Let’s start this month by remembering the absolute dumpster-fire that was last month’s . The number for November was patently ridiculous on its face, and it took mere minutes to realize that the BLS was showing 2-month changes for what were essentially one-month changes:
“Because what it looks like is that for many series the BLS didn’t calculate a two-month change based on the current price level – it looks like, especially for housing, they assumed October’s change was zero so that the two-month change reported for this month was actually a one-month change spread over two months. For example, even with the low Owners’ Equivalent Rent print in September, the y/y figure was 3.76%, so about 0.31% per month. The BLS tells us that the two-month change in OER was +0.27%. That looks more than a little suspicious to me.”
That in fact was what had happened. The BLS has clearly spelled-out procedures for what happens when they cannot collect a price. If they can collect the price for other similar items, they impute the data for the uncollected price by ‘adjacent cell imputation.’ Happens all the time, and has happened more since there have been fewer data collectors, and that has upset a lot of people…but it’s no big deal. What happens less often is that the BLS can collect no similar price, or they don’t have a statistically-significant sample; in that case the BLS procedures call for the prior price to be carried forward and then the price gets naturally corrected the next time it can be gathered. I’ll talk more about this in a week or two, but if the item was generally rising in price that unchanged estimate for monthly price change will be a little low in the first month and a little high in the second month. If the item was generally getting cheaper, you’ll be a little high and then a little low when you catch up. But that’s better than taking a wild unscientific guess.
But normally, that happens for tiny categories. In this case, since no prices were collected, the BLS realized that its procedures called for carryforward pricing. After the data were released, they were very transparent about the fact that this caused understatement in the CPI, and that while most categories will be corrected by normal sampling in a month or two, the rent and OER samples will take about six months to correct because of the way those samples use overlapping six-month survey panels. You don’t need to worry about the fine details here, but to realize that the October number is missing, the November number is garbage, and the year/year numbers won’t be “right” for a while.
Ergo, take everything in today’s number, and all the charts, with a grain of salt.
A little side note is that the BLS was able to collect some data for November, when there was historical data available, so some of the series are complete. And some series have a dash (“-“) for November. Bloomberg simply omits October for those series. The practical consequence is that this is a massive mess for anyone who has built spreadsheets based on fairly normal assumptions about data structure! And it will be for a while. Anyway, on to today’s number.
Over the last month, inflation markets have been little changed.
They’re actually even more unchanged than that looks like, because the apparent rise in short-term inflation expectations is a quirk of the fact that every day, the window covered by a 1-year swap rolls forward one day, and as it turns out the day that it loses on the front end is a day when the NSA CPI was declining sharply thanks to the garbage report we just mentioned. So, the new 1-year swap has less of that garbage dragging the y/y rate down, and so it rises slightly. The net result is that inflation expectations at the front end are not really rising.
The expectations for the December CPI were for +0.31% on the seasonally-adjusted headline, with +0.32% on . These are even more guessy guesses than normal, since economists had to figure which categories might jump back and by how much. The actual CPI came in at +0.307% (SA) on headline CPI, and +0.239% on Core CPI. We will ignore the y/y rates for now. If we take those numbers at face value, it would annualize to 2.9% on Core CPI and 3.75% on headline CPI. That doesn’t seem wildly off, with the obvious caveat that annualizing a one-month change is stupid. Sorry.
Now, the Median CPI is going to be a snap-back sort of month. I think. The median category appears to me to be one of the regional OERs, so the actual number will depend on the seasonal adjustment the Cleveland applies to that subindex. And I don’t know what the Cleveland Fed did for their last data point so they may be jumping off differently than I did. But any way you slice it, we’re going to be around 0.30-0.35% for median.
This is right about where the trend was prior to September. A word on September: while it is convenient to think that September was the ‘last good data point’ we had before the shutdown, remember that month had an outlier Owners’ Equivalent Rent number (0.14%, vs a series of 0.28%-0.40% that happened in the year prior to that) that we expected to rebound in the next month. We never saw the rebound. Median CPI was also affected by that, and so the last truly normal number was August. The upshot of it is that there may be some continued deceleration in median CPI, but it isn’t clear at all.
Core goods as of this month were +1.42% y/y. They look to be leveling off a bit, and it may be that the bump from tariffs (which, contrary to economic theory but in keeping with the way it really works, got bled into prices over a period of time rather than all at once) is petering out. Too early to tell, and part of this leveling out is due to soft Used Cars data in this month’s release. Core Services, mostly housing, continues to decelerate but see all of the caveats about rents.
And yes, rents went back to doing what they had been doing. Primary Rents were +0.26% m/m, and Owners’ Equivalent Rent was +0.31% m/m. So, yeah: that dip in OER in September was a mirage, and we’re still running at 3-4% in rents although the one-month BLS blip makes it appear that we’re still decelerating. I am not sure that’s really true.
Speaking of rents, Barclays put out a great piece earlier this week. It’s called “Apples and oranges in the CPI basket: Why market rent gauges mislead on shelter,” and if you have access to it you should read it. If you do not have access to it, you can just read my articles from the last few years. Seriously, though – it’s a very good piece and I’ll talk about it more in a week or so. But here are two of my favorite exhibits from their writeup.
Since 90% or so of rents are continuing rents, and all of the high-frequency rent indicators are recording new rents…can you see why there’s a problem?
That’s why a few years ago I migrated my model for rents to be based on a bottom-up estimate of what landlord costs were doing. Here is that model with the updated Primary Rents.
Normally, the Enduring Model has more lead time, but since part of it relies on that haven’t been released since September (and which is coming out tomorrow), the look forward is shorter than normal. Still, it says the same thing I’m saying above and approximately what Barclays is now saying – 3% on rents is about where it should be. It is not likely to decline sharply from here. And that means that getting CPI to 2% is going to depend on a collapse in goods prices or core services ex-rents, neither of which I see happening soon.
Although I should point out that core services ex-rents, aka Supercore, has been looking better of late.
And really, that’s the summary of the number: some of the effects from bad stuff (e.g. tariffs, which were never as big a deal as people treated them) are wearing off but some of the positive trends (e.g. the deceleration in rents) have also mostly run their course. The Enduring Investments Inflation Diffusion Index shows that there’s a bit of an upward trend in the distribution of accelerations/decelerations.
All of which points to the same thing I’ve been saying for a while, and that’s that once the spike was over we knew inflation would drop but it was likely to settle in the high 3s/low 4s (since amended to mid-to-high 3s). The tailwinds on inflation have turned into headwinds, so monetary policy overall needs to be tighter than it otherwise would be. The Fed doesn’t see it that way yet, and new additions to the Board of Governors are definitely more likely to be dovish than hawkish. Not only that, the federal government is also adding liquidity…or will be, if the President convinces Fannie Mae and Freddie Mac to buy $200bln in mortgages. A Federal Reserve which appreciated the inflation risks would be preparing to drain away that liquidity, no matter what it was going to do on interest rates. There’s no sign of that.
As a result: I think it’s reasonable to expect dovish outcomes from the Fed from here, although Chairman Powell will doubtless try to stick it in the eye of the President (and the American people get caught in the crossfire) before his term is up. That differs from the Fed of the last 30 years only in degree. They are going to be too loose, and there’s a good risk that inflation heads higher from here (not to 9%, mind you, but getting the sign right will matter).