Why Mortgage Rates Stay Near 7 Percent: The Monthly Intelligence Report, September 2026
Updated September 2026
The Federal Reserve does not set the thirty-year mortgage rate. Bond buyers do, and the newest US Treasury data shows foreign holdings of US government debt fell $72.1 billion in a single month, to $9.30 trillion, in the report released August 17, 2026 covering June. That is the pressure Ray Stendall of Stendall Realty Group has been tracking all year, and it is why the weekly mortgage survey printed 6.71 percent on September 3, 2026, its highest reading since June 2025, with no path to a 5-handle in sight.
All figures below are verified to primary or named sources as of the Friday, September 4, 2026 close unless dated otherwise. This is Issue 4 of the Monthly Intelligence Report. The distance between a strait in the Persian Gulf and the rate quoted to a homebuyer in Riverside County is shorter than almost anyone thinks, and this issue walks every link in that chain.
What actually sets your mortgage rate?
Most people watching interest rates are watching the Federal Reserve. The Fed sets one rate, overnight, for banks. It does not set the rate on a thirty-year mortgage. That rate is set by whoever buys long-term American debt, at whatever price they demand.
For three decades the answer included Japan. Japanese institutions borrowed at nearly zero at home and bought bonds abroad, much of it US Treasury debt. Japan is still the largest foreign holder, at roughly $1.1 trillion as of the most recent reported figures.
That arrangement is coming apart for two reasons at once. Japanese interest rates just hit levels last seen in 1996, so Japanese money finally has somewhere to go at home. And government figures reported by Kyodo show the Middle East supplied 94 percent of Japan’s crude in 2025, with 93 percent passing through the Strait of Hormuz, closed or restricted since February and the scene of American and Iranian fire through the first week of September.
A country paying more for oil while earning more at home is a country that sells foreign bonds. The newest Treasury data shows exactly that pattern.
The chain in one sentence: a strait in the Persian Gulf, an oil bill in Tokyo, a Treasury auction in New York, and the rate quoted to a homebuyer in Riverside County. It is why we think mortgage rates stay near 7 percent even in scenarios where the Fed eventually cuts.
One more thing before the details. The bond market settled an argument this week, quietly. Long-term yields pushed toward their 2026 highs while the market’s own inflation forecast did not move at all. That combination points away from inflation fear and straight at the question of who funds America, which is this issue’s argument, made by the market in public.
What changed this month?
The jobs headline was strong. The composition was not.
Payrolls rose 162,000 in August per the September 5, 2026 Employment Situation report, about three times expectations, unemployment held at 4.1 percent, and July’s initially reported loss of 23,000 was revised to a gain of 21,000. Read one layer down and 62 percent of the month came from two lines: restaurants and bars added 59,000, five times their twelve-month average, and local schools added 42,000 that BLS says largely reversed July’s drop.
Government plus healthcare produced 39 percent of the gain, lower-wage in-person services another 39 percent, and the knowledge economy shrank. Information lost 23,000, financial activities lost 11,000 for a second straight month, and professional and business services added just 10,000, a net 24,000 decline across the three sectors that pay the most. So what: the market read the headline and priced a rate hike. The composition is the displacement pattern this letter tracks, hiding inside a good number.
One strong month inside a shrinking year.
The same August 2026 report’s household survey counts 624,000 fewer Americans employed than a year ago, a labor force smaller by 973,000, and 1.2 million fewer prime-age workers between 25 and 54, against a population that grew 1.4 million. Participation ticked up to 61.6 percent in August but is down half a point since January by BLS’s own note.
The other side, printed because it cuts against us: August’s gain was full-time led, up 735,000 with part-time falling 223,000, and involuntary part-timers dropped 414,000. Still, 8.8 million people hold more than one job, 5.4 percent of everyone employed, and average hourly earnings grew 3.1 percent against 3.4 percent headline inflation. So what: a paycheck that trails prices is what an economy swapping software jobs for restaurant jobs produces in aggregate.
The AI-layoff streak broke.
Challenger, Gray and Christmas counted 52,881 announced cuts in August 2026, up 58 percent from July but down 38 percent from a year ago and the lowest August since 2022. For the first time since February, AI was not the top cited reason. Restructuring led with 16,173 cuts and AI fell to fourth at 3,462. Technology cut 6,103, its lowest month of 2026, though its year-to-date 155,126 runs 52 percent above last year. So what: the displacement story now shows up more clearly in the payroll tables, where information and finance keep shrinking, than in layoff announcements.
The Fed priced a hike, then argued with itself.
Governor Barr said on September 1, 2026 that the Fed should act decisively to raise rates if inflation is not moderating. Governor Waller countered on September 3 that he would support holding if inflation keeps easing. Futures odds of a September 16 increase swung from 66 percent Monday to 49 Thursday to 58 after Friday’s jobs report, per CME data on September 4, and the 2-year Treasury hit 4.377 percent, its highest since January 2025. So what: no cut is priced for 2026, and the decision now hangs on next week’s inflation data.
The war moved from the strait’s edge into Iran itself.
After the August 30 Larak Island strike, US forces hit Guard targets inside Iran on September 1, near Bandar Abbas, Qeshm and other coastal sites, citing fresh attempted attacks on shipping. Iranian missiles answered at US bases in Jordan, where eight were intercepted, and at US-linked sites across the Gulf through midweek. Vice President Vance said talks are off until attacks on shipping stop. The IMF PortWatch shipping tracker logged as few as six visible transits on August 30 against a pre-crisis baseline near 85 to 130. So what: oil rose about 9 percent on the week to a $96.28 Friday settle, and the diplomacy of late August now shares the stage with a widening exchange.
