The Exit Problem: Why Staying Costs More Than Arriving | HL Hunt
The Exit Problem: Why Staying Costs More Than Arriving
Our search analysis explained why people don't compare offers when they buy. This report is about what happens afterward, and it produces a stranger result. In several consumer financial products the best available price goes to people who just arrived and the worst goes to people who stayed — which inverts every intuition about what a long relationship is worth. That isn't a pricing quirk or a failure of competition. It follows directly from one fact: a firm can charge an existing customer more than a new one by exactly as much as it costs that customer to leave, and in financial services leaving is expensive in a very particular way.
In this report
The pricing logic
Two customers, identical in every respect except one has been with the firm for six years. What price does the firm charge each?
| New customer | Existing customer | |
|---|---|---|
| Comparing alternatives? | Yes, actively | No |
| Cost of choosing elsewhere | Zero | Real and repeated |
| Price that retains them | The competitive price | Competitive price plus their switching cost |
| Firm's optimal price | Low, to win | Higher, because it holds |
Charging both the same would be irrational in either direction — either you overpay to acquire, or you leave margin uncollected on people who won't leave. So firms separate them, and the separation is the entire mechanism.
The critical property: the sustainable price gap equals the switching cost. Not the customer's loyalty, not the value of the relationship — the friction. Which means a firm that makes leaving harder has directly increased what it can charge, and that's a return on investment nobody has to be cynical to pursue.
The price gap a firm can sustain is exactly the cost of leaving. Not loyalty, not relationship value — friction.
What makes leaving expensive
No single barrier is large. The accumulation is what matters.
- Payment instructions. Direct debits and recurring charges pointing at an account have to be found and redirected, and per our descriptor analysis, customers frequently can't identify what's charging them — so the first step is an inventory nobody can complete.
- Deposit and payroll routing, which involves other parties.
- Bundled pricing, where leaving one product raises the price of the others — and a bundle is a switching cost sold as a benefit.
- Saved details and stored state — cards on file, preferences, records.
- The assessment risk. Applying elsewhere means being underwritten again, and the answer may be worse. This is the barrier unique to financial services and the most underrated.
- Time and attention, which per our search analysis is the binding constraint.
- Uncertainty about the outcome, since the new provider's actual terms aren't known until you've committed to finding out.
The assessment risk deserves separate treatment because it interacts with everything else. A customer whose circumstances have worsened since they were originally approved faces the possibility that leaving means being declined — so exactly the customers most in need of a better price are the ones least able to test whether one exists. The friction is highest where the stakes are highest.
And per our timing analysis, the application inquiry itself has a small cost and the process takes time during which nothing is resolved. The whole exit path is uncertain, effortful, and potentially worse than staying — which is enough friction to hold a substantial price gap without anyone doing anything improper.
The part that can't be taken with you
The switching cost specific to financial relationships, and the one that has no analogue in other markets.
The record of how you behaved with a provider mostly stays with the provider. Per our renewal analysis, behavioural data on an existing account dominates application data — a lender knows far more about a customer of four years than any new lender can learn from a bureau file.
Which produces an asymmetry that our verification analysis frames precisely: a credit file is portable trust, and it's a compressed, lossy version of what your current provider actually knows. The rest — payment patterns, how you handled a difficult month, whether you responded when contacted — doesn't travel.
So the incumbent has an information advantage that is genuinely earned and genuinely non-transferable. Two consequences:
- A new provider must price for uncertainty the incumbent doesn't have, which means the incumbent can beat any outside offer for a good customer and still charge more than a competitive price.
- A customer who has behaved well for years cannot prove it to anyone else. They've built an asset that only one firm can read.
That second point is the sharpest version of the exit problem. The better you've been as a customer, the more of your value is locked in a form only your current provider can see — so good behaviour increases the switching cost rather than the bargaining power, which is precisely backwards from how it should work.
What it costs to stay
Work a stylized case, because the per-period amounts are small and the cumulative figure is not.
A household with a $14,000 balance. New-customer pricing is 2.4 points better than what they're being charged after four years of drift.
- Annual cost of staying: $14,000 × 0.024 = $336
- Over four years: $1,344
- Effort to switch: perhaps four hours
- Implied hourly return on switching: $84 in the first year alone
By that arithmetic everyone should switch. They don't, and per our search analysis the reason isn't irrationality — the decision is never presented as $336 a year. It arrives as a rate that drifted, on a statement, with no comparison beside it, in a month when four hours weren't available.
Which is the crucial structural point: the loyalty penalty accrues continuously and the decision to address it must be made discretely. There is no moment at which staying costs $1,344 — there are forty-eight months in each of which it costs $28. And $28 never justifies four hours.
That's what makes this different from ordinary price dispersion. The search problem is about a cost paid once at purchase; this is about a cost that never presents itself at a size worth acting on, while summing to something substantial. A firm doesn't need the customer to be inattentive — only to be attentive at any single moment to an amount too small to move.
Competition that never arrives
The market-level consequence, and it explains an appearance that misleads observers.
A market can look intensely competitive while the majority of customers experience no price competition at all. Heavy advertising, aggressive introductory offers, and visible rivalry are all competition for the flow of new customers. The stock of existing customers is insulated from it.
Which produces a specific market structure:
- Acquisition is fiercely contested and unprofitable or barely profitable.
- The existing base is where margin is earned.
- Acquisition spending is funded by the base — which per our cross-subsidy analysis is the same structure as free checking funded by overdraft: one group's benefit paid for by another's, invisibly.
- Retention effort concentrates on customers who threaten to leave, which rewards the threat rather than the tenure.
