Japan will keep making akiya until the 2040s

The relationship those numbers describe changed in 2005. That year 1,062,530 Japanese children were born and 1,083,796 Japanese people died, a natural change of −21,266 and the first natural decrease since the Ministry began counting in 1899. It was not a dramatic year on the ground. Twenty years later the same subtraction produces −918,253, which is forty-three times the size of the first crossing.

Our sister publication The Japanist has read that crossing against the full record since 1947, including the birth peaks of 1949 and 1973 and the marriage and divorce series. This article is narrower and considerably more self-interested. If you own a Japanese house, or you are thinking about buying one, which parts of that arithmetic are actually your problem?

Japan's population statistics are unusually well published and unusually easy to misquote. Four rules do most of the work, and every figure below follows them.

These are not pedantry. They are the difference between reading a figure and decorating a sales narrative with one, and the distinction matters more in this market than in most, because the buyer is usually working in a second language on someone else's summary of someone else's translation.

Deaths are the easier of the two lines to misread, because they look like a verdict on a country's health. They are mostly a verdict on its age structure. Japan's fewest recorded deaths came in 1966, at 670,342, and the line has climbed since the late 1970s not because Japanese people started dying younger but because the first postwar baby boom started arriving at the ages at which people die.

The age breakdown makes the point better than any argument. In 2025 the Ministry recorded 1,286,048 deaths of people aged 75 and over — four in five of the total — and a further 177,295 aged 65 to 74. Fewer than 130,000 deaths, under 8 percent, were people below 65.

Now run that forward. Everybody who will be 75 in 2043 has already been born, and so has everybody who will be 65. The national institute's 2023 projection therefore does not have to guess at the death line so much as read it: annual deaths peak at about 1.67 million in 2040, and the population aged 65 and over peaks at 39.53 million in 2043 as the second baby boom, born between 1971 and 1974, moves into that band.

A death is not a listing, and the gap between the two is where most of the practical difficulty in this market lives. A house passes to heirs who often live in Osaka or Tokyo, who may not register the title, who face demolition costs and a tax code that can make leaving the building standing cheaper than clearing it. We took the long way through that pipeline — including the roughly 58 percent of vacant homes in the ministry's owner survey that were acquired by inheritance, and the registration duty that has applied since April 2024 — in what foreign buyers can actually buy from Japan's nine million akiya.

But the direction is not in doubt. The 2023 Housing and Land Survey already counted 9,002,000 vacant dwellings, a record 13.8 percent of the housing stock, and the death line that feeds it has fifteen more years of climbing to do.

Scarcity is the oldest lever in property marketing, and in Japanese rural housing it is close to the least defensible one available. Whatever else is true about this market, the supply of cheap empty houses is the single variable most reliably going up. A platform, agent or consultant telling you to move before the akiya run out is arguing against the government's own projections.

There is a real scarcity here, and it sits somewhere else entirely. The 2023 survey found 9,002,000 vacant dwellings but only about 330,000 actually listed for sale, and the subset of those with clear multi-heir title, legal road frontage and rebuild rights, connected utilities, and a location where somebody will still want to live in 2045 is very much smaller again. That subset does not expand when a house in a dying valley goes dark. If anything it thins, because the towns that qualify are the towns everyone else has also identified.

This is the same distinction we drew in what listing count hides when you compare akiya platforms: volume is the easiest metric to grow and the least informative to a buyer. Demography is now doing the same thing to the akiya category as a whole. The pile gets bigger every year. The answer does not.

In most wealthy countries births and marriages have drifted apart, and the wedding count has become a lifestyle statistic with little bearing on family formation. Japan is the exception, and the number that makes it one is small and stubborn. Of the 686,173 births in 2024, 18,093 — 2.6 percent — occurred outside marriage. The other 97.4 percent were births within marriage.

That single figure is what makes the marriage count load-bearing rather than incidental. A country in which two or three births in a hundred happen outside marriage cannot offset a collapse in weddings with a rise in unmarried parenthood the way much of Europe has. In Japan the marriage series is a leading indicator of the birth series, with a lag of one to three years and a remainder that never arrives.

Existing households, singles, investors, and foreign buyers also buy houses. Marriage is the best official proxy for new family-house demand, not the only buyer. The thing that typically wants a family-sized house in a Japanese town, fifteen or twenty years from now, still has to be formed first.

The marriage series peaked in 1972 at 1,099,984 couples, the second baby boom pairing off. The 2025 approximate count is 489,119 — a second consecutive annual rise, after 485,092 in 2024, and still under half the peak. The mean age at first marriage in 2025 was 31.0 for husbands and 29.7 for wives. The people marrying are older, and there are fewer of them.

