On Friday, Israel's Central Bureau of Statistics published its monthly consumer price index, and alongside it, as it does every month, the housing price index. The index rose 0.1%, and the annual decline eased from 2.0% to 1.5%. Every financial site in the country ran almost the same headline: the fall in home prices has stopped. There's a floor. The market is stabilizing.
I want to stop on that number, because it measures something quite different from what most of us think it measures. And this is not an Israeli curiosity. The same mechanism is running right now in the American new-home market and in the American car market, which is where this letter will end up.
Before we go on, it's worth knowing which number we're talking about, because the CBS publishes two and they don't say the same thing. The housing price index is built from real transactions reported to the Tax Authority, and it is quality-adjusted: the bureau takes the price written in the contract and strips out the differences between the homes that happened to sell — locality, the socio-economic level of the area, number of rooms, floor area, and the age of the home, which in practice is the difference between a new home and a resale. Without that adjustment, a month in which more large homes or more new homes happened to sell would look like a month in which prices went up, even if no price moved at all. It's a good method, and it's the standard one around the world.
Alongside it the CBS also publishes a table of average prices by home size and city, and that one is a plain average with no quality adjustment at all. That table showed a national average of 2.43 million shekels in the second quarter, a rise of 7.9% over the year. Same bureau, same day, and one number says minus 1.5% while the other says plus 7.9%. The entire gap sits in the method and not in the market, and someone reading a headline usually has no idea which of the two it used. When someone tells you what happened to home prices, the first question isn't "by how much" but "according to which number."
The quality-adjusted index is the better of the two, and it still has one blind spot, which is what this whole letter is about. It reads the price written in the contract, so it works beautifully on one condition — that the price in the contract is actually the price that was paid. For the past two years, that has not been the case.
The discount that comes in through the back door
Research by the head of financial analysis at Reichman University, published in TheMarker in April, measured what the financing incentives that developers hand buyers are actually worth. The result: the average incentive is equivalent to a discount of 12.7% off the price of the home, which on a 3 million shekel apartment comes to roughly 380,000 shekels. According to the same research, 38.7% of new-home purchases in the preceding year included an incentive of this kind, so this is no longer the edge case of one developer stuck with a project. It is how a large part of the market works. How does it happen in practice? We'll get to that, and the most common route by far is a mechanism called 20/80.
Now watch what happens. The developer gives the buyer a discount worth 380,000 shekels, and the buyer receives every shekel of it, but the contract filed with the Tax Authority shows the full price — because the discount wasn't given on the price, it was given in the payment terms. The bureau sees the full price. The index sees the full price. And when enough deals like that pile on top of one another, you get exactly what we saw on Friday: the index reports that prices have steadied, while the price people are actually paying keeps falling. There's a fingerprint of this inside the data itself. In that same release, new-home prices fell 2.0% over the year against 1.5% for the market as a whole. The segment where the incentives are concentrated is the one falling faster, and that's before you deduct the incentive from the price.
Which means the floor everyone reported is, at least in part, not something happening in the market but something happening in the measurement. In the last letter we talked about how the first release of an economic number is really a draft, and how it gets revised afterwards. This is a different animal entirely: this number will not be revised in two months, because as far as it is concerned nothing about it is wrong. It is simply measuring one thing while all of us read it as another.
How a 20/80 deal actually works
Let's take the instrument apart, because without it there's no seeing where the discount comes from. In an ordinary purchase from a developer you pay according to the pace of construction — ten percent now, another twenty when the frame goes up, and so on until you get the keys. That means you take out a mortgage early and pay interest on it through the entire construction period, three to four years in which you are paying for a home you don't live in yet, while you go on paying rent on the place where you do live.
In a 20/80 deal you pay 20% at signing, and that's it. The remaining 80% comes due only on the day you get the keys, and in the meantime the developer is the one carrying the cost of financing the project. In real money that is a discount worth hundreds of thousands: all the interest you didn't pay for three years, plus the double housing cost you were spared. The developer calls it a financing incentive, the Tax Authority sees a contract for the full amount, and those are simply two descriptions of the very same thing.
Why would he do that, instead of just cutting the price?
The answer is less sinister than it sounds. A developer who cuts his price list by 10% isn't cutting the price of one apartment, he is repricing the entire project: the units already sold suddenly look like a bad deal and the people who bought two months ago pick up the phone, the bank financing the project looks at the collateral it took and finds it worth less, and the inventory still sitting on his balance sheet, the inventory he borrowed against, is worth less too. A single price cut hits all of those at once, and in a market where cash flow is already stretched, that is precisely the move that can break the camel's back — in this case, the developer's financial back.
A discount routed through the payment terms, by contrast, hands the buyer exactly the same money and touches none of them. From the developer's side it's a perfectly sensible business decision, and I'm not sure I would decide differently in his place. The side effect is that the official number, the one all of us look at to understand what's happening in the market, stops describing reality.
