Who Finds Out First: The Order of Discovery in Household Trouble | HL Hunt
Who Finds Out First: The Order of Discovery in Household Trouble
When a household gets into difficulty, the people who could help and the people who find out are not the same people, and they do not find out in a useful order. The bank sees it in days. Other lenders cannot see it for 45 to 90 days. The party with the earliest view can do almost nothing except charge for it, and the party that could restructure arrives fourth with the narrowest picture. That sequence is set by data plumbing rather than by anything about usefulness.
What you'll learn
The core thesis
When a household gets into difficulty, the people who could help and the people who find out are not the same people, and they do not find out in a useful order.
Our position is that the sequence of discovery determines who is able to act, that the sequence is set by data plumbing rather than by anything about usefulness, and that the party with the earliest view is routinely the one least able to do anything with it.
The order is roughly this:
| Who | When | What they see |
|---|---|---|
| The bank or card issuer | Days 0 to 3 | The balance, the failed transfer, the changed spending |
| The landlord or utility | Days 5 to 10 | A late payment |
| The lender | Days 1 to 30 | Its own missed payment only |
| The credit bureau | Days 30 to 60 | Whatever was furnished, per our reporting analysis |
| Other lenders | Days 45 to 90 | The bureau's version, second hand |
The gap between the first row and the last is 45 to 90 days. During it, every other lender continues treating the household as it was, which per our stacking analysis is the same window that makes simultaneous borrowing possible.
The party that knows least can do most
Line up the capacity to help against the position in the sequence and the mismatch is almost exact.
- The bank sees first and can do almost nothing. It can decline a payment, charge a fee, or close the account. Per our overdraft analysis, its available responses mostly make the situation worse.
- The landlord sees early and has one lever, which is the most consequential one available to anyone, per our tenancy analysis.
- The lender can restructure, defer, reduce, or forbear, per our payment plan analysis, and arrives fourth with the narrowest view.
- Other lenders can only tighten, and they act last on the oldest information.
- The household is sometimes last to name it, because the signals are spread across accounts nobody aggregates.
The third bullet is the expensive one. Per our early stage analysis, the same difficulty addressed at day ten has more options than at day sixty, and the party with the options gets the information late and partially. A lender sees that its own payment was missed and does not see that three others were, or that the pattern started two months earlier in the current account.
Why the order is what it is
- Transaction data is immediate and belongs to one party. Per our portability analysis, it is the richest view available and the least shared.
- Furnishing is monthly. A thirty-day cycle plus processing is a structural delay rather than a failure.
- Rent and utilities mostly do not furnish on time, so the earliest payment failures never enter the shared record at all.
- Each lender sees only its own account. Per our exposure analysis, nobody holds the aggregate except the bureau, and the bureau holds it late.
- The household has no consolidated view either, per our household analysis, particularly where accounts sit in different names.
Point three is the one that does the most damage and gets the least attention. Rent is the largest payment most households make and the first one they fall behind on, and because it is rarely furnished, the system's earliest and most informative signal is simply absent from the shared record.
A falsifiable claim
Current account signals should predict default on other lenders' products earlier and better than bureau data does, and the lead should be measurable in weeks. Any institution holding both transaction data and bureau data on the same customers can test it: compare the predictive performance of transaction-derived features against bureau-derived features at increasing horizons before default. If the bureau data performs comparably at the same horizon, the sequencing argument here adds little. We expect a substantial lead for transaction signals, which would mean the information required to intervene usefully already exists and sits with a party whose only available responses are fees and closure.
Why the order is defensible
The obvious conclusion is that everyone should find out sooner, and that conclusion deserves resistance.
- Faster shared signals mean faster tightening. Per our monoculture analysis, if every lender learned on day three, every lender would reduce limits on day four, which is the sequence that converts a bad month into a crisis.
- Most early signals resolve themselves. Around 62% of accounts a few days past due cure with no intervention, so a system acting on day-three information would act mostly on noise.
