Every issue of this letter has been about a gap. This one is about the widest gap in the economy as of early September, the one between how expensive life has actually become and what the official numbers are willing to admit.
Call it what it is. Forget the soft landing. This is a vibecession, an economy where the aggregate prints look survivable while the lived experience underneath them keeps getting harder. The cost of money reset higher and it is not coming back this cycle, with the 10-year at 4.73% and a 30-year mortgage at 6.66% in Freddie Mac's August 27 survey. Inflation still sits a full point above target while real wages for the people who fill my buildings run negative, and the labor market everyone called strong rolled over in public in the August 7 jobs report. The household ledger agrees: the share of credit-card balances 90 days late reached 12.8% in the second quarter, the highest since 2011, and the New York Fed notes the level is inflated by old charged-off debt even as it keeps climbing. Then the new Fed chair told you where he stands. Kevin Warsh used his first Jackson Hole keynote, one hundred days into the job, to name inflation the paramount concern and to retire forward guidance altogether, calling the market-and-Fed feedback loop a hall of mirrors. The line he chose: "At the moment of truth, there are either reasons or results." Futures flipped from two-thirds odds of a hold to favoring a hike, with a cut priced at zero. The part worth watching: the long end rallied on the hawkishness. The 30-year Treasury, which touched 5.31% in mid-August, its highest since 2007, closed at 5.22% after the speech. The bond market paid for inflation-fighting credibility. If that trade holds, the long rates my business lives on top out once the market believes the Fed will finish the job. One caveat arrived this summer from an unexpected borrower: Bloomberg, citing Nomura, put AI data-center debt issuance near a quarter of net new Treasury coupon supply, fresh structural competition for the same long-end capital my mortgages price off.
I called this a year ago. I wrote the original vibecession report in late 2025, when the aggregates still looked fine and the thesis was easy to wave off. A year later the household-distress thesis stands as written, and I have revised exactly one piece: the Sun Belt supply hangover is clearing faster than I expected. The original is here: Year(s) of the Vibecession — Original (2025).
The strain is no longer hiding in survey data. It is showing up at the courthouse steps. In early August I pulled the September foreclosure postings for DFW: 579 homes across Dallas, Tarrant, Collin, and Denton counties. Here is what makes this cycle different from 2008, and it is stranger than a crash. 86% of these homeowners have positive equity, a median cushion of about $124,000. They have equity, and they can no longer make the payment. At the GFC's peak, one mortgaged home in four in America was underwater and foreclosure was the exit because there was nothing left to sell. Today only 3.2% of mortgages are seriously underwater and filings run about 80% below 2008. 2008 broke balance sheets, the homes worth less than the loans. 2026 breaks cash flow, the homes fine and the households failing. A balance-sheet crisis crashes prices. A cash-flow crisis makes renters. Every one of those 579 households still needs somewhere to live, and the somewhere is a rental.
The fiscal backdrop says this pressure is structural, and I will spare you the full chart stack this issue. The short version: federal debt passes the World War II record around 2030 on the CBO's own optimistic baseline, and every honest exit from the debt maze, the hawkish one and the dovish one, ends with the dollar buying less. I do not need to win that debate. Both roads lead to the same trade: own the hard asset, finance it with long fixed-rate debt, and let either mechanism do the work.
Run the same dollar through the alternatives and watch what each one pays you to wait. A 10-year Treasury pays about 4.7% nominal, and on the debasement road that paper is the thing losing value. Gold is the pure monetary hedge and pays nothing while you hold it. Commodities respond fastest to a shock and yield nothing. Raw land has perpetual demand, no liquidity, and no income. Collectibles are scarcity with no cash flow and no financing market. Workforce multifamily is the one asset on that list that is scarce, throws off monthly cash, carries leverage you can fix, and shelters the income with depreciation the whole way through. Income-producing real estate is the only hard asset that hedges the debasement and pays you rent while it does it.