And before you decide this is someone else's problem
American builders do the same thing, and they barely bother to hide it. The National Association of Home Builders reported in March that 64% of builders were offering sales incentives — mortgage rate buydowns, closing-cost credits, upgrade allowances. The most common of them is the rate buydown, where the builder pays the lender up front so your mortgage rate starts lower than the market rate.
The logic is identical to the Israeli one, and American analysts say it out loud: the builder keeps the sticker price high in order to protect the value of the inventory he hasn't sold yet, and hands you the discount somewhere the price sheet doesn't record. You get a real benefit worth tens of thousands of dollars, the recorded sale price stays where it was, and every index built on recorded sale prices reports a market that is holding firm.
Where else we've seen this movie
If you've ever walked onto a car lot, this will sound familiar, because carmakers have been doing it for decades. A carmaker almost never cuts the sticker price, for exactly the same reason — a cut like that wipes value off every car sitting on every lot and infuriates everyone who bought last month — and instead it offers zero-percent financing. This August, Volkswagen, Ford, Dodge and Jeep are all offering 0% for up to 72 months, at a time when an ordinary car loan runs around 7%. On a $40,000 car that financing is worth about $4,000. A real discount, in cash, that appears nowhere on the sticker.
Take the arithmetic of an American car buyer over the past decade, because it shows where this mechanism leads in the end. The average transaction price of a new car went from roughly $35,000 to almost $49,000, close to 40% in ten years, and wages did not rise at that pace. Had loan terms stayed where they were, the monthly payment would have jumped by the same proportion and a great many people would simply have left the market. What happened instead is that the loan got stretched: nearly a quarter of new-car buyers in the second quarter signed for 84 months or more, which is seven years, on a car. The average monthly payment did reach a record $777, but that is far less than it would have been without the stretch. The price didn't come down, the ability to pay it didn't go up, and the gap was closed with time.
What's in your pocket
If you're buying new construction. The incentive is worth real money, and on average it's worth a great deal of it. But two buyers who signed at exactly the same price, one with an incentive and one without, did not make remotely the same deal — and nobody looking at the two contracts could tell them apart. The price on its own has stopped telling you much of anything, and the question "what does it cost" now has to include "and when do I pay."
If you own a home and you're watching the index. The index tracks contract prices, not what your home would fetch today. In a market where discounts are leaking into the payment terms, that reading leans high, so the home you carry in your head at a certain value is probably worth somewhat less. The gap does no harm to anyone as long as nobody touches it, and it closes the moment you actually try to sell.
If your mortgage rate was bought down for you. This one has a date on it, and it's worth knowing. In the common structure, your rate is two points below the note rate in year one, one point below in year two, and from year three you pay the full rate. The payment you budgeted around is not the payment you will be making in three years, and the home securing it may be worth less than the price on the contract. In Israel the same clock runs on the 20/80 deals: the 80% comes due at delivery, three to four years after signing, which means deals signed in 2023 and 2024 are landing now and over the next two years.
What's already happening on the ground
Israel's chief economist at the Ministry of Finance published a figure this week that illustrates that last point precisely: 1,821 purchases of new homes were cancelled, a jump of 41% against the January review. It's worth being accurate about how this works, because in Israel there is no real clause in the contract letting a buyer walk away. In practice, though, developers let most buyers out quietly, because they don't want lawsuits, buyer associations and noise in the press. So someone who put down 20% and discovers the financing doesn't work finds that the door out is a good deal wider than the contract says.
At the same time, sales were reported this month to have jumped 50% in June, and that number is also correct. It is measured against June 2025, a month of military operations when the market was almost entirely frozen, so nearly anything that came after it looks like a surge. In the free market, excluding subsidized housing, 7,550 deals were done in June — against June 2024, that is zero change.
Sitting above both of those is the inventory: 84,280 new homes waiting for a buyer at the end of June, roughly 26 months of supply. And that number, large as it is, still understates things — the bureau drops a home from the count 15 months after its construction is finished, even if it was never sold. It wasn't sold, nobody moved in, it simply stopped being counted. Industry estimates put the real inventory above 90,000 homes, and that is an estimate rather than an official figure, but the direction is clear enough. As long as that cushion is lying there, any recovery in demand gets absorbed into it long before it reaches prices.
Which brings us to two scenarios
This balloon — a price list that won't move, discounts hidden in the payment terms, buyers stretched across seven years of payments, and more than two years of unsold inventory — is not going to stay inflated forever. At some point people stop playing by these rules, either because the money runs out or because the patience does. The only question is how the air comes out.
The first scenario is 2008. The air comes out with a bang, and it can start from either end. At one end a large developer falls, and the banks discover the collateral they took is worth less than they booked. At the other end — and this is in fact what happened in 2008 — a financial institution is the one that falls first, a bank or a finance company, and then the whole market stops lending to developers at once, banks start pulling deposits from one another, and a credit squeeze forms across the system. From there it rolls to exactly the same place: projects freeze mid-construction, subcontractors and suppliers go down after them, and people lose their jobs. The damage doesn't stay in the sector either, it reaches the stock market, interest rates, and the pockets of people who never bought a home from a developer and never planned to.
