Mercari, Inc.4385.T
Pull the consolidated 14.6% operating margin apart and a different company appears underneath it: a domestic C2C marketplace earning 35.0% and converting almost all of it to cash, carrying a balance-sheet-heavy credit book that drags reported free cash flow below zero and a US arm that only just turned its first profit. The inherited reading was a masked compounder waiting to be re-rated. Valued part by part, the core is worth more than the market gives it — but the discount lives at the enterprise line, and by the time it reaches the equity it is thin. What is left to decide is narrower: whether the 35% core is a durable compounder or a rent nearing its demographic ceiling, and whether an untested credit book stays benign.
Core Marketplace Japan, at ¥42bn of normalised operating profit on the 19x its liquidity moat and 35% margin earn, is worth roughly ¥798bn.
Merpay's credit-and-payment book adds ¥95bn at a base 10x on ¥9.5bn of profit — no platform premium, because the 99.4% recovery is a favourable-cycle number ; the newly profitable US marketplace adds ¥30bn on revenue ; the unallocated corporate drag removes ¥154bn.
The parts sum to an enterprise value of ¥769bn, against a market enterprise value of ¥656bn — a ~17% discount. Net of adjusted debt of ~¥40bn, equity is ¥729bn.
On 165.04m shares that is ¥4,417 in the base case, against a spot of ¥3,929. The enterprise discount is real. By the time it reaches the share price, most of it is already gone.
The interesting thing about Mercari is that the number everyone quotes describes a company that is not quite there. On a consolidated basis the group earns a 14.6% operating margin, generates negative free cash flow, and returns 9.4% on capital — the profile of a middling internet business that spent a decade not making money and has only recently stopped. The market has filed it accordingly, at 15.8x forward earnings and 5.4x book, near the floor of a ten-year multiple range. Underneath the consolidation the picture is not uniform at all. One domestic business — the C2C marketplace — earns a 35.0% operating margin on almost no capital. A payments-and-credit graft earns 20%, but only by carrying a balance sheet. A US arm that lost ¥12bn four years ago has just turned its first profit. The question the dossier turns on is whether that 35% core is a durable compounder the market has failed to price, or a high-margin rent approaching the ceiling of a shrinking domestic market, with an untested credit book bolted to its side.
That distinction matters more than usual, because the consolidated line actively hides it. Reported free cash flow has been negative every year since FY June 2022 (−¥12.2bn in FY June 2025), which reads like a company burning money. It is not. The cash is being absorbed by Merpay's credit receivables, which went from ¥4.5bn in FY June 2022 to ¥262.7bn in FY June 2025 — the balance sheet of a lender, not of a cash-burning platform. Strip the credit book out and the marketplace converts the way an asset-light network should. Read the consolidated 14.6% margin, the negative cash flow and the 9.4% return on capital as the real economics, and you have mispriced the business by averaging three things that should never be averaged.
The inherited reading was that this masked core was a clean dislocation waiting to be harvested — a 35%-margin compounder trapped behind a frightening balance sheet, worth materially more once someone did the sum. Do the sum carefully and the first half holds: the parts add up to an enterprise value about 17% above where the market carries the group. The second half is narrower. That 17% enterprise discount sits on top of a credit book financed by debt ; net the debt out and divide by a share count that has grown two-and-a-half times over the decade, and the equity discount thins to the low single digits. The core is genuinely undervalued at the enterprise line. The question is how much of that ever reaches the shareholder.
What is left, then, is a good asset priced at roughly fair value, with the upside living in two things the price does not hold. One is whether the 35% core is a compounder or a rent: consolidated GMV grew +11%, but that number is flattered by cross-border and hobby categories, and the organic domestic core is probably closer to its demographic ceiling than the headline suggests. If it saturates, the 35% becomes a harvestable rent rather than a reinvestable compounder, and the multiple that supports it falls. The other is capital: the company has never paid a dividend, never bought back a share, and diluted holders from 65m to 165m shares, while sitting on a platform that generates real cash. Any move to return it would restore the one thing this sector rewards independently of the multiple.
The position framing is active observation, not ownership at this level. The weighted asymmetry is modestly positive but there is no margin of safety in the price, and the case that would turn it — durable core margin, a clean credit book, a first return of capital — is decided on a published calendar. The FY June 2026 annual result, due around mid-August, settles the core margin, the Merpay recovery rate and the US trajectory in a single print. Conviction is moderate. This is a watchlist name with a documented long bias, waiting for one number.