Energy stress set records into Labor Day.
Diesel reached $5.85 per the AAA reading on September 4, 2026, an all-time national record that passed even June 2022. Gasoline averaged $4.15, the first Labor Day above $4. The Strategic Petroleum Reserve fell to 286.6 million barrels in the EIA week ending August 28, 2026, about 40 percent of capacity and the lowest since November 1982. So what: the cost pipeline into 2027 goods prices is loaded no matter what crude does next week.
Bonds sat at extremes without new ones.
The 10-year closed September 3, 2026 at 4.77 percent, near its highest since January 2025. The 30-year closed at 5.25, under its mid-August peak of 5.34. Japan’s 10-year touched 3.00 percent, its first visit since 1996, then eased back toward 2.9 by Friday. So what: both ends of the cascade sit at or near multi-decade levels, and the driver is funding, not inflation fear.
Housing financing kept tightening.
The weekly mortgage survey printed 6.71 percent on September 3, 2026, its highest since June 2025, into contract demand already at its weakest since January per NAR. Private proxies stayed soft. ADP counted 38,000 August jobs, its slowest since January, and the ISM services survey showed employment contracting at 47.8 while its prices index hit 72.6. So what: costlier money is landing on frozen demand while services inflation pressure refuses to fade.
How accurate has this letter been?
We monitor 127 indicators monthly and have formally logged and graded 39 predictions. Twenty-five carry our calibration weighting, and across that group the calibration-weighted hit rate runs 46 to 49 percent. Across all 39, the raw hit rate is about 67 percent and direction accuracy near 85 percent. Three measurements over different groups, so each prints with its own count. We never blend them.
How we grade ourselves: before making a call, we write down what would prove it wrong and how big a miss would count. Easy, wide predictions count for less. Hard, specific ones count for more. Every call is graded when the data comes in, misses publish next to hits, and the full method is available to any subscriber who asks.
Where we have been wrong
The hole in the job market is smaller than we said. Through June and July we leaned on two employment reports that each cut prior months sharply, read that as proof the monthly survey was overcounting, and said the annual correction would confirm it. It came in at 79,000, against last year’s preliminary 911,000, later finalized at 898,000.
That is a real miss on the better instrument. The annual figure uses payroll tax records covering nearly every employer rather than a survey sample. When the accurate tool disagrees with the fast one, the accurate one wins.
What survives is narrower. Private employment was revised down 178,000 while government rose 99,000, and financial activities were cut further on top of the 121,000 that sector shed between May 2025 and the summer. The mechanism holds. The headline did not, and we will not pretend otherwise. This year’s 79,000 will itself move when finalized in February 2027.
Open predictions
| ID | Call | Resolves | Confidence | Tracking |
|---|---|---|---|---|
| M-260828-01 | Core PCE for November lands between 3.10 and 3.30 percent year over year | Dec 23, 2026 | 7/10 | New this issue |
| M-260820-01 | The average 30-year mortgage rate for the fourth quarter is at or above 6.50 percent | Dec 31, 2026 | 7/10 | Supportive and strengthening. Weekly survey 6.71 percent September 3, highest since June 2025, daily trackers touched 6.89 September 1 |
| M-260820-02 | The S and P 500 posts a drawdown of 10 percent or more, measured on weekly closes | Dec 31, 2027 | 6/10 | Not started. The index closed September 4 about 1.0 percent below its August 13 record |
| M-027 | Gold averages above $4,300 for the trailing quarter | Dec 31, 2026 | 7/10 | On track, margin thinning. Spot dipped under the bar intraday midweek near $4,282, then recovered to roughly $4,420 to $4,480 Friday, with December futures settling $4,476.60 September 4 |
| H-004 | The California median home price ends 2026 within 3 percent of a year earlier, a band of roughly $830,000 to $881,000 | Dec 31, 2026 | 8/10 | On track. County readings inside the band |
| Four calls resolve September 30 | Covering crude, the strait, third-quarter mortgage rates, and private credit | Sep 30, 2026 | Mixed | Two tracking toward misses, in opposite directions |
On M-260828-01, we are betting against a named forecaster. Capital Economics published in April 2026 that most of the forces pushing the Fed’s inflation gauge above the more familiar one should reverse by the turn of the year. If they are right, our range is too high. We have written down the date we find out.
The forecast matrix: 27 variables across seven categories
Read these tables with one distinction in mind. The Now column is sourced and dated, tracing to releases named in the sources line at the end of this post. The forecast, confidence and direction columns are our judgment, not measurements. We publish them because a forecast you cannot check is worthless, and we label them because a guess dressed as data is worse than worthless.