The demand side of the hedge is just as structural. Shelter is the last line item a household cuts, and at the workforce price point the demand runs counter-cyclical: in a downturn, would-be buyers stay renters and higher-rent households trade down into exactly this product. With a 30-year mortgage at 6.66% (go figure), renting a workforce two-bedroom in my markets runs hundreds of dollars a month cheaper than owning the median home in the same ring, so the renter pool is captive until rates fall materially. My tenants work in healthcare, education, and logistics, the sectors that keep paying through a recession. The rent check holds because the paycheck behind it does.
That is the vibecession. Now here is the part people find uncomfortable.
I invest in this. I buy the buildings these strained households live in, and I do it to earn a return. For a lot of people that sentence ends the conversation. Landlord, extraction, done. I understand the reflex. I also think it is wrong, and the numbers are the reason. The fifty largest apartment owners in America hold about 11% of the market combined, a smaller share than GM, Toyota, or Ford each hold of the car market. The rent problem is a supply-and-cost problem wearing a villain costume. Done right, the investor gets paid and the rent stays payable, and those turn out to be the same trade.
My return comes from two places, and both sit outside the tenant's rent check.
The first is yield. I buy the income a building produces today, priced so the year-one number clears my hurdle with zero growth story attached. The sellers who can meet that price are the ones in trouble, and the buildings are rarely the reason. They bought floating-rate debt in 2021, the rate caps expired, and the refinance math fails. The forced sales come out of the bridge and CMBS book, where delinquency hit 7.69% in July, against an agency book still healthy at half a percent. That is a fifteen-to-one gap on the same asset class, and it means the distress arrives as a conveyor belt of motivated sellers rather than a flood that crashes every price. The industry's own mid-year polls sketch the standoff in one line: operators say fundamentals turned, and capital says pricing is not there yet. NMHC's tightness index crossed 50 for the first time in a year in the same month its members marked every capital index down. Closed deals split the difference: MSCI's apartment cap rate printed 5.79% in the second quarter, the highest since 2015. Sellers anchored to 2021, debt priced for 2026, and a frozen middle in between. That freeze is the opportunity, because a patient buyer picks through the conveyor belt one deal at a time.
The basis backs the yield up. In Austin the market price sits near $220,000 a unit, a fifth off its 2022 peak. Across DFW the per-unit value is still flat to falling on a year-over-year read while the cost to build never came down. On the deals I have walked this year the arithmetic keeps landing in the same place: existing workforce product pricing at a 30 to 45% discount to replacement cost. Camden just voted with $1.6 billion, selling out of California and redeploying into seven Sun Belt markets. For me the discount to replacement cost is the margin of safety: the price has to clear the yield hurdle on today's income first, and then the basis protects the position while it pays.
The second is operations. The old value-add playbook, spend fifteen thousand a unit on granite and charge $150 more, is finished. A tenant whose raise already went to groceries and insurance will not fund your countertops. The edge moved to running the building well: keeping residents, controlling the expense line, collecting the ancillary income a tired operator leaves on the table. That is a management job, and it is the reason both sides can win. The institutions are landing on the same playbook: Elie Rieder, founder of Castle Lanterra, wrote in August that creating value in multifamily is not always about spending more, and that disciplined ownership means knowing when the economics actually support a renovation. His alternative list reads like mine: onsite teams, resident retention, the everyday experience. An operator who needs 5% annual rent hikes needs the tenant to lose. My number works without that.
My home market splits in two, and the split is the whole story. I grew up on the affluent side of Dallas, so I know both halves firsthand. In the prime corridors the typical renter earns about $86,000 and pays close to $1,900 in housing built in the mid-1990s. The Dallas I buy in is a different city inside the same metro: my renters earn a median around $54,000 in buildings from the early 1980s renting near $1,300. Two rental markets that barely touch. The concession war everyone points to, the Class A oversupply, is happening in the first Dallas. Almost nobody is building new workforce housing in the second one. The markets touch at one seam: when Class A piles on concessions, my best earners can trade up, and that caps my rent upside without touching my core demand. A new Uptown tower at $1,900 with two months free still costs more than a $54,000 household can carry. Durable demand, no new supply, and a built-in ceiling on how hard anyone should push it.
