The Thin Edge of the Tariffs Wedge is Slowly Creeping Into US Inflation

• US core inflation accelerated, pumping the 3-mo MA above target

• Where evidence of tariff pass-through is creeping into the data

• Two strong caveats on data quality

• Markets scaled back FOMC cut pricing

• Scotia’s forecast remains for the FOMC to be on hold this year

• US CPI / core CPI, m/m % change, SA, June:
• Actual: 0.29 / 0.23
• Scotia: 0.2 / 0.3
• Consensus: 0.3 / 0.3
• Prior: 0.1 / 0.1

Core CPI inflation edged up a bit last month to 2.77% m/m at a seasonally adjusted and annualized rate (chart 1). The past three months have been 2.9%, 1.6% and 2.8% for a three-month moving average of 2.4%. The FOMC is likely to say that’s hardly screaming out for rate cuts regardless of what the administration says. Furthermore, core inflation would have been higher yet if not for a cooked SA factor, while further evidence points to impaired data reliability due to BLS funding cuts.

This is also how markets generally reacted to the data. The US 2-year yield moved up by about 3–4bps post-data. Markets continue to price no move by the Fed on July 30th and edged September cut pricing a little lower to about 14bps.and full-year cut pricing was ratcheted back by about 5bps to well under a half point of cumulative cuts priced by the end of 2025. Our forecast remains no cut this year.

Some of the hottest CPI categories were groceries, energy, clothing, recreation goods and home furnishings, each of which may have had tariff-driven factors behind them given significant import-propensities (chart 2). In seasonally adjusted terms, fruits and vegetables were up 11.5% m/m SAAR, home fuel prices were up by 12.8%, gasoline jumped 12.8%, clothing was up 5.3%, home furnishings jumped by 12.4% with significant breadth across sub-categories like furniture and appliances, and recreation goods were up 9.7% and also with high breadth. 

Core services inflation which excludes energy and housing related components was subdued this time (chart 3). 

This time, the gain was driven by an acceleration of core goods CPI to 2.4% m/m SAAR (chart 4).

Shelter costs were up by 2.1% m/m SAAR with rent up 2.75% and owners’ equivalent rent up 3.7% alongside higher home heating costs. The perennial wish for cooling market rents to show up in key measures of rent in CPI remains largely elusive.

POOR QUALITY DATA AT THE WORST POSSIBLE MOMENT

US inflation data reliability is highly questionable at the worst possible moment in time. There are two reasons for this.

One is that one of the biggest seasonally unadjusted increases in core CPI compared to like months of June (chart 5) was tamped down by the seasonal adjustment factor that was applied to June core CPI which was the lowest on record compared to like months of June in history (chart 6). At any other June SA factor used in history, this June’s core CPI reading would have been noticeably warmer (chart 7). The SA factor is calculated with a strong recency bias that is skewed to developments over the latest years. All of the pandemic-era SA factors have been the lowest on record. Having faith in these estimates requires having confidence that today’s seasonality is fundamentally different than ever before and that the distortions introduced by the pandemic’s rolling shocks and the timing of stimulus measures still apply in 2025. 

Secondly, the share of the CPI basket estimated by proxy methods instead of hard data increased again. After two months in which that share was at 30%—double the prior record in the depths of the pandemic when agents couldn’t be sent into the field to collect prices—it now stands at 35% (chart 8). The BLS emphasizes the role played by the Trump administration’s budget cuts for such under-sampling and hence the reliance upon proxy methods like substituting prices for goods using comparable goods, and prices for goods from other markets when in-market prices cannot be sampled. Proving that this distorts CPI is impossible, but it would stand to reason that one should be extremely guarded toward the quality of the data when so much of the basket is being inferred and will never be revised since the data has fundamentally gone awol.

Simply put, it is wrong to blindly ignore serious question marks surrounding the quality of US inflation data at the worst possible moment.

DETAILS

Charts 9–22 break down individual components of the basket to show trends.

Charts 23–24 ranked the y/y % price changes across CPI components in raw terms and weighted contributions to CPI respectively.

Charts 25–26 do likewise for m/m price changes.

Please also see the accompanying table with further details, micro-charts and measures of deviations from historical tendencies.