# In All the Noise, Ask One Thing: Did This Cost Reach Anyone's Paycheck? Source: Realpha Blog (blog.getrealpha.com) Original article and charts: https://blog.getrealpha.com/en/blog/macrovoices-2026-10-01-macrovoices-552-david-rosenberg-navigating-the-noi/ > MacroVoices #552, October 1, 2026: David Rosenberg on why he won't buy the inflation narrative, what is actually driving real yields, and the November date nobody circles. Listening notes for education, not investment advice. Published: 2026-10-02 Locale: en Tags: inflation, Federal Reserve, bonds, macro, podcast notes ![Dusk over a finished but empty American suburban subdivision, price-cut signs at the front doors, the road running deep toward the horizon where a lit data center glows](/covers/macrovoices-2026-10-01-macrovoices-552-david-rosenberg-navigating-the-noi-cover.png) > Is it right to decide whether a thing exists, when you have neither seen it with your eyes nor heard it with your ears? > > —— Su Shi, "Record of Stone Bell Mountain" (Northern Song, 1084; translated by the author) On the October 1, 2026 episode of MacroVoices — #552 — Rosenberg Research founder David Rosenberg argued that today's inflation scare is missing one link: wages. His number is core goods CPI, which ran at 1.5% year over year twelve months ago and sits below 0.7% today, a four-year low — over the same stretch in which the U.S. net effective tariff rate posted its largest increase since the early-to-mid 1930s. His claim comes with a stated condition: unless the labor market passes energy and tariff costs through into pay, a shock like this behaves like a tax hike, and what it compresses is demand. The episode was recorded before the November 3 midterms, so the political turn he builds on has not happened yet. ## Is inflation the risk to watch right now? Watch it, yes. Treat it as the threat the business channels describe, no. His reasoning rests on an experiment that already ran. Through the spring, summer and fall of last year the same commentators warned of a mountain of imported tariff inflation. The tariff rate did move — by the most in roughly ninety years — and the place transmission should have shown up, core goods CPI, decelerated from 1.5% to under 0.7%. Somebody ate the increase: the exporter protecting market share, the importer at the border, margins along the way, or all three. The eating is visible in the data. So he turns the question around: if the tariff shock couldn't produce sustained inflation, why would energy alone manage it? ## Why wouldn't higher crude and diesel turn into lasting inflation? Two layers. The first is arithmetic. Inflation is a rate of change, not a level. If diesel simply holds here, in twelve months it drops out of the year-over-year figures; to kill the momentum, energy only has to peak, not retrace. He notes traffic through Hormuz back to levels unseen since the war began, and Gulf exports ramping, including that critical Saudi east-west pipeline. The second layer is the one he actually judges on: transmission. He takes apart the eighteen non-transitory months from 2021 into 2023 — that episode stuck because it reached wages. Two trillion dollars of stimulus checks landed as the economy reopened, and, more decisive, the government spent two years paying people more not to work than to work, so employers had to bid up to get labor back. That is a complete wage-price spiral. He separates the 1970s out too: repeated oil shocks into a unionized economy where everyone had a cost-of-living clause and labor held bargaining power. What he sees today is nominal wage growth decelerating. New York Fed's John Williams said in a September 29 speech that there is no evidence these price shocks are entering wages. Rosenberg adds proportions: labor is 30% of cost at the retail level, six times more powerful than energy, while diesel's pass-through into transportation and freight is roughly a 5% impulse. We watch energy tick by the second on a Bloomberg screen; we never see labor costs move. What happens when pay can't absorb it? You still get higher prices in fuel and food — cars need filling, people need to eat — and the other 80 to 85% comes under downward pressure as real wages go negative and demand gets destroyed in the cyclical parts of services and goods. At that point the shock does the economic work of a tax increase. Would you call a tax increase inflationary? Only if you take it to your boss and win the raise. ## Are rising yields about inflation, or about something else? One of his lines has been quoted everywhere: the rise in yields is 90% real rates and 10% inflation expectations. The part I wanted was the mechanism behind it. He offers two sources. Hyperscaler capex has outgrown internal cash flow — three of the four largest are now net free-cash-flow negative — so they are issuing at the long end, in multiple currencies, and corporate credit demand is now colliding with government credit demand. The bigger one, in his view, is regime change at the Fed. His comparison is hard to argue with: the national debt went from 38 trillion to 39 trillion between October 2025 and February 2026, when the ten-year sat below 4%, and nobody called it a debt crisis. At 40 trillion, with a fresh zero on the end, everybody does — and the AI capex boom was already running back in February. What changed is monetary policy. The market priced one or two cuts, flipped to a tightening bias, got an actual hike, and now prices three or four more. A move of well over a hundred basis points across the Treasury curve is that reset. He is blunt about the new framework. Kevin Warsh watches the share of the PCE deflator rising faster than a 3% annual rate, a statistic in which a bag of peanuts up 10% and an automobile down 5% carry the same weight. And Warsh gives no guidance, so the market reads every syllable as hawkish — he opens his mouth and the ten-year jumps six or seven