The second scenario is a long, slow leak. The air comes out over years, the price list stays where it is while the real price keeps quietly sliding. Developers turn into companies whose only function is servicing debt rather than building, buyers keep waiting on the sidelines because nothing gives them a reason to hurry, and transactions stay at low levels year after year. No drama, no headlines, not one single day you can point at and say "it happened here" — but no market either.
What I think will happen, and why
Before I answer, it's worth looking at one number from the American car market, because it settles the argument almost on its own. The 60-day delinquency rate on car loans to borrowers with weak credit stands today at 6.9%, the worst reading since 1994, and higher than the roughly 5% peak recorded at the height of the 2008 crisis itself. An entire market, built end to end on stretched financing terms, is today in deeper distress than it was in 2008 — and there has been no bang. No collapse, no contagion, no wave of unemployment. It is simply bleeding slowly, and has been for years, and most people have no idea it's happening.
From that you can draw the rule, and it isn't the intuitive one: the size of the balloon doesn't determine the shape of the pop. What determines it is who holds the risk, and whether they are forced to sell. In 2008 four conditions held at once, and that is what made it a bang. The risk was packaged and sold onward until nobody knew who was holding it; whoever was holding it was leveraged, meaning working mostly with borrowed money, and funded itself with short-term loans that constantly had to be renewed, so it had to sell at precisely the moment of the fall; the accounting rules forced it to recognize the loss immediately; and millions of households were the borrowers, and they went down almost simultaneously.
In the Israeli housing market, three of those four conditions simply don't hold. First, there is no securitization here — that's the trick where a bank takes thousands of mortgages, packages them into a bond and sells it to investors, passing the risk along. In America in 2007 that was an enormous machine, and it is exactly why, when the thing blew up, nobody knew who was holding the toxic package and nobody was willing to lend to anybody. In Israel the securitization bill has been crawling through the Knesset for some two decades and has passed a first reading, but there is no market operating at any real scale, and this credit sits on the balance sheets of the five large banking groups, around 39% of all business credit in the economy. That sounds more frightening. In practice it is more reassuring, because everyone knows precisely who is holding it, and he is supervised.
Second, those banks are under no pressure. They are among the most profitable in the world, and a bank earning like that absorbs credit losses out of current earnings rather than out of capital, which means it isn't forced to sell anything on a Tuesday morning in order to survive. There's also a difference in the timing of supervision that is easy to miss: in 2007 the American regulator arrived after the fall, whereas here the Bank of Israel has already required the banks to set aside additional capital against leveraged land financing, and capped subsidized balloon loans at 10% of the mortgages taken out each month. You can argue about whether it's enough, but it happened before the event rather than after it.
Third, there is no daily pricing here, and that point sounds technical and changes everything. A security traded on an exchange gets a new price every morning, and that price is what forces somebody to act — a fund that has to meet a capital ratio sells at the market price, whatever it happens to be. A finished apartment standing empty at a developer doesn't get a new price every morning; it simply stands there, and that standing shows up in no report as a loss. That is precisely why the adjustment in this market happens through quantity rather than through price, and it explains how a nearly frozen market and a nearly flat price index can coexist in the same month without contradicting each other.
The one condition that does hold is that the bonds of the real estate companies trade on the exchange, get priced daily, and sit inside the pension savings of a great many of you. That is the only fast channel there is here, and two real pressures are building beside it. The Bank of Israel reports that in 44% of the banking system's exposure to projects, the pace of construction is running ahead of the pace of sales, meaning they are building faster than they are selling and financing the gap with debt that only grows. And the 20/80 deals, as we said, carry a built-in expiry date that is starting to land right now.
So in my estimation, this will be a long, slow leak and not a bang. The risk here is visible, concentrated in supervised and profitable institutions, and nobody is forced to sell. And I want to say this plainly, because I'm not sure it counts as good news: a bang at least clears the market and brings prices back to where people can buy, whereas a long leak stretches that out over a decade and leaves an entire generation standing on the sidelines, waiting. Japan did exactly this for twenty years, and nobody there remembers one dramatic day.
What could turn it into a bang is one specific combination, and it's worth keeping an eye on: a large developer financed with traded bonds stops meeting its debt payments, at precisely the time when the banks are already required to hold extra capital against the sector and are therefore less willing to extend credit. That is what turns visible risk into a forced sale, and a forced sale is the difference between a leak and a bang. In the meantime, the next time you see a headline saying home prices have stopped falling, it's worth remembering that the index that measured it never saw the discount in the first place — which makes it the last one that will tell you when this is actually over.
We shall see.
See you in next week's letter, or during the week — if I can't help myself until then...