CoStar's own analysts just drew the same split with different tools. Their August vintage study found DFW's recovery concentrating almost entirely in newer product: buildings from the 1980s and earlier have lost roughly five percent of their occupied units since 2019 while 2010s product gained ten. Rent growth splits into a barbell of pain, the 1980s stock down 2.1% and the brand-new 2020s stock down 2.3%, for opposite reasons: old stock trades pricing power to defend occupancy while new stock fights a lease-up war. And the metro is turning underneath it all. DFW absorption ran 25% higher year over year, the strongest quarter since 2021, and Cushman & Wakefield's second-quarter count makes it concrete: 10,146 units absorbed, the most of any market in the country that quarter. Vacancy peaked in the first quarter of this year, completions head to a decade low next year, and CoStar calls rents flat by year-end with a real recovery in 2027. The discipline is reading which lane of the metro is actually recovering, and buying the door there.
Special feature: Austin, and the second derivative. If dispersion is the game, Austin is the board, because it is running the same cycle as DFW, about six quarters ahead.
Start with the brutal facts, because the posture demands it. Austin asking rents are still falling, down 0.8% over twelve months, the fifth-largest decline among major U.S. markets. Effective rents are down 2.6%, because three quarters of Austin properties are paying concessions, one to two months free as the going rate. Vacancy sits at 11.6%, fourth-highest in the country. Austin posted the lowest rent growth in America three years running. That is the level.
Now look at the direction. The rate of change is the trade. Asking rent growth troughed at negative 4.5% in the first quarter of this year. Two quarters later it reads negative 0.8%. Effective rents troughed at negative 7.5% the same quarter and have climbed almost five points since. Vacancy peaked at 15.8% in late 2024 and has fallen more than four points, one of the nation's sharpest two-year drops. Downtown asking rents are already growing at better than 3%. CoStar dates the zero-crossing for the whole market by the end of 2026, and its own model books next year's swing as the largest in its Texas tables. Everything is still negative. Everything is improving at the fastest rate in the state. That is the second derivative, and it turns before the sign does. By the time the number goes positive, the discount that made the math work is usually gone. I am not alone in this read: when CoStar ranked the Sun Belt's recovery in August by change in rent growth and occupancy, Austin came in second of every major market in the region while still negative, their write-up conceding the market "may be moving beyond the worst phase of its supply-driven correction."
The reason the turn holds is that both blades of the scissors closed at once. Supply: construction starts over the past year total about 7,000 units, two percent of stock, the lowest since 2012, and the pipeline collapsed seventy percent from its 2023 peak. The local count is blunter still: ApartmentTrends logged new submittals coming to an abrupt halt at 1,683 units last quarter, with 28,843 units sitting paused, waiting on financing that is not coming at these rates. Demand: Austin absorbed 21,398 units in twelve months, first in the nation as a share of inventory, and ApartmentTrends counted a second quarter that absorbed nearly three units for every one delivered, their words being an extraordinary level of demand rarely seen in this market. Rents on par with Dallas for the first time since records began reset the affordability math the boom had broken. Zillow's measure now puts Austin rent-to-income at 18.1%, the lowest of any major U.S. market, and Cushman & Wakefield counts just 15,174 units still under construction. And the demand is structural: over the past decade Austin nearly doubled its share of national apartment absorption, from 2.4% to 4.5%, the third-largest gain of the 394 markets CoStar tracks. The cyclical trough is happening inside a market whose baseline demand weight permanently re-rated.