The cleanest way to read the decade is as a single U — but a stranger one than most, because the two arms were driven by different things and the bottom is not quite where the story turns. Mercari spent its first six years buying a network: revenue compounded from ¥4.2bn to ¥76.3bn while the operating margin fell to −25.3% at the FY June 2020 trough, financed by the 2018 IPO and by debt. Then, from FY June 2021, the margin turned — a first profit at +4.9%, a wobble in FY June 2022 on US reinvestment, then a durable +9.5% in FY June 2023 as the asset-light core finally converted the user base it had bought at a loss. The recovery since has been real, and it accelerates: 14.6% in FY June 2025, 20.6% over the first nine months of FY June 2026. What makes the U strange is that halfway up it the balance sheet inverted — from ¥100bn of net cash to ¥51bn of net debt — not to fund the business, but to fund a credit book.
| Inflection | FY 2016Land-grab | FY 2020ZIRP trough | FY 2021First profit | FY 2023Core reveal | FY 2025Inflection |
|---|---|---|---|---|---|
| Revenue (¥bn) | 12.3 | 76.3 | 106.1 | 172.0 | 192.6 |
| EBIT margin | −0.3% | −25.3% | +4.9% | +9.5% | +14.6% |
| Return on equity | n/a | −53.1% | +15.4% | +28.3% | +30.5% |
| Return on capital | n/a | −23.8% | +5.7% | +7.3% | +9.4% |
| Credit receivables (¥bn) | 0.1 | 1.1 | 2.4 | 132.5 | 262.7 |
| Net debt (¥bn) | −22.8 | −88.6 | −100.1 | −44.0 | +51.2 |
| Diluted EPS (¥) | −3.2 | −147.9 | +35.1 | +77.6 | +154.8 |
Source: cellular verification of Mercari_4385.xlsm against the data pack, FY June convention. EBIT = reported operating income. Net debt shown with net-cash positions negative; the swing to net debt from FY June 2024 funds the Merpay credit book, not operations. FY June 2023 net debt anchored to the module (−44.0); the Bbg cell reads −58.0 on a wider debt definition.
Three management decisions explain the shape. The US expansion was allowed to lose money for far too long — around ¥26bn of cumulative losses across FY June 2022–2024 before break-even was imposed, capital and attention tied up for years in a business that only turned in FY June 2025. The Merpay credit book was scaled to ¥328bn (+45% year on year) largely on debt, on the strength of a 99.4% recovery rate that has only ever been observed in a benign credit cycle. And the complete absence of any return of capital, alongside the ×2.5 dilution and a convertible overhang that already pulls diluted EPS below basic, leaves the stock without the yield floor that — in this sector above all — is the only thing that protects a price when the multiple moves. The margin discipline since FY June 2023 is real and recent; whether it survives a credit downturn or a saturating core is exactly what is not yet proven.
The engine only makes sense once you stop looking at the consolidated line and look at the domains, because they are three economically different businesses wearing one ticker. The 9M FY June 2026 data is plain about it. Marketplace Japan earns 35.0% on ¥94.9bn of revenue. Fintech Japan earns 20.1%. The US earns 4.9%. A corporate line removes about ¥7bn. The consolidated 18% core operating margin is the weighted average of a genuine domestic franchise and two businesses that are ordinary or young — which is exactly why a single consolidated multiple is the wrong instrument, and why the sector's own rule is to value this name by taking it apart.
The most important thing to understand is where the cash actually is, because the reported number points the wrong way. Free cash flow has been negative for four years, and on the consolidated line it looks like a business that cannot convert. Look closer and the negative is mechanical: each incremental yen of Merpay credit lent out consumes balance sheet, funded by debt, and receivables have stretched to 441 days. The marketplace itself is almost capital-free — capex runs 0.16% of revenue — so the platform's own cash conversion is strong; the T1a proxy puts the retreated platform free cash flow near +¥43bn, about 6.6% of the market cap. It is a proxy, not a certified figure, and reconstructing it cleanly is the single most useful thing the next disclosure could provide.
The word that does the quiet work here is pricing power, and it lives in one place. The blended take rate of 13.3% looks like strength but is inflated by the gross-up of US delivery revenue; the real domestic take rate is ~10% and has been stable for years — sticky, but capped, because pushing it higher would push sellers toward the competition. Revenue grew +16.1%, but that was mix, not pricing: Fintech revenue rose +27% on the credit book, and the core's own growth is slower and probably nearer demographic saturation than the +11% headline GMV implies. The quality of the revenue is a notch below what the aggregate suggests.
The cost that can invert the whole economics is not industrial but financial: the cost of risk on the Merpay book. At 99.4% recovery and ~70bps of losses, it is a favourable-cycle figure on a young loan cohort, scored by a proprietary model that has never been tested in a downturn — and the loss lag is asymmetric, so the damage arrives late, in a recession, precisely when the ¥328bn book is at its largest. The sector's own stress case (70 to 300bps) takes roughly a fifth off group operating profit. Against that sits a business doing almost nothing wrong operationally and a balance sheet doing almost nothing at all: real platform cash, no dividend, no buyback, a share count that only grows. That dormant, un-returned cash is the one lever the price does not hold — and the one the company has never once pulled.