Monetary policy and rates
| Variable | Now | 90 days | 12 months | Conf. | Dir. | Mechanism |
|---|---|---|---|---|---|---|
| Fed policy rate | 3.50 to 3.75%, held since December | Held or one 0.25 increase | 3.50 to 4.00% | 7 | Up, hawkish | Warsh August 28, Barr September 1, decision September 16 |
| Odds of a September increase | 58% (CME, Sep 4), swung 66 to 49 to 58 across the week | Resolved Sep 16 | n/a | 5 | Up | Warsh and Barr hawkish, Waller the dovish voice |
| 10-year Treasury | 4.77% close Sep 3, about 4.79% Friday, near its highest since Jan 2025 | 4.55 to 4.95% | 4.30 to 5.00% | 6 | Up | Debt supply and who will hold it, not the Fed’s target |
| 30-year Treasury | 5.25% close Sep 3, little changed Friday, under the 5.34% mid-Aug intraday peak, 55 days settled above 5% through Sep 1, most since 2006 per Bloomberg | 5.05 to 5.50% | 4.90 to 5.60% | 6 | Up | Treasury buybacks Sep 9 to Nov 4 exist to hold this down |
| 30-year mortgage | 6.71% weekly survey (Sep 3), highest since June 2025, daily tracker 6.71% Sep 4 | 6.50 to 7.00% | 6.25 to 7.00% | 7 | Up | No path to a 5-handle in 2026 |
| Mortgage-to-Treasury gap | About 1.94 points (6.71 weekly survey over the 4.77 close) | 1.85 to 2.15 | 1.75 to 2.10 | 5 | Flat | The extra return demanded over lending to the government |
| Bank of Japan rate | 1.00%, Sept 17 to 18 meeting, an increase priced as near-certain | 1.25% | 1.25 to 1.75% | 7 | Up | The most underwatched input to US mortgage rates |
Inflation
| Variable | Now | 90 days | 12 months | Conf. | Dir. | Mechanism |
|---|---|---|---|---|---|---|
| Core PCE (the Fed’s gauge) | 3.3%, flat two months | 3.1 to 3.3% | 2.8 to 3.2% | 7 | Flat | See M-260828-01 |
| Core CPI (the familiar one) | 2.5% | 2.4 to 2.7% | 2.4 to 2.9% | 6 | Flat | Held down by housing, which counts half as much in PCE |
| Gap between the two | 0.8 points, backwards from normal | Stays above 0.5 | Narrows slowly | 5 | Widening | Capital Economics says it closes by year-end. M-260828-01 is our bet against that |
| Bond-market inflation forecast (10-year breakeven) | 2.35% Sep 4, from 2.31% late Aug, anchored while yields rose | 2.2 to 2.5% | 2.1 to 2.5% | 6 | Flat | Evidence the selloff is about funding, not inflation |
| Services inflation excluding housing | 3.9% | 3.6 to 4.0% | 3.2 to 3.8% | 6 | Flat | Must move before anything else does |
Energy and refining
| Variable | Now | 90 days | 12 months | Conf. | Dir. | Mechanism |
|---|---|---|---|---|---|---|
| Brent crude | $96.28 settle Sep 4, up about 9% on the week ($95.52 Sep 2) | $84 to 105 | $72 to 98 | 4 | Up | $93 to $88 to $96 in two weeks: diplomacy, then escalation |
| National gasoline | $4.15 (AAA, Sep 4), the first Labor Day above $4, record August average | $3.80 to 4.30 | $3.40 to 4.00 | 6 | Flat | Follows crude with a lag |
| National diesel | $5.85 all-time record (AAA, Sep 4), passing June 2022’s $5.816 | $5.20 to 6.00 | $4.60 to 5.80 | 5 | Flat | Refining capacity, not crude, sets this |
| Distillate stocks | About 10% below the five-year average (mid-August, EIA) | Below average | Rebuilding slowly | 5 | Down | The tightest point in the energy system |
| Refining margins | NOT READ THIS CYCLE from an institutional source. EIA’s outlook expects high margins through 2026 | High | Normalizing | 3 | Flat | Crude can fall while fuel stays expensive. Read as a pair |
| Strategic Petroleum Reserve | 286.6M barrels (week ending Aug 28, EIA), about 40% of capacity, lowest since Nov 1982 | 250 to 275M | 240 to 270M | 7 | Down | Roughly 103M of it cannot currently be pumped out |
Geopolitics and shipping
| Variable | Now | 90 days | 12 months | Conf. | Dir. | Mechanism |
|---|---|---|---|---|---|---|
| Strait of Hormuz | Selective closure since late February, US strikes Aug 30 and on Guard targets inside Iran Sep 1, Iranian missile response through Sep 3 | Contested, partial reopening possible | Reopened | 4 | Down, relief | Diplomacy continues, now under fire |
| Hormuz traffic | As few as 6 visible transits Aug 30 (IMF PortWatch) against a pre-crisis 85 to 130, dark ships mean true flow is higher | Escorted trickle | Recovering | 5 | Flat | Named tracker and naval figures below |
| Persian Gulf exports | About two-thirds of pre-war, per Goldman Sachs analysts Struyven and Grigsby | 60 to 90% | Near normal | 4 | Up | Rerouted and escorted, not normalized |
| Red Sea bypass | Houthi blockade of Saudi shipping since July 20, mainstream tanker traffic through Bab el-Mandeb down about 42% per Lloyd’s List | Contested | Contested | 5 | Down | Hormuz reopening without the bypass is not normalization |
| War state | Stalemate. Our primary branch at 40 percent | Stalemate | Stalemate or settlement | 5 | Flat | Neither escalation nor resolution is the base case |
Labor and AI
| Variable | Now | 90 days | 12 months | Conf. | Dir. | Mechanism |
|---|---|---|---|---|---|---|
| Monthly job change | Plus 162,000 Aug, 62% from restaurants and local schools, knowledge sectors down 24,000 | Plus 25,000 to plus 125,000 | Weak positive | 4 | Flat | Composition and participation, not the headline |
| Announced layoffs | 52,881 in Aug, down 38% year over year, lowest August since 2022 | Low | Low | 6 | Down | Displacement runs through hiring freezes, not firings |