People still say they intend to marry. The institute's 16th National Fertility Survey, taken in 2021, found 81.4 percent of unmarried men and 84.3 percent of unmarried women aged 18 to 34 expected to marry eventually — down from 85.7 and 89.3 five years earlier, but still a large majority. Among those intending to marry, the average number of children hoped for was 1.82 for men and 1.79 for women, the first time the women's figure had fallen below two.

Read that honestly and it cuts both ways. The intention is still there, which is why nobody should write off Japanese household formation entirely. And the intended family is now smaller than replacement even among people who want one, which is why nobody should underwrite a rural four-bedroom house on the assumption that a Japanese family of five will want it in 2045.

Here is the part of the arithmetic that surprises most foreign buyers, and it is the most immediately useful thing on this page. Japan's population fell by 3,096,575 people between the 2020 and 2025 censuses. Over the same five years the number of households rose by about 1,294,000, to 57,124,507. Persons per household fell to 2.15, and it fell in all forty-seven prefectures.

This changes which houses have a future buyer, and it does not favour the classic cheap akiya. The Japanese dwelling that a growing number of small households want is compact, cheap to heat, close to a station or a clinic, and does not need a car for everything. A great deal of Japan's vacant stock is the opposite: large, cold, built for a three-generation family, and twenty minutes from anything by road.

It also explains why the depreciation logic bites so hard in the countryside. A building with no structure value in a market with no household growth is worth land value at best, and sometimes less than the cost of clearing it — the mechanics of which we set out in why Japanese houses lose their value.

Japan's total fertility rate was 1.14 in 2025, a new low, against a replacement level of roughly 2.1. But no house sits in the national average. Okinawa recorded 1.52 and Tokyo recorded 0.96, and both are official figures from the same release.

Now hold two facts together, because their combination is the single most counter-intuitive thing in Japanese property. Tokyo has the lowest fertility rate in the country and the strongest housing demand in the country. It was one of only two prefectures whose population grew between 2020 and 2025. New-build condominium prices in the 23 wards set records over the same period, which we unpacked in why the Tokyo condo record is about a different Japan.

Locally, migration overwhelms fertility. Tokyo does not need to produce residents; it imports them from the 1,558 municipalities that shrank. Which means the demographic question for a buyer is never “is Japan shrinking” — it is settled that it is — but whether this particular town is on the receiving end of internal migration or the supplying end of it. We worked through how to read that at municipal level in what Japan's foreign population cap debate means for property buyers, and the short version is that immigration is a metropolitan phenomenon and will not refill your valley.

On the long birth series there is one notch so sharp it looks like a printing error. 1966 was a Hinoeuma (丙午) year, the Fire Horse in the sixty-year sexagenary cycle, and a superstition held that girls born in it would make difficult wives. Births fell from 1,823,697 in 1965 to 1,360,974, a drop of about 25 percent in a single year, then rebounded to 1,935,647 in 1967 — overshooting the year before the fall.

2026 is the next Fire Horse year. There is no official forecast that the superstition will repeat, and this article is not offering one. What is worth noticing is why the 1966 notch was possible at all: it was a timing decision available to a married generation that was still having roughly two children each and could afford to wait a year. The generation of 2026 is smaller, marries at 31, and already sits at 1.14. A superstition can move a peak. It cannot invent a peak that is not there. The confirmed 2026 count arrives in 2027, and until then anybody selling you a Fire Horse baby bust is selling a story the data has not written.

We include this because the same test applies to almost every claim you will read about Japanese property. Narratives move prices at the level of a single listing, where one motivated buyer sets the number. They do not invent demand at the level of a town, where the number of households is a matter of public record. When someone tells you a region is about to be discovered, ask which of those two things they are describing.

None of this argues against buying a house in Japan. It argues against buying one on the wrong thesis. Almost every foreign buyer runs careful diligence on the purchase — the structure, the title, the renovation quote — and almost none of them run any diligence at all on the disposal. These are the five questions we would want answered before wiring money.

The fifth question resolves the other four. If the answer is yes, you are buying use value in a country that will sell it to you cheaply for the next twenty years, and the demography in this article is working for you rather than against you. If the answer is no, you need every one of the first four to come out well, and in 90.6 percent of Japanese municipalities at least one of them will not.

Ownership cost is the part that compounds either way. Fixed asset tax, insurance, maintenance and management run every year regardless of what the town does, and a non-resident owner generally needs a tax agent in Japan to file on their behalf. Our cost calculator covers the acquisition and first-year figures so the recurring numbers are visible before you commit rather than after.

Akiya X is a property qualification engine, not a brokerage and not a forecaster. The whole product exists because of the asymmetry described above: supply that grows on a schedule, demand that concentrates in a minority of places, and a buyer who has to tell those apart from the other side of the world in a second language.