The thesis hangs on this pillar because the whole asymmetry turns on whether one demand stream holds and a second stays benign. The core demand is unusually good: a C2C resale network whose liquidity is self-reinforcing, structurally acyclical — resale intensifies in a downturn — with a TOPIX correlation of 0.23, the lowest of anything that is not a utility, and GMV still growing +11% against a shrinking population. Two things cap the score. The headline growth is flattered by cross-border and hobby categories, so the organic domestic core is probably nearer its demographic ceiling than +11% suggests; and the second demand stream — Merpay credit, +45% — is pro-cyclical and risk-bearing by construction, adding volume while lowering the average quality of demand. The base is rare and defensible; the growth leg depends on the core not saturating and the credit cycle not turning.
The moat is the second cardinal and the most decisive pillar in the dossier, because it alone decides whether the 35% is a durable compounder or a rent in remission. It is a genuine liquidity switching cost: sellers list where the buyers are and buyers come for the selection, a two-sided network effect that a decade of domestic leadership has made hard to attack head-on, reinforced by a proprietary credit-scoring data set built on the marketplace's own behaviour. The limit is that there is no contractual lock — nothing like a Zozo brand concession — and the one credible challenger, PayPay Flea Market, is backed by the SoftBank distribution machine. Deep, self-reinforcing, and without a contract to defend it; the whole thesis rides on whether that liquidity holds against PayPay.
Exceptional at the core (35% margin, capex 0.16%, ROE 30.5%), structurally dragged at the consolidated line — return on capital of 9.4% is barely 300bps over the cost of capital, and reported free cash flow is negative while the credit book grows.
Recent execution is strong — core operating profit +74.5%, US turned from −¥12.1bn to profit. The decade record is mixed: US losses were tolerated far too long, and the balance-sheet credit bet is an untested wager on a benign cycle.
No parent-subsidiary overhang, founder-led, no controlling predator (unlike Zozo under LY Corp). Against that: zero dividend, zero buyback, ×2.5 dilution and a convertible overhang. The open question is whether the cash is ever returned.
A real but not exceptional profile — a deep moat and an outstanding core pulled down by imperfect capital allocation and an untested credit book. Above a value trap, below a clean compounder such as Food & Life (19–20/25), and level with Zozo on the number, though the two get there differently. The grade fits the valuation: the core does not earn a premium on the consolidated line, and once the parts are summed the equity discount is too thin to claim. A masked compounder with a speculative graft, correctly priced.
Is the masked 35% core a durable compounder, or a saturating rent already in the price ?
The fintech graft: accretive spread, or imported credit risk ?
Opinion splits. One camp values the Merpay spread (20% margin, +45% book) as accretive optionality cross-sold at near-zero acquisition cost ; the other sees a debt-funded, balance-sheet-heavy loan book (net debt +¥51bn, receivables ¥263bn) importing an untested credit risk into an otherwise acyclical core. Both are right: the graft is accretive in margin and dilutive in cash conversion and risk profile, and the accretion reverses mechanically if the cost of risk normalises. At 99.4% recovery on a young cohort in a benign cycle, the resilience is asserted, not proven.
Does the dormant, un-returned cash ever get mobilised ?
The price assumes it does not: no dividend, no buyback, continued dilution, all cash reinvested into the core, the US and the credit book. Against that sit a platform generating real retreated cash (~+¥43bn proxy), a zero shareholder yield in a sector where the dividend is the only multiple-independent source of return, and a founder-led board with no controlling overhang to block a change of policy.
At ¥3,929 and 15.8x forward earnings — near a ten-year floor, de-rated from 41.5x in FY June 2023 and a 162x ZIRP peak — the market is pricing a mature platform at mid-teens profitability, a US at break-even, and a fintech treated as a modest margin contributor but a balance-sheet risk. What it over-discounts is that balance-sheet risk: the swing to net debt and the negative consolidated free cash flow have frightened cash-focused investors into compressing the multiple, when the negative cash flow is the artefact of financing a 99.4%-recovery loan book. What it under-prices is the 35% core, invisible under an 18% consolidated line, and the optionality of a first return of capital. Valued part by part, the enterprise value sits ~17% above the market's; by the time that reaches the equity, most of it is gone.
A double shock on one macro turn. The domestic core confirms saturation (GMV < +3%), re-rated from growth platform (19x) to mature rent (13x); and the Merpay cost of risk wakes up (70 to 150–300bps) as Japanese consumer credit slows, halving fintech profit (¥9.5bn to ¥5bn) and compressing its multiple. The floor holds near ¥2,412 because the liquidity moat survives and the core keeps generating cash — a reversible timing disappointment. Permanent loss enters only below ~¥2,000, and only if a credit cascade above 300bps couples with a structural, not merely decelerating, core decline.
