I Found a 16% Yield Apartment Near Japan's New Semiconductor Hub. Then I Checked the Land Registry.

This article is based on a real due-diligence investigation conducted in 2026. Identifying details — neighborhood, lot numbers, seller, and brokerage — have been omitted or generalized to avoid identifying the property, which may still be listed. No wrongdoing by any party is alleged; the point is how listing conventions and public records differ, and what that means for a buyer. This is general information, not investment, legal, or tax advice.

I found a fully occupied six-unit apartment in Myoko City, a short walk from one of Japan's most interesting new semiconductor investments, advertised at a 16.4% gross yield. On paper it was the kind of number that makes you clear your afternoon. This is the story of what happened when I actually underwrote it — the parts that passed, the one number that didn't, and the framework I now run every regional Japanese property through.

Let me be clear up front about where it ended: I have not bought it, and as of writing I still can't, because a title question remains open. That is the whole point.

The Macro Seduction

The pull was obvious. Israel-based Tower Semiconductor announced in July 2026 a roughly ¥600 billion expansion of its Japanese manufacturing, with a METI subsidy of up to ~¥160 billion, centered on its Uozu (Toyama) operations and the Arai plant in Myoko City (Niigata) — with regional reporting putting roughly ¥80 billion into the Myoko-Arai facility to make optical semiconductors (silicon photonics) for AI data centers. A fab means engineers, contractors, and years of employment. My initial thesis wrote itself: an aging, high-yield asset sitting on top of a decade of new, well-paid rental demand.

I won't re-argue the Myoko macro case here — I laid out both catalysts and their risks in After Hakuba and Niseko: Myoko and Furano. What matters for this article is a discipline I've learned the hard way: a strong macro story is a reason to investigate a property, never a reason to skip steps on it.

First-Pass Due Diligence — Which Passed

The listing: a 6-unit wooden apartment, 2K layouts, built 1992, two stories, fully occupied, with parking for six to eight cars. Price ¥14.8M, annual rent ¥2.424M, advertised gross yield 16.38% and an advertised "net yield" of 14.28%.

Here is the part that matters for credibility: this was not an obvious dud. The first-pass numbers were sound.

• Rent is real, not inflated. Units were let at ¥33,000–35,000/month against an independent market read of roughly ¥34,000. No over-renting to flatter the yield.
• Occupancy looks organic. Lease start dates were spread across 2022–2024 — the signature of a building that fills naturally, not one staged with related-party tenants right before sale.
• Management terms are reasonable. ¥1,500 per occupied unit (¥500 vacant), with an existing interior/equipment repair history on file.

If I had stopped at the income statement — as the advertised yield invites you to — I would have made an offer. Most of the risk in regional Japan is not in the rent roll. It is one and two layers below it.

The Building — An Open Risk

Next layer: the structure. A ~35-year-old wood-frame building in heavy-snow country is not disqualifying, but it demands answers. I could find no record of major work on the roof, exterior walls, the main water and drainage lines, or the structure itself. In Myoko, snow load and annual snow-removal are not footnotes — they are a recurring line item and a structural-fatigue question at the same time. This didn't kill the deal. It moved "major capex reserve" from optional to mandatory in my model. Call it an open risk, priced but unresolved.

The Number That Didn't Make Sense: The Land

Then I got to the land, and one figure refused to reconcile.

The listing stated 606.13 m². When I pulled the registry, the lots actually being conveyed totalled about 165.54 m². That is not a rounding difference. That is a different property.

This is where a foreign buyer has to understand that Japan runs parallel record systems that do not automatically agree — exactly the lesson, at the micro level, of Rosenka vs Market Price in Japan: the official number and the real number are different objects, and reconciling them is your job. The 606.13 m² figure traced not to a current survey but to the property's 1992 building-confirmation certificate (建築確認, kenchiku kakunin) — a document brokers commonly cite in regional listings. It is not a lie; it is a different system, describing the site as it was permitted in 1992.

To reconcile it, I pulled three documents, all Japanese-only, each obtained separately:

• the cadastral map (公図, kōzu) — the shape and arrangement of the lots;
• the registry (登記, tōki) — legal ownership and recorded areas;
• and available survey records — the measured reality.

Laid side by side, they told a more complicated story than the listing. The site was made up of several parcels:

Lots A + B + C — the parcels I could confirm were part of the sale — sum to 165.54 m². Lot D (~395 m²), which makes up the bulk of the "606 m²" impression, appeared to have been subdivided off at some point in the past, and its current ownership was still being confirmed. The premise "you are buying 606 square meters" could not be taken at face value; it had to be rebuilt from the records.

The Reframe

Two findings changed the underwriting entirely.