| AI as cited layoff reason | Streak broke: led March to July, fourth in August (3,462). Restructuring led | Returns to top three | Structural driver | 5 | Flat | Tech cuts still up 52% year over year, and information payrolls still shrinking |
| Financial-sector employment | Down 11,000 in July and again in August, down 121,000 from its May 2025 peak | Continued decline | Continued decline | 6 | Down | The most AI-exposed white-collar cluster |
Housing
| Variable | Now | 90 days | 12 months | Conf. | Dir. | Mechanism |
|---|---|---|---|---|---|---|
| Existing home sales | 4.06M a year, median $434,100, the 37th straight month of annual gains, per NAR | 3.9 to 4.1M | 3.9 to 4.3M | 6 | Flat | Contracts lead closings by one to two months |
| Signed contracts | Weakest since January, about 30% below 2019 per NAR | Flat to lower | Flat | 6 | Down | A price and financing problem, not jobs |
| Attached homes (condos, townhomes) | Selling far slower than detached in our markets | Widening gap | Widening gap | 7 | Down | New lending rules land January 4, 2027 |
Household budget
| Variable | Now | 90 days | 12 months | Conf. | Dir. | Mechanism |
|---|---|---|---|---|---|---|
| Real wages | Wages up 3.1% year over year against 3.4% headline CPI, still negative | Slightly negative | Flat | 6 | Flat | Positive against core instead. The measure decides |
| Home insurance | US average up about 46% since 2021, projected near $3,057 for 2026 | Rising | Rising | 7 | Up | Every premium dollar shrinks the qualifying loan |
| California insurance | Projected up 16% in 2026, the largest of any state, per Insurify | Rising | Rising | 7 | Up | FAIR Plan rates rise 29.1% on October 15 |
How does a strait in the Persian Gulf reach a mortgage in San Diego / Orange or Riverside County?
Part one: the weekend the standoff turned hot again
For a week in late August, the strait story was a diplomacy story. Qatar’s prime minister was in Tehran on August 27, the third mediator in a week after Pakistan and Oman. Iran’s Revolutionary Guard had listed its reopening conditions: lift the naval blockade, pay war damages, remove sanctions. Iranian officials described an agreement with Oman on a temporary shipping corridor.
Then, on Sunday August 30, US forces struck two Iranian rocket-launch sites on Larak Island, in the strait itself. Central Command said Guard units there were preparing to launch rockets carrying sea mines into the shipping lanes. Iran acknowledged casualties, promised punishment, and on Monday launched missiles and drones at US-linked targets. Jordan’s military said it intercepted eight ballistic missiles, and the UAE said it downed an Iranian drone over its waters.
Then, on Monday afternoon Washington time, US forces began striking Guard targets inside Iran itself, near Bandar Abbas, Qeshm and other coastal sites, saying Guard units had again attempted attacks on commercial shipping. Exchanges continued into midweek, and Vice President Vance said talks are off until the attacks on shipping stop.
Two details matter more than the headlines. First, a senior Iranian source told Reuters the Oman corridor agreement was not finalized, directly contradicting the Guard spokesman who said an understanding had been reached. Treat the corridor as a proposal, not a fact.
Second, oil kept moving anyway. Saudi Arabia, the UAE, Kuwait and Iraq continued shipping, some tankers with transponders off. The US Navy says it has escorted roughly 1,500 commercial vessels carrying about 750 million barrels since the campaign began, turned back 75 ships, and disabled three. IMF PortWatch tracking gives the scale: visible transits have run from the single digits to the low teens a day against a pre-crisis baseline near 85 to 130, and ships sailing with transponders off mean the true count runs somewhat higher. Goldman Sachs analysts Daan Struyven and Yulia Zhestkova Grigsby put Gulf exports at 15 to 16 million barrels a day, about two-thirds of pre-war levels, up from a March 2026 low of 5 to 6 million.
The strait is neither open nor shut. It is a contested, escorted trickle, with diplomacy and shooting both live at once. Iran’s foreign minister has described the closure as selective, applying only to ships of countries it considers enemies, which is why traffic counts and export figures tell different stories depending on whose ships you count.
Part two: the reserve America already spent
The Strategic Petroleum Reserve is the emergency crude stockpile kept in salt caverns along the Gulf Coast. In the EIA week ending August 28, 2026 it held 286.6 million barrels, about 40 percent of its capacity, the lowest since November 1982. Roughly 125 million barrels have been drawn since the strait closed in February.
Now the part that changes what the number means. The Government Accountability Office found more than a quarter of the reserve unavailable for withdrawal as of December 2025, because of construction and cavern outages. Rapidan Energy put the unavailable portion at a minimum of 103 million barrels.
Subtract, and roughly 184 million barrels are actually usable. Two cautions, ours to own: the unavailable share was measured in December 2025 and applied here to an August 2026 level, and the comparison line is our own estimate, not a published government threshold, of an operating floor near 180 to 200 million barrels, below which cavern and pipeline integrity comes under strain.