- Transaction data is the most sensitive data a household has. Sharing it faster is a surveillance expansion, not only an information improvement.
- The lag is also protection. Per our portability analysis, a household currently invisible to other lenders during a difficult month has room to recover before anyone prices it.
- Furnishing accuracy falls with speed. Per our furnishing analysis, faster reporting of unsettled situations produces records that are wrong more often.
The first bullet is the serious one and it reverses the naive recommendation. The useful change is not making the signal travel faster to everyone. It is getting the signal to the one party that can restructure, which is a different and much narrower proposition.
What follows for institutions
- Read your own transaction data for distress, not just for underwriting. Per our cash flow analysis, the institution holding the account already has the earliest view and generally uses it only at application.
- Separate the response from the detection. Detecting difficulty should trigger an offer, not a fee, and that distinction is the whole value of being early.
- Ask the customer rather than inferring. A single question at the right moment outperforms a model built to avoid asking it.
- Do not act on day three. Per our false positive analysis, a signal at that base rate is mostly noise and acting on it trains customers to ignore you.
- Where you are both bank and lender, connect the two. Most institutions that hold both views do not join them, which forfeits the only genuine advantage in this sequence.
Item five is the practical opportunity and it is largely untaken. An institution with the current account and the loan has the earliest view and the broadest set of responses in the same place, which is the combination nobody else has, and per our bundling analysis it is also the strongest honest argument for the relationship discount that bundling pays.
Scenarios and what we are watching
| Scenario | What we would expect |
|---|---|
| Open data access becomes routine | The 45 to 90 day gap compresses toward days, with the tightening problem in the first objection arriving alongside the benefit |
| Rent reporting spreads | The earliest signal enters the shared record, which helps on-time renters and exposes late ones, per our portability analysis |
| Furnishing frequency increases | Faster records and more errors, per our furnishing analysis, with the errors falling on consumers |
| Nothing changes | The party that can help keeps finding out fourth, and the party that finds out first keeps responding with fees |
The first row is the development that would resolve the mismatch and create a new one. Compressing the gap gets information to the lender who can restructure, which is the gain. It also gets it to every other lender simultaneously, and correlated tightening across every provider in the same week is a mechanism for turning a recoverable month into an unrecoverable one. Whether faster is better depends entirely on which party acts on it first, which is a question of design rather than of speed.
Frequently asked questions
The bank, within days. The lender sees only its own missed payment, and other lenders not for 45 to 90 days.
Capacity to help runs opposite to the sequence: the earliest observer can mostly only charge fees, while the party able to restructure sees least and latest.
No. If everyone learned on day three, everyone would tighten on day four, and most early signals resolve themselves without intervention.
Get the signal to the party that can restructure, not to everyone. Where one institution holds both the account and the loan, join the two views.
Key takeaways
- The bank sees difficulty in days; other lenders cannot see it for 45 to 90 days, and the gap is structural rather than a failure.
- Capacity to help runs opposite to the sequence: the earliest observer can mostly only charge fees or close the account.
- The lender able to restructure arrives fourth and sees only its own missed payment, not the pattern across accounts.
- Rent is the earliest and most informative signal and is rarely furnished, so it is absent from the shared record.
- Faster sharing to everyone would produce correlated tightening within a week, which turns a bad month into a crisis.
- About 62% of accounts a few days past due cure unaided, so acting on day-three information is mostly acting on noise.
- The untaken opportunity is institutions holding both the account and the loan joining the earliest view to the broadest response set.
- Detection should trigger an offer rather than a fee, which is the entire value of being early.
This analysis is educational and does not constitute financial or legal advice. All timings are stylized illustrations of typical sequences and vary substantially by product, furnisher, and jurisdiction. Furnishing frequency, what each party is permitted to observe or share, consumer data access rights, and the responses available to each institution differ by institution type and by jurisdiction and change over time.