A residual-lot subdivision history is a yellow flag, not a red one — but it must be resolved before, not after, contract. Land that has been carved up over the decades can leave slivers, shared parcels, and ambiguous ownership behind. Until Lot D's status is confirmed in writing, the actual land you would own — and therefore the actual value and future development potential — is unknown.

The private-road lot (Lot B) matters more than its size suggests. A lot registered as a private road (私道) can be included in a sale, but its status under Japan's Building Standards Act determines something crucial: rebuild rights. A building's right to be reconstructed often depends on its frontage to a qualifying road. If the road lot's legal status is unconfirmed, the building's future — and any exit that assumes a buyer can rebuild — carries a question mark.

Then the contract terms, which are typical for regional Japan but heavy for a foreign buyer to absorb:

• no boundary demarcation (境界未確定);
• no survey;
• sold as-is (現況有姿);
• and the seller's contract-nonconformity liability (契約不適合責任) waived.

Individually, each of these is common. Stacked on top of an unresolved land question, they mean the buyer assumes essentially all of the title and boundary risk. This is precisely the situation where a judicial scrivener (司法書士, shihō shoshi) and, for boundary issues, a licensed surveyor (土地家屋調査士) are not optional — they are the people who tell you what you are actually buying before you sign.

Financing Reality

Even setting title aside, the financing profile is structurally hard. A Tokyo-resident buyer purchasing an old wooden building in regional Niigata at a sub-¥15M ticket is a combination most banks quietly decline: the collateral is small, old, and far from the buyer. The realistic paths are Japan Finance Corporation (日本政策金融公庫), local/regional institutions, or loan-broker platforms — and those platforms can charge a success fee (roughly ¥495,000 minimum on approval), which is material against a ¥14.8M ticket.

For non-resident readers, be honest with yourself: this is harder still. Standard Japanese mortgages are largely unavailable to buyers living abroad, and a small, old, rural asset is the least bankable profile there is. In practice this is a cash purchase for most non-residents — which changes the return math and the liquidity of any exit.

Recomputing the Yield Honestly

The advertised "14.28% net" does not survive contact with a full cost stack. A defensible model for this asset has to carry a major-repairs reserve (35-year-old wood frame, no capex history), insurance, a realistic vacancy assumption, snow removal, and leasing/turnover fees. Add those and the true NOI — and the honest net yield — lands meaningfully below the headline. Still potentially a good number; just not the one on the flyer. (Depreciation on old wooden buildings can produce real tax effects, but a tax tail should never wag the investment dog — if the deal only works on paper losses, it doesn't work.)

For the broader reality of what regional ownership actually costs year to year, see our Japan property running costs guide; and for context on aging regional stock generally, our data-driven akiya area ranking. One framing note: this is a long-term-rental asset, and I underwrote it as one — not a minpaku play. The tenants are residents on standard leases; the semiconductor thesis, if it plays out, is about durable local employment, not tourism.

The Lesson, Stated Plainly

Two sentences summarize the whole investigation:

Japan is easy to buy, but not always easy to underwrite. The transaction mechanics are open to foreigners; the diligence is where deals are won or lost.

A strong macro story makes investors overlook micro-level property risk. A macro thesis does not fix a title defect. The fab will not resolve Lot D's ownership. The subsidy will not confirm the road lot's rebuild rights. Those are property-level facts that exist regardless of how good the regional story is.

This is the same discipline, one level down, that separates official data from market reality across all of Japanese real estate — and it is why this property was worth a week of investigation even though I haven't bought it.

The Reusable Framework: A Property-Level DD Hierarchy

Every property gets checked in this order, because each layer can veto the ones above it. A great rent roll cannot save a defective title. Here is where this specific property landed:

The investigation is ongoing: as of publication, Lot D's ownership confirmation is still pending, which is exactly why the property sits at "critical unresolved" and not "cleared" or "rejected."

The semiconductor story made the property worth investigating. It did not make the property worth buying — yet.

Related Reading

• Rosenka vs Market Price in Japan — the same "official number vs real number" lesson at the market level
• After Hakuba and Niseko: Myoko and Furano — the full Myoko macro thesis and its catalysts
• The Kumamoto Playbook: 4 Years of TSMC and Japanese Real Estate — what a completed semiconductor-town cycle teaches
• Japan Property Running Costs: Monthly & Annual — the recurring-cost reality behind any yield
• Best Akiya Areas in Japan: Data-Driven Ranking — context on aging regional stock

Tower Semiconductor / METI figures are from the July 14, 2026 announcement (¥600B total, up-to-¥160B subsidy, ~¥80B Myoko-Arai portion as reported by regional media); they are announced plans subject to change. Property details have been anonymized. Title, boundary, and road-status questions in Japan should be handled by a licensed judicial scrivener (司法書士) and land surveyor (土地家屋調査士). This article is general information, not investment, legal, or tax advice.