The Search Problem: Why Identical Borrowers Pay Different Prices
The Search Problem: Why Identical Borrowers Pay Different Prices
Two people with the same score, income, and loan request approach different lenders and receive quotes several percentage points apart. Neither lender is behaving improperly. Both are pricing rationally given what they know about the market — which is that most borrowers will accept the first acceptable offer. The standard reading of this is that consumers fail to shop and should be educated. We think that reading is wrong, and the correction has real consequences: for most consumer loans, not shopping is the correct decision, because the expected savings are smaller than the cost of finding them. Dispersion isn't a failure to be fixed by trying harder. It's a stable equilibrium, and it responds to different interventions than the ones usually proposed.
In this report
The core thesis
Competitive markets compress prices because customers move to the cheaper seller. That mechanism requires customers to know which seller is cheaper, and knowing costs something.
Where the cost of comparing is small relative to the stakes, comparison happens and prices converge. Where it's large relative to the stakes, comparison doesn't happen and prices don't converge — and the sellers' prices reflect that reality rather than each other.
Formally, a borrower should search when:
Expected savings × Probability of finding a better offer > Cost of searching
The critical structural point: expected savings scale with loan size while search cost does not. Comparing five lenders takes roughly the same effort on a $2,000 loan as on a $300,000 mortgage. The savings differ by two orders of magnitude.
This is the same asymmetry that drives our selection analysis — fixed costs against size-scaling benefits — operating on the consumer's side of the transaction rather than the lender's. In both cases small loans are the casualty, and in both cases the mechanism is arithmetic rather than behavioral.
Three claims follow. That non-shopping is usually rational, not a deficit. That dispersion is therefore self-sustaining — no search means no pressure means dispersion persists means still not worth searching. And that interventions targeting motivation will fail while interventions targeting search cost will work, which is a testable and consequential difference.
Expected savings scale with loan size. Search cost doesn't. Everything about consumer shopping behavior follows from that single asymmetry.
When search pays
Work it through. Assume a borrower can find a rate 3 percentage points better by comparing several lenders — a realistic figure in personal lending, where dispersion at the same credit tier is substantial.
| Loan | Amount | Term | Approx. savings from 3 points | Worth 3 hours? |
|---|---|---|---|---|
| Small personal loan | $2,000 | 2 yr | ~$65 | No |
| Personal loan | $8,000 | 4 yr | ~$510 | Marginal |
| Auto loan | $28,000 | 6 yr | ~$2,700 | Clearly |
| Mortgage (0.5 pt) | $320,000 | 30 yr | ~$33,000 | Overwhelmingly |
Read the first row carefully, because it's the one that matters. A borrower who spends three hours comparing lenders to save $65 on a $2,000 loan has earned roughly $22 an hour — before accounting for the cognitive effort, the friction of multiple applications, and the perceived risk to their credit file. For most people that's a bad trade, and declining it is not a mistake.
Note also the interaction with risk tier. Dispersion is widest at the bottom of the credit distribution, where pricing is most idiosyncratic and lenders differ most in risk appetite. So the borrowers facing the widest dispersion — where search would pay most in percentage terms — are frequently taking the smallest loans, where it pays least in dollar terms. The two effects work against each other and the size effect wins, which is why the population that would benefit most from comparison shops least.
The natural experiment
The strongest evidence that this is economics rather than psychology: the same person behaves completely differently depending on loan size.
A household that will compare four mortgage lenders, request written estimates, and negotiate points will take the first auto loan the dealer offers and accept a store card at checkout without reading the rate. If the explanation were financial literacy or attentiveness, that person would be consistent. They aren't — and the direction of the inconsistency is exactly what the arithmetic predicts.
Three features reinforce it:
- Mortgage disclosure is standardized, which lowers comparison cost directly — the loan estimate exists to make offers comparable, and it's the one market where a borrower can lay quotes side by side and read the same fields. Our closing analysis notes that even here, the components that aren't standardized — title, settlement services — go unshopped by nearly everyone. Same transaction, same consumer, and the shopped items are exactly the ones made easy to compare.
- Rate shopping windows exist for mortgages and auto loans specifically, treating multiple inquiries in a short period as one event. The market with the deliberate search subsidy is the market with the most search.
- Where comparison is mechanical, it happens. Deposit rates, which are trivially comparable and require no application, show far more consumer movement than loan rates at equivalent stakes.
The conclusion we'd draw: consumer shopping behavior tracks the cost of shopping, not the sophistication of the consumer. Which means "consumers don't shop" is a statement about market design.
What search actually costs
The dollar-value of time understates it. The full cost has components that are individually small and collectively decisive.
- Time to find lenders, in a market where search results are dominated by paid placement and lead generators rather than by rate.
- Application effort per lender, repeated — each one wants the same information entered again.
