Research

Why a Starter Home Costs So Much

Two forces set the price of a first home, and they feed each other: the country stopped building small houses, and owners holding cheap mortgages stopped selling.

By the Euphoria team · 2026-07-22 · 10 min read

Nobody is selling

Most explanations of house prices treat the buyer as the entire story: what a buyer earns, what a buyer can borrow, what a buyer will agree to pay. But in the market for existing homes, which is where almost every first-time buyer actually shops, the supply is other households. Every listing is somebody's decision to move out. Since 2022, a very large number of those households have quietly decided not to.

The reason is a number on a document they signed years earlier, and it has almost nothing to do with the house.

• 554,000 housing starts in 2009, the fewest in any year since the series began in 1959
• 7% of new homes built in 2019 were under 1,400 square feet, down from 40 percent in 1980
• 6.95% average 30 year fixed mortgage rate in the week of September 17, 2026

There are two stories here, one about how few houses got built for a decade and a half, one about interest rates. Most coverage picks one. The interesting part is where they meet, because the rate story is what turns a general shortage into a shortage of one specific kind of house.

The decade the country stopped building

Housing starts count the new residential units on which construction began. The Census Bureau has tracked them monthly since 1959, so the series is long enough to show what normal looks like.

Normal was roughly 1.5 million a year. Then 2005 reached 2.07 million, which was a bubble, and 2009 fell to 554,000, the lowest annual figure in the history of the series. The collapse is the part everyone remembers.

The part that matters more is the flat stretch after it. From 2009 through 2015 the country built at rates it had not seen since the early 1980s, and the climb back was slow enough that 2019 came in at 1.29 million, which is roughly what 1990 produced.

Compare those two years honestly, though, and 1990 looks better. The 1990 census counted about 249 million people; by 2019 there were about 328 million. Per thousand residents, the country started about 4.8 homes in 1990 and about 3.9 in 2019. Building the same number of homes for 79 million more people is not a recovery.

A gap like that does not close in a year. Houses last about a century, so a slow decade sits inside the housing stock for decades. Nothing built in 2026 restores a home that was never built in 2012.

Why the small house vanished

Here is the part that usually gets skipped. The shortfall was not spread evenly across house sizes. Builders did not build fewer of everything. They built fewer cheap ones.

The share of new homes under 1,400 square feet, which is the usual working definition of a starter home, fell from about 40 percent of construction in the early 1980s to around 7 percent in 2019, according to Freddie Mac. The median new single-family home completed in 2025 was 2,142 square feet, per the Census Bureau.

This is not a story about taste. It is arithmetic, and it runs in three parts.

• The lot does not shrink. A finished lot carries a price: the land, the grading, the street, the water and sewer connections, the permit. If that comes to $120,000, it comes to $120,000 whether the house on top is 1,200 square feet or 2,600. Per square foot of finished house, expensive land is half as expensive on the bigger house.
• Most of a build is fixed cost. Plan review, impact fees, moving a crew to the site, the foundation, the roof structure, the heating and cooling system, and above all the kitchen and the bathrooms, which cost several times more per square foot than a bedroom does. Doubling the floor area comes nowhere near doubling those, so the marginal square foot is much cheaper than the average one.
• Minimum lot sizes put a floor under the house price. If the smallest lot a zoning code permits is a quarter of an acre, and a quarter acre in that town costs six figures, then no legally buildable house there is inexpensive, whatever the builder would have preferred. Setback rules and parking minimums work the same way.

Put those together and a builder holding one lot has an obvious answer.

A builder is not choosing between a small house and a large house. A builder is choosing between a small house and a large house on the same lot, and the lot does not get cheaper.

Buyers press the same way, because the larger house is cheaper per square foot, so anyone comparing on that basis stretches. The National Association of Home Builders has measured the resulting mismatch: about 26 percent of buyers said they wanted a home under 1,600 square feet, while only about 16 percent of single-family homes started in 2023 were that small.

The loan that will not travel

Now the second story, which turns on a feature of American mortgages that is genuinely strange by world standards.

A standard US mortgage is fixed for 30 years, can be paid off early without a penalty, and cannot be carried to a different house. Those three properties together produce something unusual. When market rates rise above the rate you already hold, your old loan becomes valuable, and the only way to keep it is to stay exactly where you are. Economists call this mortgage rate lock-in.

