Archive
Inflation Guy’s CPI Summary (September 2025)
Well, it seems like it’s been a while since the last CPI update! Thanks to the government shutdown, it has been since this data is a week and a half later than it was scheduled to be. The importance of the CPI release is obvious, but it was reinforced by the fact it’s the only one the government is calling people back to release. It isn’t that we don’t have reasonably-accurate alternative ways to measure price pressures, though – it’s because unlike Payrolls and most other government releases that are important touchpoints for economists, the CPI is an important legal touchpoint for contracts, bonds, and legal obligations of the federal government. In this case, September’s data is a crucial number needed to calculate COLA adjustments for Social Security for next year. If this had been October’s data? I’m not sure they call back workers to release it. But that’s next month’s problem.
Speaking of next month’s problem: the government shutdown did not affect data gathering for this month’s number; they had to recall the people to collate the data and publish it but not the collectors. So the quality of the data should be fine. The data-quality question is much murkier when we look forward to next month, but since much less of the data collection is done by guys with clipboards these days, it might not be as bad as you think. Still, that will be the concern for the October CPI released next month. Like I said: next month’s problem!
Heading into the release, consensus was for +0.37% on headline CPI (SA) and +0.29% on core. I have to admit that I was confiding to people that this seemed sporty because the prior month had seen a surprising acceleration in rents that could be reversed, indications are that Used Cars would be a drag, and Food at Home also looked soft (I was right on 2 out of 3 – Food at Home was an add). That told us going in that if we were going to get to +0.29% core, either I had to be wrong on most of that or core goods ex-used cars was going to have to be pretty strong. Tariffs definitely are helping to push that narrow group of the consumption basket higher. But is that enough? Let’s see.
The backdrop going into the data was that rates have been generally softening, and the inflation swaps curve has been steepening (lessening its inversion, with near-term inflation pricing dropping more than longer-term inflation expectations). That’s consistent with a return to normalcy…but it’s really happening because energy prices have dropped quite a bit until the last couple of weeks, and that has a more immediate impact on the front of the inflation curve. The mean reversion time for energy prices is something like 15 months, so by the time you’ve gotten a few years out the curve today’s lower gasoline prices shouldn’t much affect your expectations of inflation forwards. But it affects inflation spot, which propagates through the forwards.
Actual print: SA CPI +0.31%; SA Core +0.227%. Softer than expected, and it took only a moment to see that a big part of that was due to a sharp deceleration in Owners’ Equivalent Rent (see chart). Some of that was the give-back I expected, but it was more than that and so we should put this in the back of our minds for next month – we’re probably due a reversal in the other direction over the next month or two.
Interestingly, even with the miss the Core CPI time series doesn’t look terribly weak. I mean, +0.227%, repeated for a year, still gives you 2.7% core CPI. And we won’t get downside drags from cars and housing every month.
Interesting jump in apparel this month. It’s a small category and always volatile, but since we also import almost all of our apparel it’s one place I look for tariff effects. But note the y/y numbers are still very low and in fact decelerated with this jump, implying last year’s bump in September was even larger. That’s a seasonally-adjusted figure, but I wonder if the problem here is that seasonal adjustment is failing us. Maybe pre-holiday mark-ups (from which we can show great discounts in a month!) are happening earlier. In any event it’s only 2.5% of the CPI so probably not worth too much computational cycles.
Core goods inflation rose slightly to +1.54% y/y, and core services declined to 3.47% y/y. The latter is mostly and maybe entirely due to housing, which is a core service. The former is interesting because Used Cars/Trucks was -0.41% m/m. That was expected, but it means that other core goods were more buoyant.
So here are OER and Primary Rents. 3.76% y/y (only +0.13% m/m) and 3.4% y/y (only +0.2% m/m). You can’t really tell a lot about the miss today from this chart – I showed the m/m series earlier, and the bottom line is that this continues to level out. I think the flattening is going to be more dramatic over the next 3-6 months but we’ll see. Lodging Away from Home rose again, +1.3% m/m, and is now flat y/y.
At this point, I’m thinking: with rents a downside surprise and Used Cars a downside surprise, this isn’t that bad a miss. In other words, if you’d told me we were going to get those numbers from rents and cars I would have thought core would be a lot weaker than +0.23% m/m.
Earlier I showed the last 12 Core CPIs. My guess at Median looks better, but that’s mostly because the median category is West Urban OER and even split up, an aberration in OER – and that’s what I think this is – is enough to sway Median CPI. It also means my estimate of median, +0.213% m/m, might be off because the Cleveland Fed separately estimates the seasonals for the regional OERs and so I have to guess at that part. My guess will take y/y Median CPI to 3.5% from 3.6%. And the Fed is easing. Hmm.