- Perceived credit damage. Substantial and largely mistaken. Scoring models treat multiple same-purpose inquiries within a window as one event, and prequalification frequently uses a soft inquiry with no effect at all. The belief operates as a search cost regardless of whether it's accurate, which makes correcting it one of the cheapest available interventions — and it's an information problem, not a motivation problem.
- Comparison difficulty. Offers differ on rate, fees, term, and structure simultaneously. Working out which is cheaper requires computation most people won't do — and the products designed around fees rather than rates, per our earned wage access and lease-purchase analyses, are frequently not comparable at all.
- Uncertainty about approval. A borrower unsure of qualifying anywhere has a strong incentive to accept an offer in hand.
- Timing pressure. Much borrowing responds to an immediate need — the repair, the deposit — and urgency collapses the search window to zero. The borrowers most likely to face urgent needs are the ones facing the highest prices, which is the same population identified in our liquidity analysis.
- Cognitive load, which is highest for households already managing financial stress.
Two of these compound in a way worth isolating: urgency and low liquidity are correlated, and both suppress search. A household with a buffer can take three days to compare. A household without one is choosing between the first offer and no repair — which means the liquidity constraint is also a search constraint, and it hits the population already paying the most.
Why dispersion is self-sustaining
The loop is what makes this an equilibrium rather than a transient inefficiency:
- Search is costly relative to the stakes, so few borrowers compare.
- A lender pricing above the market loses few customers, because few are looking.
- Competitive pressure toward the low price is therefore weak.
- Dispersion persists.
- With dispersion persisting but individual savings still small relative to search cost, borrowers continue not searching.
Each step follows from the last, and no participant is behaving irrationally. The lender pricing high is responding correctly to demand elasticity; the borrower not searching is responding correctly to their own cost-benefit. That's what makes it stable and what makes exhortation useless against it.
Two consequences that follow and are worth stating separately:
Marketing substitutes for price. If customers aren't comparing rates, the way to win them is to be the first offer they see. Acquisition spend rises, and the cost is embedded in the price — which raises the fixed cost per funded loan in the way our selection analysis describes, further raising prices at small sizes. The two mechanisms reinforce each other.
Lenders can price by channel rather than by risk. Where search is absent, the same lender can offer different rates to identical borrowers arriving through different channels, because there's no arbitrage. This is legal and standard, and it means a borrower's price partly reflects how they found the lender — a variable with no relationship to their creditworthiness, and one that deserves fair lending attention where channel correlates with demographics.
The prediction that tests it
A frame that only explains is weak. This one predicts, and the predictions are checkable against public data.
Dispersion should vary inversely with loan size, holding risk constant. If search cost drives dispersion, then markets where search pays should show tighter pricing. Ranked by expected dispersion, tightest first: mortgages, auto loans, large personal loans, small personal loans, small-dollar credit.
That ordering matches observation, and — importantly — it's not the ordering a pure risk explanation predicts. Risk-based dispersion would track the variance of borrower outcomes, which doesn't line up with loan size nearly as cleanly. The size ordering is the search explanation's signature.
Further predictions:
- Standardized disclosure should compress prices in the product it covers, and not in adjacent unstandardized products in the same transaction. The title and settlement services example is exactly this case and it points the right way.
- Soft-inquiry prequalification should increase comparison rates and narrow dispersion where adopted.
- Aggregators returning multiple real offers from one application should compress prices among participating lenders, since they collapse search cost to near zero.
- Financial education should have negligible effect on shopping rates for small loans — because the decision not to shop is correct and education doesn't change the arithmetic. This is the sharpest test, and the education literature's generally weak results on downstream financial behavior are consistent with it.
- Urgency should suppress shopping independent of financial capability, which predicts that liquidity-constrained borrowers shop least at every income level.
The fourth is the one that could most cleanly falsify this. If a well-designed education intervention substantially raised comparison rates for small loans, the search-cost account would be wrong and a knowledge-deficit account would be right. We'd expect it not to, because a borrower who fully understands they could save $65 for three hours of work will still, correctly, decline.
What compresses prices and what doesn't
| Intervention | Mechanism | Expected effect |
|---|---|---|
| Standardized disclosure | Lowers comparison cost directly | Strong |
| Soft-inquiry prequalification | Removes application friction and the perceived credit cost | Strong |
| Multi-offer aggregation | One application returns several real prices | Strong, if offers are firm |
| Rate shopping windows | Removes the inquiry penalty | Moderate — already exist, poorly understood |
| Correcting the inquiry misconception | Removes a cost that's mostly imaginary | Moderate and very cheap |
| Rate caps | Truncates the distribution | Compresses at the top; removes supply at small sizes per our selection analysis |
| Financial education | Addresses motivation | Weak — the decision was already correct |
| Exhorting consumers to shop | None | None |
The distinction running down that table is the report's practical output: interventions that reduce the cost of comparing work, and interventions that try to increase the desire to compare don't.
