehension rather than consent — per our cosigning guide, cosigners frequently don't understand that they're liable for the full amount from the first missed payment, that they may not be told of trouble, and that the obligation appears on their file. Consent to a thing you've misunderstood is still consent and it's a weaker defence than it sounds.

"Without guarantees, these people get nothing." Correct, and conceded explicitly above. Guarantees expand access and the argument is not to remove them — it's that they're carrying a load created by a measurement failure, and that fixing the measurement would reduce the load. Removing guarantees without fixing the assessment problem would simply be rationing, per our rationing analysis.

"You haven't measured the exposure." Correct and it's the central limitation. Nobody has, because contingent household obligations aren't recorded — so the claim that the chain is large, correlated, and concentrated is inference from its structure rather than measurement of its size. That gap is itself the report's point, and it's the thing that would need fixing before the rest could be argued properly.

Testable implications

1. Access should correlate with family wealth beyond what income and file explain, among thin-file applicants specifically.
2. Guarantee calls should cluster in downturns more than individual assessments predict, reflecting correlation.
3. Guaranteed portfolios should show worse tail behaviour than the sum of their individual risk assessments implies.
4. Recovery on small business guarantees should depend on the owner's assets unrelated to the business, not on the guarantee's nominal strength.
5. Guarantors should be unable to state their exposure accurately — testable directly, and relevant to whether disclosure could work.
6. Where broader payment history is available, guarantee requirements should fall, which is the direct test of the substitution thesis.

The sixth is the one that matters most and is becoming testable. If adding rent and utility history to a file reduces how often a guarantor is required, then guarantees were substituting for a measurement gap rather than for genuine unreliability — and the chain would shrink without anyone restricting anything.

The conclusion we'd hold: the guarantee chain is how the unassessable get assessed, and it works by moving credit risk from institutions built to hold it onto households that aren't. It distributes access by who your family is, it's weakest when it's most needed, and the total exposure is invisible because the system records obligations only when someone owes money now.

Frequently asked questions

---

This report presents an analytical framework and the authors' interpretation; it is not legal or financial advice, and nothing here is guidance about whether to provide or require a guarantee. The obligations created by cosigning, guaranteeing, or pledging collateral vary substantially by jurisdiction and by the terms of the specific agreement, and anyone considering such an arrangement should obtain advice on their own position. No measurement of aggregate contingent household exposure is offered; the implications identified as testable are hypotheses.