US Core CPI is Too Warm to Cut

• Core CPI was warm and the breadth of price increases rose

• Core services drove most of the heat

• Tariff pass through evidence could take a while yet

• Markets reacted dovishly. The FOMC majority will not.

• US CPI / core CPI, m/m %, SA, July:
• Actual: 0.2 / 0.3
• Scotia: 0.2 / 0.2
• Consensus: 0.2 / 0.3
• Prior: 0.3 / 0.2

US core inflation warmed up in August as the breadth of price increases rose. Most of the heat came from core services with limited evidence of any obvious tariff pass through effects that could take a while to burn through old pre-tariff inventory, transition through to recent implementation of tariff hikes, and first-round effects on margins awaiting potential pass through. There are serious concerns about data quality that I’ll come back to and I’ll leave it to you to review the disturbing political aftermath.

DETAILS

First the details. Core CPI was up 0.32% m/m SA. That’s the warmest reading since January. It equated to a seasonally adjusted and annualized reading of 3.94% m/m SAAR after the 2.77% reading in June, 1.57% in May, and 2.88% in April (chart 1). These are not light trend readings. The three-month moving average has now pushed up to 2.76% and is rising. 

The breadth of price increases picked up (chart 2). 

Core services—that exclude housing and energy services—accelerated to 0.48% m/m SA which is the hottest reading since January. That works out to 5.9% m/m SAAR (chart 3).

Core goods inflation—that excludes food and energy commodities— was up by 0.21% m/m SA and 2.54% m/m SAAR. That’s not ripping, but it’s also not light (chart 4). 

Owners’ equivalent rent was up by 0.3% m/m SA which is the same as the prior month (chart 5). Rent of primary residence picked up a tenth to 0.3 as well (chart 6). Insurance for renters and homeowners soared (chart 7).

New vehicle prices were flat, and used vehicle prices were up by 0.5% m/m SA after two months of decline (chart 8). Auto insurance was subdued at 0.1% m/m SA (chart 9) and gasoline prices fell 2.2% m/m SA (chart 10).

Food prices were flat as groceries (aka food at home) slipped –0.1% m/m SA and food away from home (eg. Take-out, seated dining etc) was warm again at 0.3% m/m (chart 11).

Apparel prices were subdued at 0.1% m/m SA (chart 12). 

Prices for household furnishings soared again at 0.7% m/m SA after a 1.0% m/m SA gain in June (chart 13).

Travel and leisure related categories like recreation goods, recreation prices and airfare were all warm (charts 14–16).

Also see prescription drug prices, hospital services, and financial services in charts 17–19.

DATA QUALITY ISSUES

There are several data quality issues.

First is that all BLS data from this point onward is sketch given the overt politicization of the agency. CPI, PPI, PCE and nonfarm will have a dark cloud over each of them because of the firing of the former BLS head, her replacement by a politicized Trump acolyte, the fear that likely exists across BLS staff when it comes to giving numbers that Trump may not like, and the possibility of coming methodological changes that will be viewed skeptically.

For example, recall this earlier announcement from the BLS that amid the turmoil they've delayed the rebasing of the CPI figures that were supposed to be delivered with today’s report and with no date set for doing so.

Second is that a large share of the CPI basket is being guesstimated through proxy methods due to the Trump administration’s cuts to the BLS budget. Recall that the issue is that budget cuts impaired the ability to collect data and so they are using alternative methods like using prices from other cities when they can't get them for some cities, or using prices of alternate products deemed somewhat close to products for which they cannot obtain prices etc. We usually get the imputed share of the basket for the latest numbers by about 11:15amET here. It has risen in each of the prior three months to 29% in March, 30% in May, 35% in June to record highs that far surpass even what was guesstimated in the depths of the pandemic when restrictions prevented data collection (chart 20). More detail is here and here. In short, an alarmingly high share of the CPI basket is made up data, even if not for the fact we have a Trump acolyte now running the BLS which impairs the reliability of all US inflation and jobs data henceforth.

Third is that the seasonal adjustments continue to reflect a recency bias that is tamping down the estimate. Chart 21 shows this month’s SA factor that continues the pattern of relatively low SA factors in the pandemic era when comparing like months of July over time. Chart 22 shows that at alternative historical SA factors for July the reading would have been higher.

FOMC IMPLICATIONS

Nothing is settled by this one report as we'll get another round of August inflation data in September before the next FOMC, plus other data (nonfarm) and developments. That said, nothing in here says cut.

MARKET REACTION

So why the market reaction with lower yields on the back of it? That's unclear, but some possibilities:

• markets were braced for something higher yet and covering now
• markets are watching y/y headline readings that were unchanged which would be silly to do imo, versus looking at m/m and also y/y core that accelerated to over 3 (3.1)
• markets didn't see much tariff effect on the commodities ex-food&energy
• markets ignored the rest of the sources of heat behind core and the pick-up in the moving average trend

I don't think the majority of the FOMC will have the same reaction as markets.

Also see charts 23–26 for breakdowns of the basket in y/y and m/m terms and in weighted contribution terms.

Finally, see the accompanying table that provides further detail.