The capital confirmation is already in the tape. Sales volume is up 50% year over year. Institutional buyers expanded Austin acquisitions 45% since mid-2025. The market price per unit rose quarter over quarter in the second quarter, the first increase since the 2022 peak, while the year-over-year read is still negative, which means a headline writer can truthfully say Austin values are falling while the quarterly tape says the bottom is in. Both are true. I put more weight on the quarterly series. The transaction quality says the same thing: of the nine Austin sales ApartmentTrends logged in the second quarter, not one was foreclosure-related, against a year-ago quarter where nearly half were. Class C, the product I hunt, still prices 35% below its peak. The distress is clearing out of the transaction mix while the discount is still on the table. Windows like that have not stayed open long in past cycles.
I can tell you what this looks like from the ground, because we underwrote it. In July my team worked an Austin portfolio and pulled the comp set: seventy percent of competing properties offering concessions, effective rents down 3.1% in the submarket. That is the concession war in its late innings, and the operators can see the end of it. On Camden's earnings call their CFO walked the arithmetic that turns the corner into a coiled spring: a market at two months of concessions moving to one month books roughly an eight percent effective gain before face rents move a dollar. The burn-off is the rent growth, and it never shows up in the asking-rent series until it is over. Jessett gave the proof from Camden's own book on the Rent Roll podcast in August: a downtown Austin asset that sat in the high 80s on occupancy a year ago runs, in his words, at like 97, 98% today, with new-lease rents up 13%. There is a bear case, and it is honest: absorption is forecast to cool by about a quarter as the delivery wave empties out, and one analyst on the same circuit thinks Austin could stay negative into 2027. That could happen. Even the cooled forecast keeps Austin third in the country on absorption rate, and my underwriting does not need the plus sign this year.
And so the incentives are on the table: I am buying in this market, and you should assume I talk my book. Here is the check on that. In my own Austin underwriting, the recovery gets zero credit. The price has to work at a 7.50% untrended year-one yield with rent growth at zero for two years, because the defensible read of every current series is stabilization emerging, not recovered fundamentals. The second derivative is why I am looking at Austin now. It does not get to pay for the building.
The decade trade underneath all of it: rail. DFW quietly crossed 200 miles of passenger rail in October when the Silver Line opened, 26 miles through seven cities with 210,000 jobs within a half mile of the corridor. What rail does to nearby real estate is among the best-documented effects in this business, and the region ran the experiment on itself. DART's own 25-year study, run by UNT and released last fall, counts $18.1 billion of development within a quarter mile of its stations, with rents 10% higher on residential and 12.6% higher on commercial than product half a mile out. The academic meta-analysis puts station premiums at 1 to 27%, with commuter rail at the high end. Grapevine's sales tax within a five-minute walk of its TEXRail station is up roughly 40% since service began. And the national math on the spend itself: every billion dollars of transit investment creates or sustains about 41,400 jobs and returns roughly five dollars of economic value per dollar in.
The pipeline is real money. The region's adopted long-range plan recommends passenger rail in the McKinney and Alliance corridors. The Frisco Line study prices 37 miles from Irving to Celina at $2.9 billion with 17,000 daily riders projected. The McKinney line study runs $1.8 billion and names Anna and Melissa beyond it. Fort Worth has said in public it would welcome a Cleburne line if funding appears. My own map runs further out than any funded study: commuter lines to the ring towns, Sherman, Gainesville, Greenville, Tyler, Cleburne, Granbury, Decatur, with new stations built on cheap land where a town can grow around them instead of bolted onto intersections that are already priced.
The stations are only half the thesis. The other half is what feeds them, and DFW is sitting on two land banks it treats as dead weight. The first is campuses. Dallas College is expanding El Centro from 130,000 to 800,000 square feet in downtown Dallas for roughly 30,000 students, and the university evidence says that is the right direction: Brookings measured downtown campuses producing 71% more startups and 123% more invention disclosures per student than their suburban peers, and San Antonio expects 10,000 students living and studying on UTSA's downtown campus by 2028. Move the smaller colleges to the centers rail serves, recycle the freed acreage into mixed-use, and each station gets a built-in daytime population. The second is golf. More than 800 U.S. courses have closed since 2006 and two-thirds of municipal courses lose money every year, while each one sits on roughly 150 acres. Las Vegas just sold a 95-acre municipal course that becomes about 1,500 homes. Prairie Village, Kansas split a 136-acre club into 412 homes plus a county park that doubled the city's usable park space. Every underwater course near a future corridor is a station village waiting for a rezone.
