When the Resale Price Becomes Part of the Sale

01 / The Answer

Buyers price the second number whether or not you show it. What moved primary transactions was making it certain, not making it visible.

The strongest identified evidence in this cohort says buyers are already computing the second number. In the college textbook market, student purchase behaviour tracked the probability that a new edition would appear and destroy resale value, not the price on the shelf. Demand elasticity for a book certain not to be revised was about −0.9. For the same book facing certain revision, it was about −3.7. Fact The implied discount rate was low enough to rule out myopia and most forms of hyperbolic discounting.

That result is the load-bearing wall of this study, and it is worth stating precisely, because it is easy to overclaim. What it establishes is that buyers incorporated expected resale value into willingness to pay before modern price discovery existed, in a market where nobody surfaced the number for them. Students in the late 1990s were doing this arithmetic without a marketplace, a chart, or a checkout module.

What it does not establish is that making the number salient, credible, precise or immediate cannot increase the effect. The evidence here does not show that simply showing buyers a number creates incremental demand. It also does not show that it fails to. No public source tests it directly, and that gap is real rather than rhetorical. Unresolved

That distinction sorts the entire cohort. Every case where the second number measurably changed the first sale involved a company removing uncertainty, removing delay, or removing the transaction entirely, and being paid for doing so. Every case where a company simply published or facilitated a residual value produced either nothing measurable or a reverse-logistics cost centre.

So the answer is conditional, and it is narrower than the market's answer. Surface the second number where independent data already says yours is better than the alternative. Facilitate where the customer is already standing in your channel. Capture where verification is expensive enough that buyers pay a premium for certainty. Underwrite where the uncertainty is about whether the category holds value rather than about the price level of a mature one. Otherwise leave it alone: a used unit that substitutes cleanly for a new one is a competitor, and dressing it up as a program does not change that.

These carry very different evidential weight. Section 10 grades each one, and only a single item in the framework is supported strongly enough to state as a rule.

02 / The Four Choices

Four different decisions get filed under one word, and the confusion is where the money is lost.

Companies say "we're doing resale." In the record, they are doing one of four things with completely different economics, different risk owners, and different failure modes. The most expensive errors in this cohort came from executing one while believing they were executing another.

The four are not a ladder. They are alternatives with different preconditions, and the cohort contains companies that moved between them in both directions. What the cohort does not contain is a company that made residual value visible and captured the resulting demand without also owning either the certification or the redemption instrument.

03 / The Sequence

The producer is almost always the last participant to notice, and the first move is almost always defensive.

The order below is the one the evidence supports across autos, textbooks, games, phones and watches. It is not the tidy version. In particular, steps three and four happen without the producer's participation, and step six fails often enough to be treated as a predictable error rather than bad luck.

The timing evidence is unusually clean at both ends. The theoretical case that killing the used market does not pay was published in 1974. The empirical confirmation arrived in the 2000s. Pearson announced its digital-first restructuring, which removes the resale option entirely rather than shortening it, in 2019. Fact Rolex operated for decades alongside a vast independent secondary market before launching certified pre-owned in December 2022, and acquired the retailer that piloted it the following year. Fact

04 / When the Second Number Helped the First

The mechanism was certainty, immediacy, or a currency the buyer could only spend with you.

Category uncertainty, priced and then retired

Tesla introduced a resale value guarantee in 2013, committing to buy back a Model S after roughly three years at a set fraction of the purchase price. The problem it solved was not depreciation. It was that nobody knew whether a used electric car from an unproven manufacturer would be worth anything at all. Three years later, with retained values reported above the guaranteed floor, the program ended. Reported Tesla disclosed the associated liability at $1.58 billion as of March 2016. Fact

Read the entry and the exit together and the lesson is precise: the guarantee was priced against category risk, and it was withdrawn the moment the category stopped being risky. It was never a permanent feature. It was the cost of getting a new asset class its first resale prices.

A currency redeemable in one place

GameStop's trade-in program is the cleanest example of capture in the cohort. Gross margin on pre-owned product ran near 48 cents on the dollar, against roughly 21 cents on new software and single digits on hardware. Estimate The company has consistently told investors that trade-in drives market share and widens its price points. Company claim

The structural detail is the one that matters. Trade credit was routinely worth materially more than cash for the same item, and it could only be redeemed in GameStop. That converts a customer's residual value into a closed-loop currency: the second number becomes demand for your own inventory rather than money leaving the building. No producer in this cohort achieved that without owning the redemption.