Which reframes a policy debate. The consumer protection instinct on dispersion is disclosure-plus-education. The analysis says the disclosure half works and the education half doesn't — and that the highest-return unexploited intervention is unglamorous: make prequalification with a real rate and no hard inquiry universal. That single change removes application friction, removes the perceived credit cost, and makes the offer comparable, which is most of the search cost in one move.
The strongest objections
"Consumers systematically underestimate the savings, so the decision isn't rational." The best objection. If borrowers believe dispersion is 1 point when it's 4, they're declining a better trade than they think. Genuinely plausible, and it argues for publishing dispersion data as a cheap intervention. But note it doesn't rescue the education approach — the correction needed is a specific fact about market dispersion, not general financial capability, and it's an information intervention rather than a motivational one. So the objection modifies our account without changing the policy conclusion.
"This excuses lenders who price opportunistically." It explains the equilibrium; it doesn't endorse any participant's conduct. Two things remain criticizable regardless. Pricing by channel rather than by risk is a practice the analysis illuminates and doesn't justify, particularly where channel correlates with protected characteristics. And deliberately raising search cost — non-comparable fee structures, offers that expire under time pressure, quotes that require an application to obtain — is a strategy the framework predicts and condemns rather than excuses.
"Dispersion reflects genuine risk differences, not search." Partly true and the two are hard to separate in observational data, since lenders differ in what they can see about the same borrower. Our response is the size ordering: if dispersion were purely risk assessment differences, it wouldn't vary so cleanly with loan size at constant borrower quality. Risk assessment doesn't get easier because the loan is bigger. Search does get more worthwhile.
What a borrower should actually do
The analysis is only useful if it changes behavior, and the honest advice is size-dependent rather than uniform:
- Above roughly $10,000 or any long term: shop. The savings clear any plausible cost of time by a wide margin, and this is where the money is.
- Use prequalification, which returns real rates without hard inquiries at most lenders and eliminates most of the cost.
- Know the inquiry rules. Multiple same-purpose inquiries within a window count as one, and prequalification frequently doesn't count at all — the mechanics are in our inquiries guide.
- Compare total cost, not the monthly payment. Payment comparison is how longer terms get sold as cheaper, and it's the single most exploited comparison error.
- Check a credit union, which is the highest-expected-value single additional quote for most borrowers and takes one call.
- On small loans, don't feel obliged to exhaust the market. Two quotes captures most of the available benefit; the third and fourth rarely pay for themselves.
- Shop before you need it. Urgency is the largest search cost of all, and it's the one you can eliminate in advance — which is a further argument for the buffer our savings analysis describes.
- Improve the file rather than the search where you're at the bottom of the distribution. Moving a tier is worth more than finding the best lender within a tier, and it's the mechanism our option analysis describes.
That last point is worth stating plainly because it inverts the usual advice. For a borrower facing wide dispersion because they're priced as impaired, the return on improving their file exceeds the return on shopping — a tier move is worth multiples of the within-tier spread, and it's durable across every future transaction rather than applying to one loan.
Frequently asked questions
Because little competitive pressure forces convergence. In a market where most customers accept the first acceptable quote, a lender pricing above the best available rate loses very few of them.
Almost entirely a function of size. Savings scale with the amount while effort doesn't — three points on $2,000 is about $65, and on a mortgage the same comparison is worth tens of thousands.
Far less than believed. Same-purpose inquiries in a short window count as one event, and prequalification often uses a soft inquiry. The misconception itself functions as a search cost.
Anything lowering comparison cost — standardized disclosure, soft-inquiry prequalification, multi-offer aggregation. Education targets motivation, and the decision not to shop was usually correct to begin with.
Key takeaways
- Expected savings scale with loan size while search cost doesn't, which makes not shopping the correct decision on most consumer loans.
- Three points of dispersion is worth about $65 on a $2,000 loan and tens of thousands on a mortgage — the same borrower rationally treats them differently.
- Dispersion is a stable equilibrium: no search means no pressure means dispersion persists means still no reason to search.
- Where search is absent, lenders can price by channel rather than by risk, which deserves fair lending scrutiny.
- Dispersion varying with loan size at constant risk is the search explanation's signature, and not what a pure risk account predicts.
- Interventions that lower comparison cost work; interventions that raise motivation don't — universal soft-inquiry prequalification is the highest-return unexploited one.
This report presents an analytical framework and the authors' interpretation; it is not legal or financial advice. Worked figures are stylized illustrations of the arithmetic rather than quoted market rates, and the magnitude of price dispersion varies by product, tier, and period. Predictions identified as testable should be treated as hypotheses.