Caveats stated: on a usable-barrels basis, the reserve looks to be at or near its floor now. The reserve is what a government reaches for when oil spikes. The United States has largely spent that option, days before a weekend that showed why it might want it back.
Part three: crude is noise, refining is signal
Watch crude over two weeks. Roughly $93 on August 20 and $88 on August 28 as diplomacy advanced, per TradingEconomics pricing, then a $96.28 settle on September 4 after the strikes, up about 9 percent in a week. Headlines moved it both directions before escalation resolved it upward.
Now watch diesel. It never blinked. It held near record levels through the entire diplomatic round trip, then set an all-time national record of $5.85 on September 4, 2026, passing even June 2022. California diesel sits above $7.
Crude oil is not fuel. It has to be refined into gasoline and diesel, and refining is a separate bottleneck with its own limits. Distillate inventories, the tank covering diesel and heating oil, sit about 10 percent below their five-year average, and in late August 2026 US refineries were running at 97.4 percent of capacity, per EIA, and still not rebuilding the cushion. We could not obtain a current refining-margin reading from an institutional source this cycle, and the matrix says so at low confidence. The government’s own energy outlook expects margins to stay high through 2026.
Diesel is the number that matters most. Nearly everything in a store arrived on something burning it. KPMG notes diesel costs spill into goods prices with a lag of one to four months. Our own estimate for the full pass-through into grocery prices is longer, two to four quarters. That is a 2027 grocery bill being set by an August 2026 refinery run rate.
Part four: how a strait reaches a mortgage, in six links
Link one. Hormuz closes or restricts. Roughly a quarter of the world’s seaborne oil normally passes through it.
Link two. Import-dependent countries burn strategic reserves and buy replacement oil at wartime prices. Japan is the extreme case, with 93 percent of its crude transiting that strait.
Link three. Paying more for oil damages a country’s trade balance. A nation that used to run a surplus and recycle it into foreign assets no longer has one to recycle.
Link four. What such countries sell is US Treasury debt, the most liquid thing they own.
Link five. Fewer foreign buyers means the US Treasury must offer higher yields to sell its debt.
Link six. Thirty-year mortgage rates are priced off long-term Treasury yields plus a spread. Higher yields, higher mortgage rates.
For months this chain was a hypothesis. The evidence has now arrived in official data. The Treasury’s report on foreign holdings, released August 17, 2026 covering June, showed total foreign holdings of US government debt fell $72.1 billion in one month, to $9.30 trillion, the latest in a string of monthly declines from February’s record. China’s holdings dropped 4 percent to $633.4 billion, the lowest since September 2008, per Reuters. Japan, still the largest holder, cut about 2.3 percent in the same month per market reporting of the Treasury data, to roughly $1.12 trillion. At August’s auctions, the share bought by foreign-type bidders ran consistently below average.
The US government is behaving like it knows. Expanded Treasury buybacks of 10-to-30-year debt, at least $4 billion per operation, run September 9 through November 4, 2026, built to support the long end. Investor Stanley Druckenmiller’s public verdict: the buybacks will lose a contest against the bond market.
Every link in this chain operates independently of the Federal Reserve. The Fed could cut its policy rate and mortgage rates could still rise, because the Fed sets the overnight rate for banks and the bond market sets everything past about two years.
Part five: Japan is running two engines at once
Japan would be the pressure point on oil alone. It is not on oil alone.
On September 1, 2026, Japan’s 10-year yield touched 3.00 percent, its first time there since 1996, and reached 3.015 the next day. The 5-year set a record and the 2-year a 31-year high. By Friday the 10-year had eased back toward 2.9 percent after a midweek auction of 30-year debt drew solid demand, and the yen rallied about 2 percent Thursday to around 156 per dollar. A Bank of Japan increase at the September 17 to 18 meeting is priced as close to certain, and Governor Ueda has hinted at one.
For thirty years the trade was simple: borrow yen at nearly nothing, buy something abroad that yields more. That only works while borrowing at home is free, and it no longer is. Japanese institutions now have three reasons pointing the same way. Yields at home, a weak yen to defend, and an oil bill payable in dollars.
And Japan is acting on exactly the pressure we describe. Prime Minister Sanae Takaichi set an end-of-August deadline for an energy resilience package, and Industry Minister Ryosei Akazawa announced it in the first week of September: cost-sharing for crude bought outside the Gulf, government-backed insurance, pipeline cooperation that bypasses the strait, strategic naphtha stockpiles, and a reaffirmed push toward 11 to 14 nuclear reactors. A government does not rebuild its energy procurement over a disruption it expects to end soon.
Even Washington is leaning in. At the G20 finance meeting, US Treasury Secretary Scott Bessent publicly predicted Japan’s government and central bank will do the things that lead to a stronger yen, and agreed markets were pricing exactly that.
So mid-September holds a live possibility. The Federal Reserve deciding on a hike September 16, and the Bank of Japan hiking September 18, in the same week, while the US 30-year sits near its highest yields since 2007.
Part six: the tell almost everyone missed
The single most useful fact in this issue hides in a number most people have never heard of. The bond market publishes its own inflation forecast every day. It is called the breakeven rate, read from the gap between the price of a regular Treasury bond and an identical one whose payments adjust with inflation. If investors feared inflation, that gap would widen.
