• Inflation fell sharply below consensus in y/y terms
  • Implied m/m average changes over October–November were broadly soft
  • Why it’s all made up data you should view very skeptically…
  • ...as markets largely faded it
 
  • US CPI / core cpi, y/y %, November:
  • Actual: 2.7 / 2.6
  • Scotia: 3.1 / 3.0
  • Consensus: 3.1 / 3.0
  • Prior: 3.0 / 3.0 (from September, October missing)

Oh come on. Tell me you don’t believe what the BLS just threw at us in chart 1. Bonds sort of did by driving a small rally, but it was pretty minor in relation to the apparent magnitude of downside surprise. Markets only have another cut fully priced by the April FOMC. Perhaps enough folks in markets view the numbers as unreliable as I do, but there is a case for the FOMC to accept the evidence including payrolls and cut in January.

Chart 1: US CPI Inflation

Why? I’ll get to that after reviewing the readings. Perhaps the only one who accepted the numbers at face value was Kevin Hassett. He arguably failed the bond market trial balloon test with very dovish views that are very close to the administration (chart 2). The 10s yield has ebbed and flowed with his chance at winning. If President Trump is truly concerned about the housing market, then given the 30-year mortgage is priced off 10s he might not like what the bond market is telling him about Hassett’s candidacy. Expect more ongoing market trial balloons from the administration.

Chart 2: Hassett Failed the Bond Market Test

The Results

Year-over-year CPI fell back to 2.7% versus consensus at 3.1% with core CPI falling back to 2.6% (consensus 3.0%). That looks a lot softer than expected and the magnitude of the miss is unusual for y/y figures.

We did not get m/m figures with this report because no October estimates were provided given the government shutdown from early October to November 12th which prevented data collection.

We can loosely infer the average m/m readings between September and November by taking the SA index levels by component and backing out the implied percentage changes over the two months divided by two.

Chart 3 does this by component and ranks the estimates from highest to lowest. There was widespread softness across the whole basket. Core CPI’s implied average monthly change over this period was 0.08% m/m SA. 

Chart 3: November Changes in US Headline CPI Categories

Core goods ex-food and energy prices were up by just 0.06% m/m SA.

You get the idea. Food 0.03% m/m SA, gas 0.4%, OER 0.14%, rent 0.06%, clothing -0.36%, newe vehicles 0.16%, used vehicles 0.52%, all services 0.11%.

Why it’s Totally Fabricated Data

I have very little faith in the BLS’s data collection for several reasons.

Most of October's prices were not collected given the shutdown.

This spawned guidance that the "BLS could not collect October 2025 reference period survey data, so survey data were carried forward to October 2025 from September 2025 in accordance with normal procedures."

Also recall that the shutdown extended to mid-November (12th). There were lags in getting back to work that probably went beyond guidance that “CPI data collection resumed on November 14, 2025.” The time and resources they had to collect data in November should be viewed with high caution. For one thing, the DOGE packages that expired in September likely left the BLS further strained for resources. Let alone the impact on morale from having its leader fired and into the broader layoffs.

As per the BLS, “Collection began on Friday, November 14. By authorizing additional collection hours, BLS attempted to collect data for the entire month of November.” This is important because some would argue that perhaps the figures were biased lower by seasonal sales like Black Friday on November 28th and the lead up to it. I’m not a believer in that argument; these sales happen every year, there is spartan evidence they were greater or lesser than in the past, and they would mainly affect some goods with little emphasis on services. For example, it’s not like OER and rent go on sale for Black Friday!

So this says two things. November is missing data for the first half of the month that simply could not be collected. And I think we should be cautious toward the resources they had to competently collect data in the second half of November.

Further, at about 11:15amET or so, we'll get the share of the basket that was imputed through alternative methods (here, ‘different cell’ column). It had risen to 40% in September for an all-time high (chart 4).

Chart 4: BLS Use of Alternate Estimation Methodology in US CPI

The SA factor wasn’t a stand-out like it has been for nonfarm payrolls (chart 5).

Chart 5: Comparing US Core CPI SA Factors for All Months of November

How Will the Federal Reserve React?

I think the Committee will sound cautious but run with the data that we have so far which is bad for jobs as explained here, and soft for cpi if you believe it and subject to PCE translations including PPI.

In other words, if they had to make the decision today, then they would be very likely to cut by another 25bps on January 28th. Lots of data and developments between now and then, but still, I can't see them having the confidence to throw it all out and dig in. The Board is generally more dovish than several of the regionals who rotate next month

Please see the rest of the collection of charts for further perspective if desired.

Chart 5: US Inflation Breadth; Chart 7: US Core Goods Inflation; Chart 8:US CPI Core Services Ex-Housing; Chart 9: US Housing Inflation
Chart 10: US Apparel; Chart 11: New vs Used Vehicle Inflation; Chart 12: US Food Prices; Chart 13: US Financial Services
Chart 14: US Airfare
Chart 15: November 12-Month Change in US Headline CPI Categories; Chart 16: November Weighted Contributions to the 12-Month Change in US Headline CPI