The last point produces something worth naming. Where firms offer better terms on request or on a threat to leave, the price becomes a function of assertiveness rather than of circumstance — and the customers who ask are systematically not the ones paying the most. That's a distributional consequence of a retention practice that looks entirely reasonable from inside the firm.
Who pays it
The pattern is not random across customers.
Who stays longest and therefore pays most:
- People with less time and attention, per our liquidity analysis — those under financial pressure have least capacity for anything discretionary.
- People who fear being declined, which is exactly those whose circumstances have deteriorated.
- People with more products bundled at the same institution.
- Older customers, through longer tenure.
- People for whom the relationship carries non-price value — a branch they can reach, staff who know them.
Three of those five correlate with vulnerability, which makes the loyalty penalty a regressive charge that nobody set and that arises entirely from the interaction of ordinary pricing with unequal capacity to leave.
And it compounds with the premium our premium analysis documents. The same household pays more for the product initially and more again for staying with it — two separate mechanisms, both fixed-cost or friction driven, neither about risk, both landing on the same people.
Portability, not disclosure
What follows, and it's a narrower conclusion than the usual reform proposals.
Disclosure helps less than expected here. The problem isn't that customers don't know they could switch. Telling someone their rate is 2.4 points above a new-customer offer informs them of a $28 monthly amount that still doesn't justify four hours. Information doesn't change the arithmetic that's actually binding.
What works is reducing the cost of leaving:
- Automatic transfer of payment instructions, which removes the largest and most tedious barrier.
- Portable transaction and payment history, so a new provider can assess you on what you've actually done rather than on a compressed file — this addresses the information asymmetry directly, and the mechanism is the one our data access analysis describes.
- Conditional decisions before commitment, so the assessment risk is bounded and the customer knows the answer before starting.
- Unbundling, so leaving one product doesn't reprice the others.
- Prompted comparison at renewal, which creates the discrete moment the continuous cost never produces.
The second is the one this desk would weight most heavily, because it converts an asset only the incumbent can read into one the customer owns. That doesn't just reduce friction — it removes the mechanism by which good behaviour increases lock-in, which is the most perverse feature of the current arrangement.
And the fifth deserves attention because it's cheap. The loyalty penalty survives on the absence of a moment, so manufacturing one — an annual prompt that presents the cumulative figure rather than the monthly one — attacks the actual mechanism at almost no cost.
The strongest objections
"Introductory pricing is normal and pro-competitive." Largely conceded. Discounting to acquire is ordinary, exists in many markets, and can genuinely benefit customers who move. The argument isn't against introductory pricing — it's about what happens to the people who don't move, and specifically that the sustainable gap is set by friction rather than by cost, which means it isn't disciplined by anything.
"Incumbents have earned their information advantage." True and important. A provider that has served someone for four years genuinely knows more, and that knowledge was costly to acquire. The response isn't to deny the advantage but to note who owns the underlying behaviour — the customer generated the record, and portability transfers it rather than expropriating anything the firm created.
"You haven't measured the penalty." Correct. The $336 figure is stylized, gaps vary enormously by product and provider, and comprehensive data isn't publicly available for most markets. The mechanism holds regardless of magnitude — wherever switching is costly, the sustainable price gap is bounded by that cost rather than by anything competitive — but the size of the effect is an empirical question this report doesn't answer.
Testable implications
- Price gaps between new and existing customers should track switching cost across products, being largest where payment instructions and bundling create the most friction.
- Tenure should predict price paid, controlling for risk — the direct test, and answerable from any large provider's own book.
- Portability mechanisms should compress the gap where introduced, comparing across jurisdictions and products.
- Customers who ask should receive better terms than equivalent customers who don't, making price a function of assertiveness.
- Switching rates should be lowest among customers whose circumstances have worsened, consistent with assessment risk rather than with satisfaction.
- Presenting the cumulative rather than periodic cost should increase switching materially, testable by randomizing how a renewal notice frames the figure.
The sixth is the cheapest test of the core mechanism and could be run by any provider or regulator. If framing the same information as an annual total rather than a monthly rate substantially changes behaviour, then the penalty survives on presentation rather than on ignorance — and the remedy is a formatting requirement rather than a structural one.
The conclusion we'd hold: in markets where leaving is costly, price is set by friction rather than by competition, and loyalty is not rewarded because there is no mechanism that would reward it. The customer who stays is not being punished for staying — they're being charged what staying is worth to them, which is precisely the problem.
Frequently asked questions
Because the sustainable price gap equals the cost of leaving. A new customer compares and must be won; an existing one won't move for less than the hassle is worth.
Accumulated small barriers — payment instructions, bundling, saved details — plus the risk that a new provider assesses you and says no.
It's competition for new customers, which isn't the same as competition for customers. The existing base can experience none of it while the market looks fiercely competitive.
Portability — of payment instructions and of behavioural history. Disclosure helps less, because the problem isn't ignorance but an amount too small at any moment to act on.
Key takeaways
- The price gap a firm can sustain between new and existing customers equals the switching cost, not the relationship's value.
- Assessment risk is the barrier unique to financial services: exactly the customers needing a better price are least able to test for one.
- Behavioural history doesn't port, so years of good behaviour build an asset only the incumbent can read — raising lock-in rather than bargaining power.
- The penalty accrues continuously and must be addressed discretely; $28 a month never justifies four hours, though $1,344 would.
- A market can look intensely competitive while the existing base experiences no price competition, with acquisition funded by the people who stayed.
- Portability of payment instructions and behavioural history attacks the mechanism; disclosure mostly doesn't.
This report presents an analytical framework and the authors' interpretation; it is not financial or policy advice. The worked figures are stylized illustrations rather than estimates of any market's pricing, and comprehensive data on price differentials by tenure is not generally published. The implications identified as testable are hypotheses.