While ten-year yields held near their highest since January 2025, the ten-year breakeven read 2.31 percent in late August 2026 and 2.35 percent on September 4. Four basis points of drift against a week of war headlines and a 9 percent oil move is a market whose inflation forecast is barely moving. Chairman Warsh said it himself at Jackson Hole: those expectations “are not pushed around easily, and right now they are well anchored.”
Read that carefully. If yields rise while the market’s own inflation forecast stays flat, the increase comes from investors demanding more compensation to lend to America for a long time. That is precisely what a market losing its largest foreign lender looks like. The bond market just endorsed the cascade, in its own handwriting.
It cuts one way against us too, and we will say so. Flat breakevens mean the market does not expect the oil shock to become a lasting inflation problem. If the market is right, our core-inflation prediction is more likely to miss low than high. That is the two-sided read, and it is on the record.
Why does America have two inflation numbers that disagree?
America publishes two inflation measures. The Consumer Price Index is the one on the news: 3.4 percent in July 2026, or 2.5 percent core, excluding food and energy. The PCE index is the one the Fed targets: 3.7 percent, or 3.3 percent core. Core CPI now sits eight tenths below core PCE. That is backwards. Since 2000, the Cleveland Fed puts core CPI about half a point higher on average.
Three things opened the gap. One has already expired, and the gap widened anyway.
Housing weight. Shelter is roughly a third of the CPI basket and a sixth of the PCE basket, because PCE also counts things consumers do not pay for directly, mostly employer-paid and government-paid healthcare. Housing inflation has nearly stopped, and it drags the familiar index down twice as hard as the Fed’s.
A data glitch that corrected itself. The 43-day government shutdown that began in October 2025 halted rent collection for one sample batch, artificially suppressing measured housing inflation through the winter. Researchers at Texas A and M documented the correction arriving in April 2026, when the missed increases entered the index at once. July’s data compares against pre-shutdown months, so the glitch is gone from the current numbers. The gap should have narrowed. It widened instead, to eight tenths from six.
Artificial intelligence. Software, medical care and financial services weigh more in PCE than in CPI. Analysis by the Belgian bank KBC traced accelerating software prices directly to the AI boom, while rent and vehicles, weighted more heavily in CPI, decelerated. KPMG found the same reaching consumers in July 2026: data-center demand pushed up memory-chip costs and consumer electronics prices jumped. One caveat against our own argument. KPMG expects a September quality adjustment for electronics, which lowers measured inflation when a product improves at the same price, to temper some of those gains.
So the AI investment boom that Governor Barr cited as economic strength on September 1 is, through software and chip prices, helping hold up the specific inflation measure the Fed uses to decide whether rates rise. Housing is producing the good inflation news and collecting none of the benefit.
The variable integration map
The Strait of Hormuz row drives Hormuz traffic and Gulf exports, which drive Brent, which drives gasoline but only weakly drives diesel, because refining margins and distillate stocks sit between them. Diesel feeds core PCE on a lag. The same closure drains the reserve and inflates Japan’s oil bill, which, together with the Bank of Japan rate, pressures the 10-year and 30-year Treasury rows through foreign selling, now confirmed in the Treasury’s own June data. Those set the 30-year mortgage through the mortgage-to-Treasury gap. High mortgage rates suppress signed contracts and existing sales, which suppress housing inside core CPI, which widens the gap between the two inflation measures, while the breakeven row certifies that none of this is inflation panic. The loop breaks in exactly two places: the strait genuinely reopening, or services inflation excluding housing finally falling from 3.9 percent.
What a subscriber does with this
Stop underwriting rate relief off CPI headlines. The friendly number is the one the Fed does not target.
Watch Tokyo the way you watch Washington. September 16 and September 18 are the pair to circle.
Treat diesel as a 2027 cost signal, not a 2026 nuisance. Record price, thin inventories.
Distinguish crude headlines from fuel prices. They separated visibly this month, in both directions.
Read the jobs report by its blocks, not its headline. Where the jobs came from, full-time against part-time, and who left the labor force. One quality full-time job beats three part-time jobs, and the headline cannot tell them apart.
Price the insurance line before you price the mortgage. Lenders cap the share of income spent on housing debt. Insurance counts inside that cap, so every premium dollar shrinks the loan a buyer qualifies for.
What are the scenarios from here?
Probabilities carried directly from our model, unrounded by design.
The Shock. 44 percent. Base case.
Inflation stays in the low threes on the Fed’s gauge. The Fed holds with an increase permanently on the table, or raises once. Long yields stay high on debt supply and thinning foreign demand rather than Fed policy, which is what the first week of September looked like. Mortgage rates hold in the high sixes. Housing volume stays near multi-decade lows without a price collapse, because sellers hold equity and are not forced to sell. Job losses run through non-hiring, appearing in participation rather than unemployment. The weekend’s exchanges are this branch’s texture: contained fire that keeps the strait restricted without tipping into general war.
The Grind. 37 percent.
Same rates, better jobs. The small benchmark correction moved probability here. Employment stabilizes, AI investment supports growth, and the economy absorbs high rates without breaking. This is roughly the economy Warsh and Barr described, and they have better data than we do. August’s jobs headline is this branch’s best single piece of evidence, and we printed it above without spin.
The Resolution Path. 19 percent.
The strait reopens on accepted terms, energy falls hard, core inflation follows with a lag, Japan’s pressure eases, and the Fed is cutting by mid-2027. Mortgage rates break below 6 percent into real pent-up demand. NAR’s chief economist put the counterfactual plainly this month: the housing market would be thriving, he said, if average rates returned near 6 percent.