The plan executes without surprise — the core holds 35% on gently decelerating GMV (+8%), the US stays profitable and scales modestly, the credit book grows at a stable ~70–90bps cost of risk. The SOTP at quality multiples (core 19x, fintech 10x, US 0.75x sales) delivers ¥4,417 on normalised operating profit of ~¥42.5bn, in line with guidance ≥¥40bn and consensus ¥42.7bn. A consolidated re-rating may or may not follow; the fair value does not need one.
The un-priced levers fire together. Finer segment disclosure makes the core margin visible and re-rates it toward 22x; the US scales to a double-digit margin and becomes a recognised value pole; the credit book stays clean at ≥99.4% recovery; and a first return of capital removes the yield penalty. The path needs the operational lift and the allocation decision at once, and neither is signalled today.
| KPI | Latest value | Status | What it tells us |
|---|---|---|---|
| Core Marketplace JP operating margin | 35.0% 9M FY2026 | Cardinal | The number the case rests on, four points above the T1a proxy. Two prints below 30% confirm saturation or moat erosion and pull fair value toward ¥2,400–2,900. |
| Merpay recovery rate | 99.4% | Cardinal | A favourable-cycle figure on a young cohort. Below 99%, or a cost of risk above 150bps, breaks the fintech accretion and turns a timing disappointment toward permanent loss. |
| US geographic operating profit | +¥1.19bn 9M FY2026 | Priced | From −¥12.1bn in FY June 2022; the accounted inflection that drove the +53% re-rating. A relapse would question the fundamental read. |
| Core GMV JP like-for-like | not isolated · consol. +11% | Watch | Headline flattered by cross-border and hobby. Organic core below +3% reclassifies the core from compounder to mature rent. |
| Merpay credit balance | ¥328bn · +45% | Watch | The balance-sheet-risk vector. Debt-funded; growth into a deteriorating credit cycle is where accretion turns to destruction. |
| Consolidated core OP growth | +74.5% on +16.1% revenue | Priced | The asset-light operating leverage, already largely captured in the +53% move. |
| Shareholder yield | 0% | Trigger | Zero dividend, zero buyback, ×2.5 dilution. An initiation of any return is the clearest un-priced re-rating lever in the name. |
| EV/EBITDA fwd · P/E NTM | 11.9x · 15.8x | Reference | Ten-year floors. The sector rule forbids reading the group on one multiple — the SOTP is the tool. |
The case moves from watchlist to long if the FY June 2026 print confirms the core and the company touches its capital. A domain Marketplace JP margin holding above 30% with organic core GMV durably above +5%, alongside a US that stays profitable, would validate the compounder and support the base and bull paths; an initiation of any dividend or buyback would remove the sector's yield penalty and re-rate the multiple in its own right. Both are observable at a dated event; neither is signalled today.
The case validates the bear if the narrow engine stalls or the credit book turns. A core margin falling below 30%, or a Merpay recovery rate below 99% over the same print, would reset fair value toward ¥2,412 — the first signalling moat erosion in the liquidity core, the second the start of an adverse credit cycle that inverts the fintech economics. A permanent impairment below ~¥2,000 needs both a credit cascade above 300bps and a structural, not cyclical, decline in core GMV — a conditional tail, not the central bear.
The allocation risk is the one to watch most, because the company's instincts run that way. Continuing to grow the ¥328bn credit book at +45% into a deteriorating cycle would convert the graft's accretion into destruction and burn the very cash that underwrites the bull case. The mirror risk is the opposite — that the cash stays dormant indefinitely, denying the stock the yield floor the sector rewards and leaving it fully exposed to a multiple that this bucket moves on the second derivative regardless of the underlying quality. Neither is signalled today.
The information provided on this website is for informational and educational purposes only and should not be construed as financial, investment, legal, or tax advice. All content reflects the personal opinions, interpretations, and analyses of the author at the time of writing and is subject to change without notice. Nothing contained herein constitutes, or should be interpreted as, a recommendation, solicitation, or offer to buy or sell any securities, financial instruments, or other investment products. The author is not a licensed financial advisor, broker, or investment professional. Any references to specific assets, markets, or strategies are illustrative in nature and do not constitute personalized investment advice. Investing in financial markets involves risk, including the potential loss of capital. Past performance is not indicative of future results. Readers are solely responsible for their own investment decisions and should conduct their own independent research and due diligence before making any financial commitments. You are strongly encouraged to consult with a qualified financial advisor, legal professional, or other relevant specialist before making any investment or financial decisions. By accessing and using this blog, you agree that the author shall not be held liable for any direct or indirect losses, damages, or consequences arising from the use of, or reliance on, the information presented herein. All content is provided "as is" without any warranties of completeness, accuracy, or reliability.