US inflation through year-end: Goldman Sachs expects a two-month core-inflation bump before renewed disinflation into year-end

Overview

Goldman Sachs' US economics team expects inflation data to look less favorable over the next two releases before disinflation resumes later in 2026. The forecast rests on temporary energy and statistical effects in early autumn, followed by softer shelter and insurance components, while core PCE remains more persistent than core CPI.

Core views

Goldman Sachs characterizes the most recent three months of inflation as favorable overall: annualized core CPI averaged 2.0%, while estimated core PCE averaged 2.5%, including anticipated methodological revisions. The team expects a near-term interruption to that improvement. Its preliminary forecasts place monthly core CPI at 0.24% in September and 0.22% in October, compared with 0.29% in August, before slowing to 0.12% in November and 0.11% in December. The first near-term pressure point is energy-related. Goldman Sachs expects higher energy and other commodity costs to add 3-5bp to monthly core inflation for the next few months. Higher oil and refined-product prices are expected to affect airfares most directly: its September airfare nowcast has risen to +3%, using jet-fuel prices and an equity-analyst measure of online fares. The report therefore emphasizes that its headline inflation forecasts, in particular, are highly sensitive to oil prices, with upside risks if oil-market disruptions intensify. The second factor is a technical boost to October core CPI. Because the Bureau of Labor Statistics had missing October data during the prior year's government shutdown, it approximated unadjusted October values using the geometric mean of September and November index values when producing updated seasonal factors. Goldman Sachs argues that this approach conflicts with normal October seasonal patterns across many components and estimates that the resulting residual seasonality will artificially add 5bp to October seasonally adjusted core CPI inflation. October core PCE should receive a slightly smaller boost. After September and October, the report expects a more benign trend. Tariff pass-through is judged mostly complete, having added only about 1bp on average to monthly core CPI in recent months. Shelter inflation has already slowed below its pre-pandemic norm, and Goldman Sachs sees potential for particularly soft November shelter inflation because a May increase appeared concentrated in an outlier-driven Northeast sample; the six-month rotating sample structure means that panel will be resampled in November. The report also expects CPI-specific insurance categories to decline. Auto insurance has fallen 5.3% since December after premiums apparently more than closed the gap with repair and replacement costs, and the team's regulatory-filings model points to modest further net declines through year-end. Health insurance is another expected source of downward pressure on core CPI. Goldman Sachs says narrowed health-insurer profit margins, which affect the CPI health-insurance component with a lag, could lead the October semiannual source-data update to produce larger monthly declines from October 2026 through March 2027. It estimates a 2-3bp monthly subtraction from core CPI over that period. These CPI-specific insurance effects do not translate fully to PCE, because PCE does not use CPI as its source for health and auto insurance data. Accordingly, core PCE is expected to be somewhat more stubborn than core CPI. Goldman Sachs forecasts monthly core PCE increases of 0.23%, 0.23%, 0.16%, and 0.16% from September through December. Upcoming PCE methodology changes should reduce the contribution from the software and accessories category by averaging the current CPI source data with two PPI series that saw more moderate increases during the prior year. However, the report still expects modest sequential upward pressure in that category from the lagged effect of higher memory prices. The report also highlights measurement uncertainty in wireless telephone services. The CPI component has become more volatile since the BLS began using secondary source data last year, lowering June inflation sharply and then more than reversing in August. Goldman Sachs notes that the August rebound brought the index roughly in line with average revenue per user reported by major wireless providers, its proxy for phone-plan prices. PCE is less affected because the wireless component carries roughly half the weight it has in CPI. For the Federal Reserve, the report notes that only the September CPI and August PCE reports will arrive before the October FOMC meeting, while the December meeting will have October CPI and September and October PCE reports. Under Goldman Sachs' forecasts, the inflation-breadth measure cited by Chairman Warsh—the share of PCE prices increasing by more than 3% at a six-month annualized rate—would decline from 49% in July to about 43% in the October data available for the December meeting. That would remain above the 1990-2019 average of 36%, but be well below the post-pandemic peak of 76%; the report also views the current breadth as similar to the measure after removing estimated tariff effects.

Analysis framework

Goldman Sachs builds a component-level inflation forecast, separating temporary energy-price and seasonal-adjustment effects from underlying disinflation in shelter and insurance. It uses nowcasts for airfares, regulatory-filings-based modeling for insurance, CPI and PCE source-methodology differences, historical seasonal patterns, and comparisons with market-implied core CPI paths and long-run inflation-breadth benchmarks.

Methodology notes

• Industry AnalysisVolume-price decomposition

  Component-level CPI and PCE contribution analysis

  The report decomposes monthly core inflation into categories such as airfares, shelter, insurance, software and wireless services to identify the sources of temporary pressure and later disinflation.

• Other

  Seasonal-adjustment and data-methodology analysis

  Goldman Sachs estimates how BLS treatment of shutdown-related missing observations distorts October seasonal factors and assesses how PCE source-data changes alter software and accessories inflation.

• Other

  Inflation breadth measure

  The report tracks the share of PCE components rising more than 3% on a six-month annualized basis to assess how broad inflation pressures are ahead of FOMC meetings.

Key data

• Recent annualized core CPI inflation2.0%Average over the last three months.
• Recent annualized core PCE inflation2.5%Goldman Sachs estimate over the last three months, including anticipated revisions.
• Core CPI forecast, September 20260.24% MoMPreliminary Goldman Sachs forecast.
• Core CPI forecast, October 20260.22% MoMBoosted by energy effects and estimated residual seasonality.
• Core CPI forecast, November and December 20260.12% and 0.11% MoMReflects softer shelter inflation and limited additional tariff pass-through.
• Core PCE forecast, September-December 20260.23%, 0.23%, 0.16%, 0.16% MoMExpected to be somewhat stickier than core CPI.
• Estimated October core CPI seasonal distortion5bpEstimated upward effect from shutdown-related missing-data treatment.
• Energy and commodity contribution3-5bpExpected addition to monthly core inflation for the next few months.
• September airfare nowcast+3%Reflects higher jet-fuel prices and tracked online airfares.
• Auto insurance CPI decline since December5.3%Goldman Sachs expects modest further net declines through year-end.
• Health insurance effect on core CPI-2 to -3bp monthlyExpected from October 2026 through March 2027.
• PCE inflation breadth43%Forecast share of PCE components rising more than 3% annualized over six months in October data available by the December FOMC meeting; versus 49% in July, 36% 1990-2019 average, and 76% post-pandemic peak.

Impact & implications

The report's central implication is that September and October inflation readings may appear firmer than the underlying year-end trend because of oil-linked costs and statistical distortions. Goldman Sachs expects the later-year combination of limited tariff pass-through, softer shelter, and falling CPI insurance components to restore disinflation, although PCE should remain relatively firmer because it does not receive the same insurance-related relief.

Risks

• Inflation forecasts, especially headline forecasts, face upside risk if disruptions to oil markets intensify further.
• Wireless telephone services inflation is an additional source of uncertainty because the CPI component has become more volatile following the use of secondary source data.

What to watch

• September CPI and August PCE data ahead of the October FOMC meeting.
• October CPI and September-October PCE data ahead of the December FOMC meeting.
• Oil, refined-product and jet-fuel prices, given their expected effect on airfares and consumer prices.
• The October CPI seasonal adjustment and the impact of shutdown-related missing data.
• Shelter inflation in November as the potentially outlier-driven May sample rotates out.
• The October health-insurance source-data update and resulting CPI insurance declines.
• The evolution of PCE inflation breadth toward Goldman Sachs' 43% forecast.