Honesty about this branch after the weekend. The diplomacy is real, with three mediators in a week, Iran drafting conditions, and a proposed Oman corridor. It is also unfinalized by Iran’s own account, and it now shares the stage with airstrikes and retaliation. Two standing rules bind here. Reopening Hormuz is not sufficient while the Red Sea bypass sits under a blockade that has cut mainstream tanker traffic through Bab el-Mandeb by roughly 42 percent. Both routes have to clear. And we do not move scenario weights on a weekend of headlines. The scheduled review is this coming week, and any change prints in Issue 5 with reasons.
What would change our mind?
Published in advance so they cannot be revised afterward. Status through September 4, 2026: none of the four rewrite triggers has fired. The labor escalation signature printed its mirror image in August, and the cascade trigger meets its first data on September 16.
Toward The Resolution Path: commercial Hormuz traffic sustained above half of pre-war volume for three consecutive weeks, and the Red Sea blockade lifted. Core PCE at or below 3.0 percent twice running. The 30-year Treasury closing under 4.75 percent for a month.
Toward The Shock: a September increase followed by core PCE holding at or above 3.3 percent. The 30-year sustained above 5.50 percent. A monthly job loss worse than 100,000 with participation falling rather than unemployment rising.
Against the cascade thesis, this issue’s headline claim: foreign holdings of US Treasuries rising for two consecutive monthly reports while the strait stays closed, with Japan’s position stable or growing. The first test arrives with the next Treasury report in mid-September. We would rewrite this section and say so.
Against the inflation-gap finding: the gap between core CPI and core PCE narrowing below four tenths for two consecutive months, which would mean Capital Economics read it correctly and we did not.
Against our labor case entirely: February 2027’s final correction landing at or above the preliminary 79,000, combined with the share of 25-to-54-year-olds in the workforce recovering more than three tenths of a point.
What are we watching next?
September 10. August producer prices, the first inflation print of the week.
September 11. August consumer prices. Our threshold: housing inside the index above 0.3 percent monthly would mean the frozen-market effect is ending.
September 16. Two events, one day. The Fed decision at 2:00 p.m. Eastern with fresh member forecasts, priced near 58 percent odds of an increase as of Friday, September 4. And the Treasury’s foreign-holdings report covering July, the cascade’s first direct test. Our threshold there: a quarterly Japanese decline above $30 billion.
September 17 to 18. The Bank of Japan, with a hike priced as near-certain. Our threshold: any move past 1.25 percent, or Japan’s 10-year holding above 3.0 percent for a month. It touched that line September 1 and eased back by Friday, so the clock has not started.
October 2. The September jobs report. We will read the three blocks, the full-time share, and participation before the headline, as this issue just did for August.
October 15. California FAIR Plan rates rise 29.1 percent.
January 4, 2027. The condominium reserve rule below.
What does this mean for your market?
Tier-1 coastal metros
Scarcity markets are decoupling. Where inventory is tight and buyers finance less of the purchase, liquidity at the top of the ladder has improved from a year ago, the cleanest evidence the freeze is a financing problem rather than a demand problem. It runs both ways. This tier pulls away while rates sit near 7 percent and loses the advantage quickly if rates fall, because cheaper money brings the financed buyer back. For the current San Diego and Orange County reading on that split, see the September 2026 Southern California Housing Brief.
Tier-2 suburban markets
The rate-sensitive middle, where the housing loop does its work. Time on market is stretching while price growth flattens. Sellers hold equity, so the adjustment shows up in duration rather than price. One thing most reporting misses. Compare the sale price against the original asking price, not the final one. When those separate, the market is clearing through cuts and concessions, and published price indexes understate it.
Rate-sensitive commuter markets
The heaviest combined burden. Longer drives into four-dollar gasoline, the steepest insurance increases, and the most buyers who need the monthly payment to work. In the markets Ray Stendall tracks directly across San Diego, Orange and Riverside counties, the first negative year-over-year price readings appeared here, around one percent, led by attached homes. That is our own market data, not a national statistic.
Buyers and investors: three items
Condominiums are being repriced by lending rules on a published calendar. In March 2026, Fannie Mae issued Lender Letter LL-2026-03 with a matching Freddie Mac bulletin. Since August 3, 2026, the Limited Review shortcut is gone for projects over ten units, so every condo loan now gets a full examination of the association’s budget, reserves, delinquencies, insurance and litigation. Lending specialists say the shortcut covered roughly 40 percent of condo transactions. On January 4, 2027, minimum reserve funding rises from 10 to 15 percent of budgeted assessment income, triggered by the loan application date, not the closing date.
Two provisions cut the other way. An association can qualify instead with a reserve study, a professional estimate of future repair costs, if it is under three years old and funded at its highest recommended level, and the old rule blocking financing when investors owned over half a building is gone. A July 9, 2026 delay petition could still move the date. Sellers of attached homes face a known deadline. Buyers should read the reserve study before the listing.
Insurance is a pricing input now, not a closing cost. The US average premium is up roughly 46 percent since 2021, about three times general inflation, and premiums rose in 95 percent of US ZIP codes from 2021 to 2024, per the Consumer Federation of America. Insurify projects California up about 16 percent in 2026, the most of any state. The sharpest example is the California FAIR Plan, the state’s insurer of last resort. Regulators approved a 29.1 percent average increase effective October 15, 2026 on just under 700,000 policyholders, 696,562 as of June 2026, concentrated in the wildfire portion, so low-risk policyholders may see little change or even decreases. A property quoted in August that closes after October 15 carries a different premium than the one in the file. Pull the quote before the offer.