Dallas already built the museum piece for what hesitation costs. Under the Cityplace tower sits Knox-Henderson, a subway station shell the neighborhood rejected in the late 1980s. Adding the shell cost about a million dollars while the tunnel was open. Finishing it today would run on the order of $100 million, and there are zero plans to do it. The neighborhoods that said no to stations bought the region's most expensive lesson.
I want to be precise about what is fact and what is mine. The studies, the premiums, and the corridor price tags above are on the public record. The ring-town lines, the college moves, and the golf conversions are my map, on nobody's plan of record, and I hold that position on one belief: a metro that added 177,922 people in a single year will not move them all on highways. For this letter the point is simple. The workforce buildings I buy have to clear a 7.25% yield on today's income with zero credit for any of this. A corridor landing nearby is the same species as a renovation premium: never banked, always welcome. The difference is that the rail evidence base is thirty years deep, and the map only grows in one direction.
All of this depends on doing the underwriting yourself: underwrite the stabilized yield instead of a marketed proforma, reassess the taxes a new buyer triggers, fund the reserves nobody funds, and walk the moment a deal stops paying.
Here is the paper trail. I have screened more than 600 deals across the Sun Belt this cycle, more than 82,000 apartment units, with a median deal size of 100 units. Of the 580 fully worked in the tracker, I dropped 322 outright, fifty-five percent. Thirteen sit on the shortlist today, more than that reached it and got dropped along the way, and thirty-eight more sit as targets; fewer than one in eleven got even that far. Of the 473 deals with a recorded vintage, 74% are pre-1990 workforce product, the older stock the institutions will not touch. My rents sit at the line a strained workforce cohort can carry. Where a mismanaged building rents below what its own submarket already pays, bringing it to market is the job. Past that line I stop, because raising rent the way the area-median math says I could tips households already at the margin into cost-burden. Jay Parsons ran that wall from the other end in August, his own construction from Census income data and industry expense surveys: roughly a third of American renters cannot cover what it costs just to operate a typical rental. The affordability ceiling in this business is the expense line, and it binds the landlord too.
Why I haven't bought yet, after 615 deals since April. The paper trail has to include this: I have signed well north of a hundred confidentiality agreements since April, shortlisted thirteen deals, taken a handful of swings that did not land, and I have nothing under contract. The swings ended the way swings end in a frozen market: a seller who refinanced instead of selling, a higher bidder, a number declined. I would rather tell you that than let the discipline claim float free of it.
Here is where the 322 dropped deals died, counting each deal once by the reason that killed it. The income and demographic screen killed 149: the renter base in the submarket could not carry the rent the pro forma needed. Price and seller expectations took 67. Another 53 never produced the financials an underwrite needs, so they never got one. Size took 14, market 11, and the remaining 28 split across crime, taxes, occupancy, debt, and one-off disqualifiers. Notice how little of that list is about the building itself. The deals died on the renter's budget, on the seller's price, or on the seller's paperwork.
The tracker holds more than drop reasons. Where the public record gave up the loan, the picture got dark. Of 257 deals where I found the recorded mortgage, 94 are worth less than the debt on them by my normalized numbers, and roughly two dozen more barely clear it. One seller in five is marketing a property whose trailing revenue is already falling. And in the 290 operating statements clean enough to parse for fee income, the median property leaves about $90 per unit per month of ancillary income uncollected. That is the market this letter is written from: over-levered, slipping on revenue, and still leaving money on the table.
