Most English-language writing about renovating Japanese vacant houses is either a single person’s project diary or a number with no provenance. Neither helps you decide. What follows is the distribution from our own surveys, with the counts stated so you can see how thin or thick each part of it is.
All figures are from the 94 houses we had surveyed and listed as of 7 October 2026. They describe those houses. They are not averages for Japan.
The headline distribution
For each property we record a low and a high renovation estimate. Across the 92 houses where we had both:
| Low estimate | High estimate | |
|---|---|---|
| Minimum | ¥300,000 | ¥700,000 |
| Median | ¥1,500,000 | ¥4,000,000 |
| Maximum | ¥8,000,000 | ¥20,000,000 |
The useful information here is not the median. It is the width. Within the low-estimate column alone, a factor of about 27 separates the cheapest house from the dearest (¥300,000 to ¥8,000,000); across both columns the spread is a factor of about 67 (¥300,000 to ¥20,000,000). The band inside a single property is wide too: the median high-to-low ratio is 3.0, 38 of the 92 houses have a ratio above 3, and the widest is 5.0. Anyone quoting you “akiya renovation costs about X” is averaging across a population that does not have a centre.
The number that should change how you shop
In 21 of those 92 properties, the midpoint of our renovation estimate was higher than the asking price.
That is not a rare pathology. It is almost a quarter of what we surveyed. It means that for a meaningful share of cheap Japanese houses, the house is the small part of the transaction, and the listing price is close to irrelevant as a signal of cost.
The practical consequence is a change in how you compare. Stop comparing asking prices. Compare asking price plus renovation midpoint, and then ask what else is available at that total. A ¥2,800,000 house with ¥9,000,000 of work is an ¥11,800,000 project, competing with every ¥11,800,000 house on the market — many of which are habitable on handover.
For reference, the asking prices across all 94 ran from ¥1,500,000 to ¥125,000,000 with a median of ¥9,800,000. Across the set, the ratio of renovation midpoint to asking price had a median of about 0.32 — but a maximum of 8.67.
What condition grade actually predicts
We grade condition A to E. The distribution across all 94:
- A — 3 houses
- B — 19
- C — 47
- D — 23
- E — 2
Nearly half sit at C. That is the market: not derelict, not ready, needing real work with real uncertainty. The grade correlates with scope, which we record separately:
- Partial renovation — 45 houses
- Full renovation — 25
- Cosmetic only — 12
- Light — 6
- None — 3
- Major — 2
- Structural — 1
Only 21 of 94 — the cosmetic, light and none categories — are houses where you could reasonably expect to spend under a couple of million yen and be finished.
The four things that actually drive the number
Age is the variable everyone fixates on and it is not the most useful one. Our set spans 1952 to 2019 with a median build year of 1987, and the cost spread inside any decade is wider than the spread between decades. What moves the estimate is more specific.
1. Water, past and present
Evidence of water ingress is the single most expensive thing to find. A stain on a ceiling is a question about the roof, the frame beneath it, and how long it went unnoticed in a house nobody was living in. Rot in a structural member turns a cosmetic job into a structural one, and that is the jump from the ¥1,500,000 band to the ¥8,000,000-plus band.
2. Drainage, because it is probably a septic tank
73 of the 90 houses where we recorded it drain to a septic tank (浄化槽), not a municipal sewer. Only 17 were on public sewer.
A septic tank is a normal and workable arrangement in rural Japan. It is also a machine. It has a service life, statutory inspection obligations and periodic emptying. If the house has stood empty for years, the unit may be past useful life, and replacement is a line item of its own — plus the groundwork to get at it. When you budget a rural Japanese house, budget the drainage as a system, not as a connection.
3. Water supply, which is not always the town’s
62 of 90 are on public water. Of the rest, 23 draw on a private association supply (a shared village or estate system) and 5 on a private source such as a well or a spring.
None of those is disqualifying. All of them carry obligations the municipal supply does not: association fees and governance, pump and tank maintenance, water quality testing. If a house has been empty, pipework may need replacing before it will hold pressure, and on a private source you want a water quality report before you commit to the house being habitable.
4. The 1981 line
27 of the 93 houses with a recorded build year were completed before 1981. We record an applicable-standard field per property: 61 post-1981, 27 pre-1981, 5 post-2000.
One caveat on our own data. Japan’s current seismic standard took effect on 1 June 1981, and whether it applied to a given building is determined by the date of its building confirmation certificate, not its completion year. A house completed in 1981 or 1982 may well have been confirmed — and therefore designed — under the previous standard. We classify by completion year because that is what listings record; for any individual house, the confirmation certificate is the document that settles it.
Pre-1981 does not mean unsafe and does not compel you to reinforce. It does mean the design basis is the older one, which affects your risk, sometimes your insurance terms, and your options if you later want to finance or let the property commercially. Reinforcement costs vary enormously with the frame and the foundation, and it is the clearest case for getting the building opened up before you fix a budget.
What our estimates are, and what they are not
This is the part that most published renovation figures leave out, so we will be explicit.
None of the 94 houses had a completed building inspection on file. Where we recorded the field, it recorded that no inspection had been done. Nobody had lifted the floor, opened a wall, or put a camera in the roof space.
Our estimates are therefore made from observation: what is visible inside and out, the age and construction type, the utility arrangement, and comparable work. They are good enough to tell a ¥1,500,000 house from an ¥8,000,000 one. They are not good enough to be a contract price, and we would not present them as one.
The honest sequence is: use an observation-based estimate to decide whether a property is worth pursuing, then get a proper inspection before you commit to a number. The inspection costs a fraction of the variance it removes.
Building a budget that will survive contact
- Take the high estimate, not the midpoint. On a building nobody has opened up, surprises are asymmetric — they are almost always in the expensive direction.
- Price the utilities as systems. Septic tank, water supply, and the pipework between them and the house. If the building has stood empty, assume this work exists until shown otherwise.
- Add the cost of finding out. An inspection, and for a distant buyer, somebody’s travel. This is the cheapest money in the project.
- Then compare totals, not asking prices. Price plus renovation high estimate plus the first year of holding costs. That is the number to set against alternatives.
- Decide what you will not do. A full seismic upgrade on a 1965 house may be the right call or may be the thing that makes the project never work. Deciding in advance stops the budget drifting.
Getting a house looked at from overseas
Every one of the 8 overseas enquiries we received in the 120 days to 7 October 2026 came from someone living outside Japan, naming a specific property they could not go and inspect. If that is your position, what makes it workable is having a record in front of you.
For a property you are considering — ours or anyone else’s — we will go to it, make a Matterport 3D record inside and out so you can walk the building and take your own measurements, and write down what we could verify on site. We write only what we could see; anything we could not check is recorded as not verified. It is billed under its own contract as a scanning and recording job, quoted per property on travel and floor area. It is not a brokerage fee and it does not oblige you to buy through us.
For most of the houses in the distribution above, that record is what separates a ¥1,500,000 assumption from an ¥8,000,000 reality.
Nothing here is construction, legal or investment advice. Costs vary by region, contractor and year, and the figures above describe our own surveyed properties as at 7 October 2026.