The scale is not subtle. Freddie Mac found that as of June 2023, nearly six in ten borrowers held a mortgage rate at or below 4 percent. The 30 year average in the week of September 17, 2026 was 6.95 percent, according to the Primary Mortgage Market Survey.

Consider one of those households. On a $300,000 loan at 3 percent, monthly principal and interest is about $1,265. The same $300,000 at 7 percent is about $1,996. Moving to an identical house at an identical price would cost that family roughly $731 more every month in exchange for nothing. So they stay, and their house never becomes a listing.

Researchers at the Federal Housing Finance Agency put numbers on the aggregate. In working paper 24-03 they estimated that for every percentage point by which market rates exceeded a borrower's own rate, the probability the home sold fell by 18.1 percent, and that lock-in prevented about 1.33 million sales between the second quarter of 2022 and the fourth quarter of 2023.

Where the two stories collide

Rising rates are supposed to cool a housing market. A buyer who can afford less bids less, and prices soften. That mechanism is real.

What most coverage misses is that the same rate increase also removes sellers, and the seller effect can be larger. The same paper estimated that the supply reduction from lock-in pushed home prices up by 5.7 percent, outweighing the direct effect of higher rates, which pushed them down by 3.3 percent.

Read that slowly, because it inverts the usual intuition. Higher rates made homes more expensive rather than cheaper, by freezing the people who would otherwise have sold. Both effects trace to the same cause, and the one nobody itemizes won.

One further step follows, and here the line between measured and inferred matters. The households locked in most deeply bought or refinanced in 2020 and 2021, at record low rates. Those buyers skew toward people early in their owning lives, in smaller and cheaper homes, whose next move would have been a trade up. The listings that never appeared were therefore weighted toward exactly the modest homes a first-time buyer wants. That reasoning is sound, but it is reasoning rather than a measurement, and it should be read as such.

The result shows up in who ends up buying. The National Association of Realtors reported the first-time buyer share of purchases falling to 21 percent, a record low in its survey.

The payment, worked out

Everyone shopping watches the price. The rate is doing more of the work. Take the same $300,000 loan on a 30 year fixed schedule and change nothing but the rate.

The payment climbs from about $1,265 at 3 percent to about $1,996 at 7 percent. Total interest across the full 30 years goes from roughly $155,000 to roughly $419,000, on the identical loan for the identical house.

Now run it the other way, which is the comparison that actually matters to a buyer. At 7 percent, how far would the price have to fall to produce the payment you would have had at 5 percent? The 5 percent payment is about $1,610. Reaching it at 7 percent takes a loan of about $242,000. That is roughly $58,000 off a $300,000 loan, a price cut of about 19 percent.

So a two point move in rates is worth a nineteen percent discount on the house. Discounts that large are rare. Two point rate moves happen inside eighteen months.

What nobody actually knows yet

Three questions will decide how this resolves, and none of them is settled.

• Whether lock-in melts or snaps. It decays as rates drift toward the average rate on loans already outstanding, and as households move for reasons that overwhelm money: a birth, a death, a job in another state. Whether that release is gradual or arrives as a wave of listings is unknown, and the two point in opposite directions for prices.
• Whether zoning changes actually produce small houses. Several states have rewritten minimum lot sizes and legalized more units per parcel, recently enough that there is very little evidence yet on how much gets built, how fast, or at what price.
• Whether builders would return to the small house. The demand gap the builders' own association measured suggests buyers want them. Whether lot costs, construction costs, and the supply of skilled labor permit them at a profit is the question that decides.

One more piece of the monthly payment goes missing from most affordability talk. Property taxes and homeowners insurance ride on top of principal and interest, and in some states insurance has become a large and fast-moving share of the bill.

Reading the next housing headline

Two questions will keep you oriented when this story comes around again, which it will.

Ask whether the figure covers new homes or existing homes. Those are different markets with different medians, and the median price of a new home can fall simply because builders shifted toward smaller houses. A lower median is not automatically a cheaper house.

Ask where the supply in the story comes from. If the answer is existing owners, the number that governs is not what buyers can pay. It is what sellers already owe, and at what rate.

Why this matters for you

Stop treating a monthly payment as one number. It is the output of a formula whose inputs move one at a time, and once you can see which input is doing the work, most housing commentary becomes easy to grade.

Euphoria's lessons let you change the rate, the price, and the term separately and watch the payment answer, so the next time somebody announces that the market has turned affordable, you already know which input they moved.