The trade-in that is not really a trade-in

Smartphone trade-in is the case most often cited as proof that residual value drives upgrades. The published values do not support that reading. US carriers advertise trade-in credits that far exceed the standalone market value of the same handset, delivered as bill credits spread over 24 or 36 months, conditional on porting a number, activating a line, staying on a qualifying plan, and not paying the device off early. Fact

That is not residual value. It is a subscriber acquisition subsidy denominated in residual value, and the leash is the point. The device is the pretext for the contract. Inference

Certainty as the product

Certified pre-owned is the oldest formalised version of Choice C. The first manufacturer program launched in 1993, and the motive was residual defence: manufacturers who had underwritten lease residuals needed a channel that would support the values they had promised. Reported The certification premium on cars is modest, roughly 4.3% to 4.7% for a three-year-old vehicle. Reported

The same structure priced very differently in watches, where authenticity risk is severe and verification is genuinely hard. Certified pre-owned Rolex watches carried premiums reported between roughly 16% and 42% depending on retailer, against the same models sold by non-participating dealers. Reported The buyer is not paying for the watch. They are paying to stop worrying about the watch.

Why the trade-in makes buyers less price sensitive

There is a behavioural mechanism underneath all of this that is worth naming, because it explains why trade-in is such an effective margin tool. Experimental work published in 2008 found that consumers in a transaction involving a trade-in act as both buyer and seller, and prioritise getting a good value for the used item over getting a good price on the new one, producing a higher willingness to pay for the new product. Fact

That is the commercial case for a trade-in desk stated honestly. It is not that residual value creates demand. It is that a trade-in moves the customer's attention off your price.

05 / When It Hurt

Three failure modes: surplus captured by someone else, residuals underwritten into a soft market, and resale run as a marketing program.

The producer funds the market and the middle captures it

College textbooks is the best-identified negative case available anywhere. Publishers spent decades shortening revision cycles to suppress the used market. The demand estimates say it did not work: students repriced the book, and simulated faster revision cycles produced flat or lower publisher revenue before revision costs. Fact

Meanwhile the surplus was real and it went to the channel. Bookstores bought back at roughly half the new price and resold at roughly three-quarters, on gross margins around 34% against roughly 23% on new books. Fact A single textbook could pass through several users after the initial sale with no further revenue to publisher or author. Company claim

The resolution is instructive. Publishers eventually stopped fighting the second number and eliminated it. Pearson's 2019 shift to digital-first moved print to rental only, broke the print revision cycle, and cut list prices sharply toward digital price points. Fact The consistent reading is that they finally accepted the theory: if you remove resale, you must also cut price, and what you gain is not margin but control of the channel.

Underwriting a soft market with a known asset

Auto leasing is the largest quantified loss in the cohort, and it happened twice.

Through the 1990s, captive finance arms set inflated residual values to lower monthly payments and drive lease volume. When used prices fell, the gap became real money. A worked example from the period: a 1997 full-size SUV listed at $36,580 with an expected three-year residual of $25,606 was worth $16,500 at auction in 2000, a shortfall of about $9,100 per unit. Reported Industry-wide losses from mis-set residuals were estimated at $10.5 billion for 2000 and around $5 billion for 2003. Estimate

Then it happened again with better data and bigger numbers. In the second quarter of 2008, Ford Credit recorded a pre-tax impairment of $2.1 billion on North American operating leases after concluding that lease-end residuals on full-size trucks and traditional SUVs would come in far below expectation. Fact Ford Credit ended 2008 with a $1.5 billion net loss, more than explained by that charge and related items. Fact Chrysler Financial exited leasing entirely within days; Ford repriced leases on the affected lines to the point of discouraging them. Fact

Note what was and was not uncertain. Nobody was unsure whether pickup trucks had resale value. The uncertainty was about the level of a mature market, and that is exactly the uncertainty a manufacturer is worst placed to hold, because its own incentive and production decisions move the number it guaranteed.

Trying to abolish the second number

Microsoft's 2013 attempt to restrict used-game resale on Xbox One is the clearest natural experiment on suppression. The policy was announced, the competitor publicly committed to leaving disc resale alone, preorder positions moved against Microsoft, and the policy was reversed within roughly a week. Fact Preorder rankings shifted back after the reversal. Reported

The buyers were behaving exactly as the textbook demand system predicts. Remove the resale option and willingness to pay falls, immediately and visibly.

Branded resale as a marketing program

Apparel and outdoor gear produced the largest number of launches and the least evidence of primary-market benefit. Two quiet signals are more informative than the launch announcements.

First, REI closed its online used-gear shop in late 2025 while keeping used gear in stores. Reported That is the shape you would expect if the economics break on shipping, grading and handling rather than on demand. Second, the largest resale-as-a-service vendor eliminated upfront and monthly fees for branded resale storefronts entirely in 2025. Fact A software layer given away is a software layer the customers were not paying for.