basis points. ## November has two dates. Why does he say people only remember one? He asks audiences what November 4 is, and everyone answers midterms — which is the day before. November 4 is the Treasury refunding announcement. Go back to the fall of 2023. Coming off the most aggressive tightening cycle on record, with the AI trade already running, the ten-year rallied a hundred basis points in barely more than two months. The cause was Janet Yellen cutting the supply of notes and bonds at the refunding and flooding the market with bills. His division is clean: demand for fixed income is the Fed's lever through the balance sheet, supply belongs to Treasury, which chooses maturities. So he suspects Scott Bessent does something similar after a similar run-up, and that it matters far more than a 6-billion-dollar buyback program in a market that turns over a trillion a day. On November 3 he talks only about what he has confidence in: the calendar. Historically, when one-party power swings to a two-party split after a national election, then over the two years to the next election the economy slows 80% of the time, inflation falls 80% of the time, and bond yields fall 80% of the time. He'll take those odds. The mechanism is fiscal: the big bill passed because one party held Congress, and depreciation allowances and outsized tax refunds are one-offs. After November 3, no new fiscal stimulus enters the economy. He asks whether the bond bears have that in their models. ## What could break the AI trade? The host raises Chinese open-source models: roughly as good as frontier models were three months ago, free, run on your own hardware, which also air-gaps your proprietary data from somebody else's cloud. Rosenberg calls that global competition, and says we are watching in real time that this industry has no genuine barriers to entry — with China once again acting as a source of global disinflation. Then he moves to a corner nobody discusses. Warsh mentioned AI eight times at Jackson Hole and housing once, so Rosenberg's question is: do you live in a data center, or do you live in a house? U.S. housing is in recession. Builders are discounting hard, there is still eight to twelve months of new home inventory, and average new home prices are down 8 to 12% year over year — a 50-trillion-dollar asset class deflating. Multifamily is still overbuilt against demographics that no longer exist, since out-migration has left the population flat, so rents keep sliding. Plug real-time rents and home prices into CPI and the core lands well below 2%, and shelter outweighs energy by a wide margin. He wants the same look beneath the surface in equities. The conversation is Mag 7 and Nvidia, while regional banks, homebuilders, retailers, restaurants, cruise lines, theme parks and small caps are hurting, and the average and median U.S. stock is down about 15% from its 52-week high. The trade desk segment adds that only 20% of S&P 500 members are above their 50-day moving average — four in five have broken trend. The parts of the index most tied to Fed policy are already 10% off their highs. ## So how do I use this? Three habits of mine got adjusted. First, when frightening news arrives, look for the transmission chain before the conclusion. The useful thing in this episode is that chain: for a cost increase to become durable inflation it has to enter somebody's income; if it can't, it is a one-time tax, it compresses demand, and the thing to worry about afterwards is growth. With the chain in hand I know which cell to watch — average hourly earnings and collective bargaining agreements, not the price of oil. My own failure mode is to feel something first and then go collect numbers that agree. Second, the denominator. I hold a few names well off their highs, and I used to read the index near records as evidence that my stock picking was the problem. The median stock down 15%, four in five below trend — that doesn't make my losses smaller, but it changes my attribution. Figure out whether a holding is falling with a whole cohort or falling on its own news, because the two call for different responses. Third, go find the cell nobody is discussing. "Do you live in a data center or a house" is a good tool. When one topic absorbs all attention, mispricing tends to sit where nobody looks — his example being a 50-trillion-dollar asset class deflating while appearing once in a central banker's speech. I try the same thing on my own list: write down the word I read most this week, then ask what went a whole month without a glance because of it. ## Worth a look - MacroVoices #552, October 1, 2026, hosted by Erik Townsend and Patrick Ceresna, with guest David Rosenberg - New York Fed President John Williams' September 29, 2026 speech on whether price shocks are entering wages - Three public data series: core goods CPI year over year, new home sales and prices, ADP wage growth - Treasury's November 2023 refunding announcement under Yellen, and the ten-year's path over the following two months - David Rosenberg's new book, *Bear in the Bull Ring*, published October 6, 2026 ## One Thing to Take With You Whether a shock becomes a durable problem depends on whether it reaches someone else's income or behavior. If it can't, it is a one-time pain. Once it does, it starts feeding itself. Here is something I tried. Take a sheet of paper and write down the thing worrying you most right now — it needn't involve money; a recurring argument at home or a situation at work counts. Underneath, write this sentence: "For this to become a long-term problem, whose behavior has to change?" Name the person and name the behavior, then go check whether they changed anything this month. If nothing changed, the worry will probably dissolve on its own; if something did, you now know where to act. I did this once and found that the thing I had been watching for three weeks had an empty first cell.