Here are the four-pieces charts. Food and Energy +2.99% y/y. Core Commodities +1.54%. Core Services less Rent of Shelter (Supercore) +3.37%. Rent of Shelter +3.53%. These are the four pieces that add up to CPI. None of them looks terrible except for Core Goods, and there’s limited upside to that – and it has a short period, so in a year it’s likely to be lower. I do think that going forward, core goods remains positive instead of the steady deflation it was in for decades, but not big positive. However, you need it to be negative if you want inflation at 2%, unless you get core-services-ex-rents a lot lower (but that’s highly wage-driven, and reversing illegal immigration helps support that piece somewhat) or rent of shelter a lot lower. The latter is certainly receding but it’s not going to go a lot lower.
I don’t usually spend a lot of time talking about energy, because that’s a hedgeable piece (largely – gasoline is a big part of energy and that’s easy to hedge with a little lag; electricity is harder). This month, Energy was +0.12% NSA. But next month, we’ll see a decent drag because of the sharp drop in gasoline over the last few weeks. That’s a little early compared to the usual seasonal, and it may mean we get the usual December drop in gasoline in October CPI.
Except…that I think the White House has teased that we might not get October CPI at all, just skip it, because of the difficulty gathering data. If that is true, the fallback mechanisms will kick in. See my piece on what that means, here, but the bigger point is that you wouldn’t get my scintillating commentary. I guess again that’s not this month’s problem.
Now, I have to show this almost by habit, and because the economists expecting housing deflation will be dancing in the streets. Take pictures, and show them again next year. They never learn. Housing inflation is slowing but there is no sign rents are going to come anywhere near deflation. Except maybe on a weighted basis if Mamdani gets elected Mayor of New York City and freezes rents. But then we’ll have to start looking rents ex-NYC.
How disinflationary a period are we in? Wellllll…of the item categories in the median CPI calculation, there were zero core categories that decelerated faster than 10% annualized over the last month (-0.833% or faster). On the plus side, there were Personal Care Services (+11.9% m/m annualized), Footwear (+12.0%), Motor Vehicle Fees (+14.2%), Tenants’ and Household Insurance (+15.2%), Lodging Away from Home (+17.5%), Miscellaneous Personal Goods (+17.9%), Men’s and Boys’ Apparel (+19.3%), and Public Transportation (+21.5%). These are small categories for the most part – but not all import goods and interesting in that the tails are all to the upside. That’s not the way a disinflationary economy usually looks, although I don’t want to overstate the importance of a single month!
Here’s the observation about long tails compressed into a single number, the Enduring Investments’ Inflation Diffusion Index. It’s signaling upward pressure.
Below is a chart of the overall distribution. The two big spikes in the middle are mainly rents and OER. But take those away and you can see there’s not a lot of categories in the 1-3% range, and a decent weight in the 5-6% range. This doesn’t really look like a price system settling back down placidly to 2%.
Now, the stock market clearly loves this, which makes sense. The Fed is going to ease, probably twice more this year. But that was already baked into the cake in my mind, because the Fed no longer targets 2% inflation. Remember that in the most-recent change to the 5-year operating framework the Fed, in Chairman Powell’s words, “…returned to a framework of flexible inflation targeting and eliminated the ‘makeup’ strategy.” I talk more about that here: https://inflationguy.blog/2025/09/02/the-fate-of-fait-was-fated/ Ergo, the Fed doesn’t really care if we get to 2%. They’d prefer to not see inflation head higher, but they can spin a story to themselves that even though median inflation is in the mid-to-high 3s, “the process of inflation anchoring is underway” or somesuch nonsense. As long as it’s not hitting them in the face that inflation is going up, they’ll keep relying on their models that say it should be going down. N.b., those are the same models that said inflation shouldn’t have gone up that much to begin with, and should have been transitory, but we all know “Ph.D.” stands for “Pile it higher and Deeper.”
Eventually, inflation going up probably will hit them in the face. But that’s such a 2026 problem.
Does Crypto Expand the Money Supply?
We live in interesting times, and let’s face it: mostly, in a good way. It doesn’t have to stay that way, naturally, and it won’t stay that way naturally.
This has always been the weak spot in any system that insists on centralized management of certain functions. Of course, that’s the fundamental flaw and conceit of socialism: it relies on the active intercession of omniscient beings to order activities better than the masses of private actors can. Usually, “better” means “less volatile” to the policymakers who set up the committees of omniscient beings (personally, I would say “better” means “less fragile,” which is the opposite of “less volatile”).
The best argument for using the collective wisdom of the anointed few is to prevent the tragedy of the commons, where individuals making private decisions can impact the use of public goods. And that brings us to money.