The honest reading is that most branded apparel resale is Choice B executed for reasons that are not commercial, and that the public evidence for a primary-market effect is thin. Unresolved

06 / Who Actually Made the Money

In the majority of the cohort, residual value stimulated behaviour that someone other than the producer monetised.

This is the finding an executive should sit with longest. Across the cases where a visible second number demonstrably changed buying behaviour, the producer captured the economics in a minority of them, and only where it owned certification or redemption.

The two producer wins in this set, watches and games, share a structural feature rather than a category feature. Both involved a step the customer could not perform themselves and would pay to avoid: authenticating a watch, or converting a shelf of used discs into spendable value today. Where no such step exists, the producer has nothing to sell into the second transaction. Inference

07 / When the Second Number Became the Product

The boundary is real. The threshold is not, and the evidence does not support pretending otherwise.

The brief asked whether something changes when the expected resale price exceeds the purchase price. It does, but not as a clean line. What changes is the identity of the buyer. Below the line, the buyer is a user protecting a downside. Above it, a meaningful share of buyers are holding inventory, and inventory demand behaves nothing like consumer demand.

Two cases carry this section. In sneakers, the average premium paid on new releases of the signature retro model fell from a peak near 61% in 2020 to about 4% in 2023, with some 2023 colourways trading below list. Reported Across the market, the share of releases trading above retail fell from roughly 58% in 2020 to about 47% in 2024. Reported Industry participants attribute the compression to oversupply and heavy discounting rather than to taste alone. Reported

In collectibles, the 1990s plush toy boom is the fully completed version of the same arc. Annual sales at the manufacturer went from around $280 million in 1996 to roughly $1.4 billion in 1998, driven by deliberate scarcity, retirements and a fast-growing online resale market. Reported Production was then raised into the demand, a blanket retirement announcement in 1999 triggered panic selling rather than a price rise, and resale values collapsed. Reported At the peak, the category accounted for a large share of activity on the dominant auction platform, which flagged fad concentration as a business risk in its own filings. Fact

What the cohort cannot establish is causation in the reverse direction. Falling secondary prices and falling primary demand moved together in both cases, but supply increases, discounting, macro conditions and simple taste change are confounded. Whether the resale collapse caused the primary decline, or both were caused by the same oversupply, is not separable from public data. Unresolved

The usable conclusion is narrower and still actionable: once appreciation is part of why people buy, supply policy stops being an operations decision and becomes a pricing decision, and the producer is now managing a market rather than a product line.

08 / The Earliest Signal

The signal is elasticity, not participation. Participation is the one everyone watches and it is the one that misleads.

Ranked by how well the historical record actually supports them, with the false signals included because they are the ones companies act on.

The practical version of signal one: find a moment when the expected resale value of your product changed without your price changing. A competitor launched a buyback. A platform started publishing prices. A model was discontinued. Measure conversion around it. If conversion moved, the second number is already in your customer's arithmetic and you are choosing whether to participate, not whether it exists.

09 / The Recurring Mistake

Companies underwrite while believing they are advertising.

This mistake appears often enough, and expensively enough, to be the single most useful thing in the study. A subsidised residual set to hit a monthly payment target does not read internally as risk transfer. It reads as a promotion. It is booked as marketing. But it is a written put option on your own used market, at a strike you chose for sales reasons, over a horizon you cannot hedge, on an asset whose price your own future output and incentive decisions will move.

The contrast with participants who understand the instrument is stark. In aircraft, residual value guarantees are disclosed as off-balance-sheet arrangements, capped, and conditioned on tight asset return standards. One manufacturer's average guaranteed residual was around 15% of original sale price in one year's disclosure and around 11% the next, with exercise windows a decade or more out. Fact That is a company that knows it is short an option and has priced, capped and reserved accordingly.

Two supporting errors travel with it:

• Launching Choice B and staffing it like Choice A.
  Branded resale is intake, grading, cleaning, authentication, storage, fulfilment and a materially harder customer service load than new goods. The programs that retrenched retrenched on logistics, not on demand.
• Treating trade-in participation as evidence of demand effect.
  Participation responds to promotion generosity. In the best-measured category, participation and value returned rose while the replacement cycle lengthened. The two moved in opposite directions.

10 / The Decision Framework

Gate first, then choose. And keep the tiers separate, because most of this is not a rule.