Underwrite against a rate path that does not improve. Our base case holds mortgage rates in the high sixes through 2027, and the weekly survey just printed 6.71 percent, its highest since June 2025. If a purchase, refinance, or hold-versus-sell decision only works at 5.5 percent, what you have is a wager on a strait reopening in the Persian Gulf, not a plan. If you want the local version of that math, the San Diego real estate page carries the current county picture.
The bottom line
The Federal Reserve does not set your mortgage rate. Bond buyers do, and the newest official data shows the largest foreign buyers stepping back. $72 billion out in one month, China at an eighteen-year low, Japan cutting while its own rates touch levels last seen in 1996 and its oil transits a strait where the shooting reached the Iranian mainland this week.
The week’s loudest number was 162,000 new jobs. The quieter numbers underneath it: 62 percent of the gain was restaurants and local schools, the three highest-paying sectors shrank by 24,000, and 624,000 fewer Americans are working than a year ago. The market heard the headline and priced a rate hike. Homebuyers got a 6.71 percent mortgage survey, the highest since June 2025, and a diesel record that will work through 2027 grocery prices while the Fed debates.
The bond market kept telling the same story in its clearest language. Yields near multi-decade levels, its own inflation forecast barely moving at 2.35 percent. Inflation fear does not look like that. Lenders stepping away from thirty-year American debt looks exactly like that, and the repricing lands on the number a mortgage is built on.
Four of our calls resolve September 30. The cascade meets its first direct test in the September 16 Treasury data. The results run in this space, hits and misses together.
Frequently Asked Questions: Mortgage Rates and the Treasury Cascade
Does the Federal Reserve set mortgage rates?
No. The Fed sets the overnight rate that banks charge each other. Thirty-year mortgage rates are priced off long-term Treasury yields plus a spread, so the Fed can cut while mortgage rates rise. The weekly survey printed 6.71 percent on September 3, 2026, its highest reading since June 2025.
Why would a conflict in the Persian Gulf change a Southern California mortgage rate?
Japan moves 93 percent of its crude through the Strait of Hormuz, per government figures reported by Kyodo. A larger oil bill drains the surplus Japan used to recycle into US Treasury debt. Fewer foreign buyers means higher yields at auction, and thirty-year mortgage rates are priced off those yields plus a spread.
Will mortgage rates drop below 6 percent in 2026?
Nothing in the current data points there. Ray Stendall, who has tracked North County San Diego for more than 20 years, holds a base case of high-6 percent rates through 2027, with the fourth-quarter average at or above 6.50 percent. A purchase that only works at 5.5 percent is a wager on the Strait of Hormuz reopening.
What is the breakeven rate and why does it matter to homebuyers?
The breakeven rate is the bond market’s own inflation forecast, read from the gap between a regular Treasury and an inflation-adjusted one. It sat at 2.35 percent on September 4, 2026 while yields ran near multi-decade highs. Flat breakevens beside rising yields point at funding pressure rather than inflation fear.
What changes for condominium financing on January 4, 2027?
Minimum reserve funding for condo associations rises from 10 to 15 percent of budgeted assessment income, triggered by the loan application date. Since August 3, 2026, the Limited Review shortcut is gone for projects over ten units. Buildings that miss the bar become harder to finance, which shrinks the buyer pool.
How much are California home insurance premiums rising in 2026?
Insurify projects California premiums up about 16 percent in 2026, the largest increase of any state. The California FAIR Plan’s approved 29.1 percent average increase takes effect October 15, 2026 on just under 700,000 policyholders. Lenders count insurance inside the qualifying ratio, so every premium dollar shrinks the loan amount.
If you want a specific read on how this rate environment lands on your San Diego, Orange County, or Riverside County home, I offer a private seller strategy review. No pitch, just an honest look at your options. Call or text 858-877-0484, or visit booking link. Ray Stendall, Stendall Realty Group, eXp Realty, DRE #02038682.
Sources and disclosure. This publication is for educational and informational purposes only. It is not investment, tax, or legal advice, and it is not a recommendation to buy or sell any security or property. Past performance and prior forecast accuracy do not predict future results. Our published hit rates are calculated under a fixed method in which prediction criteria and failure conditions are recorded before resolution, and figures are stated separately for the full prediction set and the calibration-weighted subset, with each count shown. The full method is available on request. Forecast columns in the matrix are the author’s judgment, labeled as such. Sources include the Bureau of Labor Statistics, Bureau of Economic Analysis, US Treasury, Energy Information Administration, Government Accountability Office, Federal Reserve, Federal Reserve Bank of Cleveland, CME Group, ADP, the Institute for Supply Management, IMF PortWatch, US Central Command, Freddie Mac, Fannie Mae, Mortgage News Daily, the National Association of Realtors, Challenger Gray and Christmas, AAA, Reuters, Kyodo News, Lloyd’s List Intelligence, Goldman Sachs, KPMG, KBC, Capital Economics, Rapidan Energy, Insurify, the Community Associations Institute, and Stendall Realty Group’s own market data where noted. Ray Stendall, DRE 02038682. Brokerage services offered through eXp Realty.