The price problem starts with what a broker package calls a cap rate. On older product the marketed number is built on last year's tax bill, no reserves, and a revenue line that may not sit on the same basis as the expenses under it. I rebuild it every time: reassess the taxes, fund reserves per unit, confirm the income basis, and only then call it a stabilized NOI. That NOI at a 7.25% yield, less the all-in capex the walk-through demands, less closing costs, is my price. The one exception the formula allows: when a seller's below-market fixed debt can be assumed, the debt does work the cap rate cannot, and the yield test moves to the levered return instead. A seller's ask arrives on only 288 of the 580 deals I track. Only 14 carry both a normalized NOI and an ask, and at the ask, 9 of those 14 fail the hurdle. That is the frozen middle in one line. So patience is the position, and patience here means work: the underwriting is done on all 600-plus so the number is ready the day the seller's situation changes. A rate cap expires, a lender stops extending, an insurance quote clears. Every trigger is on the tracker with a date next to it.
I want to close with the part that does not show up in a spreadsheet.
Everyone quotes the Buffett line about being greedy when others are fearful. Almost nobody does the second half of the work, because being greedy when others are fearful means doing the most underwriting of your life in the exact stretch when every headline tells you to stop. A mentor I hold in high regard put a name to the posture that survives a cycle like this. It comes from Admiral James Stockdale, who endured more than seven years as a prisoner of war in Hanoi and watched the optimists break first. They were the men who kept setting dates for rescue, home by Christmas, home by Easter, and died of a broken heart when the dates passed. The ones who made it held two things at once: unblinking honesty about the brutal facts of their present, and unshakable faith that they would prevail in the end. That is this market. The syndicators chanting survive till 25, then 26, then 27, the industry's own stay-alive-to-whenever hashtag, are the optimists setting dates. The brutal facts are everything above: the payroll prints, the foreclosure postings, the debt math, the frozen middle. The faith is that disciplined buyers of essential housing, bought on the cash flow it produces today with fixed-rate debt, come out the other side owning the thing everyone needs.
I can tell you what the work looks like, because I kept the receipts. In my brokerage days I finished number one in call volume out of more than four hundred agents across an eighteen-office region, week after week through the deadest stretch of 2023: 2,476 calls and 77 hours on the phone across six weeks, marketing a seventy-seven-unit 1973 deal in Irving with cast iron plumbing and soil issues. Twenty-seven days between Thanksgiving and Christmas produced fifty-two offers, forty tours, and a closing, then fifty-four broker opinions of value the next quarter.
Those receipts carried me to a seat leading acquisitions for an established Dallas family office, eleven hundred deals underwritten in thirteen months, two closed, both off-market, both the same trade this letter is about. And this year the test got personal. During the heaviest underwriting stretch of my life I collapsed in the dark, blood pressure gone, face first on the bathroom tile. I woke up in a hospital gown with my eye swollen shut and a cervical collar on. I kept working. The next week, still in the brace, laptop propped against a pillow, I found one of the best purchases in DFW of this entire cycle. The deals do not wait for you to feel ready.
So yes, I am on both sides of this. I want the return, and I want the rent to be payable, and in this market those two wants pull in the same direction. Buying right lets me charge less, and operating well lets me earn more. If you underwrite one thing this cycle, underwrite the gap between the story and the rent roll. The story says pick a side. The rent roll says both sides are the same trade.
Two slower currents belong on the watch list. Harvard's Joint Center counts net immigration collapsing toward 300,000 this year, a drag of roughly 420,000 households a year on formation through 2027, a genuine headwind for everyone's demand math including mine. And Greg Willett reads that same data as the setup for a wave of forced Class C sales in 2027 and 2028. That is his framing rather than Harvard's, and it happens to describe the conveyor belt this letter is built to catch.
The next tests arrive fast: Friday brings September payrolls, CPI lands on the 11th, and the new chair chairs his first FOMC on the 17th. The numbers will move. The setup underneath them moves slower.
Data current as of September 1, 2026, publish date September 1, 2026. Rates are the August 28 close, the last trading day before publication; markets are closed for Labor Day today.