The gate. Is a credible second number for your product already discoverable from a source you do not control? If yes, you are deciding whether to participate in an existing market. If no, you are deciding whether to create one, which is a different and much worse-evidenced decision. Nothing in this cohort shows a producer successfully manufacturing residual-value salience from nothing. The markets formed first, independently, and the producers arrived later.

• Heuristic, weakly held
  If independent data already shows your retained value beats the alternative → surface it.
  You are citing evidence, not creating it, and if your retained value is average or worse you are handing a competitor a comparison axis. Weakly held because the underlying demand effect is unmeasured in both directions. The defensive half of this — don't publish a number you lose on — is better supported than the offensive half.
• Pattern, small sample
  If the customer is already in your channel and the goods do not have to move → facilitating is more likely to survive.
  In-store take-back persisted across the cohort; mail-in and online-only take-back is where retrenchment shows up. But this is a handful of cases, closures are systematically under-reported, and the online programs were also the newer and more marketing-driven ones, so channel and vintage are confounded. It is a good question to ask about your own program, not a finding that settles it.
• Heuristic
  If condition or authenticity risk is high enough that buyers pay for certainty → capture.
  The observed certainty premium runs from roughly 4% where verification is easy to more than 40% where it is hard and the brand is the only credible authority. The durable part is not the range, which is two categories, but the arithmetic: does the premium you can charge cover refurbishment, authentication, inventory and returns? Run that before building anything. If it does not, you are subsidising the secondary market.
• Heuristic, n=1
  If the uncertainty is about whether the category holds value at all → a guarantee may buy something real.
  One clean case supports this, and one case is not a finding. What makes it worth carrying is that the entry logic and the exit logic agree: the guarantee was written against category risk and withdrawn once the category had its own price history. If you cannot describe the condition under which you would withdraw yours, you are not using it for this purpose.
• Pattern, strongly supported
  Guarantees written against the price level of a mature market are where the losses cluster.
  Two large episodes, one industry, roughly a decade apart, with billions in documented and estimated losses. The mechanism is specific: you cannot hedge it, your own volume and incentive decisions move the number you guaranteed, and the strike gets set by whoever needs the monthly payment to work. This is not an argument that it can never be done — the aircraft cases underwrite mature-market residuals and do not produce comparable losses. It is an argument that doing it demands the structure below.
• Rule — the only one here
  If you guarantee a residual, cap the exposure, condition it on return standards, reserve for it, and set the strike outside the sales organisation.
  This is the one place the cohort speaks with one voice. Every documented loss came from a guarantee that was uncapped, unreserved, or struck for volume reasons. Every guarantee that was capped, conditioned and disclosed as what it is produced no comparable loss, including in a category with far longer horizons and far more volatile asset values. The variable that separated the outcomes was structure, not category, not asset, not size.
• Heuristic
  If the used unit substitutes cleanly for the new one and someone already handles the transaction cheaply → stay out.
  You have no step to sell into the second transaction, so there is nothing to capture. Your leverage is in product architecture and price. The textbook resolution — remove the resale option and cut price to match — is the coherent version of this, and it took the industry decades to reach.
• Pattern, two completed cases
  If expected resale exceeds purchase price → you are running a market, not a product line.
  Both completed arcs show the producer raising output into the peak and compressing the premium that was driving demand. Whether the compression caused the primary decline is not separable from the supply expansion that accompanied it, so treat this as a warning about supply policy rather than a proven causal chain. The transferable part: revenue peaks before the secondary price does, and a meaningful share of your buyers are holding inventory.

One cross-cutting heuristic, strongly held. Before any of the four choices, decide what instrument the second number is paid in. Cash leaves. Store credit stays. Bill credits chain the customer to a contract. Certification sells trust rather than goods. Across seven categories the instrument tracked who captured the surplus more reliably than the category, the product, or the size of the residual. It is an association across cases rather than a tested mechanism, which is why it stops short of a rule, but it is the most consistent regularity in the study.

11 / The Cohort

Fourteen cases, selected for variation in outcome and quality of evidence.

What would change the answer

Three findings would move this materially.

A controlled test showing that surfacing residual value at checkout lifts conversion or average selling price, holding price and promotion constant. Nothing in the public record establishes this, and it is the claim the resale industry makes most often. If it exists inside a company's own experiment data, it would upgrade Choice A from a comparison weapon to a demand lever.

Published unit economics for a branded apparel resale program showing contribution after intake, grading, logistics and service. Every figure available today is either category-level market sizing or vendor-sourced.

Evidence separating the resale premium collapse from the supply expansion that accompanied it in sneakers. That would convert the study's weakest section from directional to identified, and would tell producers whether the second number is a cause of primary demand or a symptom of it.