I think it is a fascinating question whether ‘money’ is a public good, which should be regulated and controlled. Or is a particular currency, such as the US Dollar, the public good which should be regulated and controlled? The argument the Federal Reserve would make is that, absent the control of the Federal Open Market Committee, the money supply would grow or shrink in dangerous and random ways. Or at least, that would be the argument they would make, if they cared about the stock of money any more.
There is no plausible argument in my mind that “interest rates”, which is what the Fed now works to control, is a public good that is better managed by the Smart Guys. So, weirdly, the Fed now manages something which they don’t have any knowledge about that should supersede private market actors (rates), but does not purport to manage something they could plausibly argue is a common good that no one directly controls (money).
** Separate question: are the Cognoscenti at the Fed any good at it? Chairman Powell said yesterday that the Fed is likely to stop running down its balance sheet soon. With the balance sheet still at 22% of GDP, compared with the pre-GFC normal of about 6% – see chart – “Until the job is done” has apparently become “until it’s time for my smoke break, and then you’re on your own.” What’s the matter with kids today?
So the answer to this ‘separate question’, as inflation remains at the highest level of this millennium and is now headed higher, is “of course they’re not. Why are we even asking that question?”
I actually want to go slightly further. The Fed no longer tries to control the money supply, which at least they might have an argument for doing, in preference to managing interest rates against the market-clearing actions of private actors. But over time (and accompanied by the whining and moaning of central bankers), the concept of money has gotten squishier and squishier. One of the reasons that central bankers want to control crypto is that they fear the power of money loose in the wild (ironically, given that they stopped worrying about money a long time ago), untamed by the Anointed Stewards of Money.
The question is, does crypto expand the money supply? For the purposes of this question, let’s ignore the official definitions of money, M1, M2, M3, etc and just focus on ‘spendable balances.’
If you give me a dollar, in exchange for something that feels like a dollar and that you can spend (say, a stablecoin like USDC), have we increased the money supply? The answer depends on what I do with that dollar. If it is deployed to a vault, then obviously the number of ‘dollarish’ units in circulation haven’t changed. You have minted $1000 USDC, but there are now $1000 USD that are sequestered in a vault and not spendable. The amount of spendable money hasn’t changed. If instead that $1000 goes to buy a Treasury bill from the government, then it is going to the government to spend. Normally, buying Treasuries doesn’t change the amount of spendable dollars, because in buying a Tbill I am deferring my decision to spend (instead, I hold securities) and delegating that decision to spend to the government. I exchange my future spending for the government’s current spending, and in the future that transaction is reversed when the Tbill matures. Some people think that means that Treasury issuance increases inflation because it increases money, but it doesn’t. The Treasury bill is just a token representing my deferral of spending into the future.
But if I was able to buy that Tbill because I issued a USDC token, which you can spend, and then gave the fiat money I received from you to the government in exchange for a Tbill, then I have doubled the number of spendable dollars in circulation: $1000 in the form of USDC, and $1000 in the form of dollars sent to the Treasury which will be spent. Essentially, what has happened is zero-reserve banking. If I were a bank and you deposited $1000, I could lend out only, say, $900 of that (“fractional reserve banking) and in principle the Fed can control that multiplier by changing the reserve requirement.[1] But now you’ve deposited $1000 and I am lending 100% of that to the government. Stablecoin manufacturers in this way are basically banks issuing their own currencies. Now, a lot of that money is going abroad, but it looks like money to me.
Worse are the vaporware crypto issuers who simply create supply out of thin air. If people accept bitcoin as money, rather than as a speculative chip to trade around, then I have created money with no reserves whatsoever, and no limit on how much ‘money’ I can so create.
If this is true, then the irony is that crypto – which was inspired originally by the desire to remove money from the ministrations of the Very Smart Bankers who could ruin money by creating too much of it – could be the very tool that creates the inflation its originators wanted to protect against. In that kind of world, I really don’t understand the use of a nominally-anchored stablecoin. If the overall money supply growth is unbounded and now essentially uncontrollable (once the size of the crypto world gets sufficiently big), then holding something that is pegged to the sinking ship seems counterintuitive to me.
While I didn’t start this article with the intention of pointing out that our USDi coin is a raft rather than an anchor (like stablecoins), it does seem to be relevant here to mention that you can now mint USDi directly from our website: https://usdicoin.com/coin . And, while the increase of USDi will contribute to the overall money supply – at least it has a built-in defense!
[1] …but it doesn’t really work like that any more. The Fed still has a dial to turn that limits how much lending can happen on a given depository base but it isn’t as clean as it was when there was a simple reserve requirement. This is well beyond the point of this article.
