Global core inflation persistence: US inflation appears less persistent than headline core measures suggest

Overview

This global macro report tests whether the US is truly the main remaining inflation problem by separating core inflation into goods, non-shelter services and shelter. Goldman Sachs concludes that the apparent US outlier status is misleading: much of the US overshoot is likely temporary or measurement-related, whereas stickier service-sector pressures remain more pronounced across other economies.

Core views

Goldman Sachs begins with the observation that core inflation is only modestly above target in most countries but remains more elevated in the US, particularly in PCE terms. Rather than relying on headline core measures, it decomposes inflation into core goods, non-shelter core services and shelter. This decomposition leads the report to the opposite conclusion from the headline comparison: the US may have less of a persistent inflation problem than the rest of the world. For core goods, inflation has broadly returned to long-run trends in emerging markets and non-US developed markets, but US PCE core-goods inflation remains about 3 percentage points above its pre-pandemic trend, versus 1 percentage point in US CPI. Goldman Sachs attributes all of the PCE overshoot to two temporary US-specific factors. Tariffs are estimated to add 2.4pp to year-over-year core PCE goods inflation, an effect expected to mostly disappear by H2 of next year. Separately, rising memory prices are producing a large AI-related and mismeasured boost because PCE gives a much larger weight to software and accessories than other inflation measures, likely including some business purchases at retail outlets; the category is adding 1pp to year-over-year PCE core-goods inflation. The report expects this impulse to fade in 2027 as further memory-price upside becomes limited and anticipated BEA methodology changes reduce the category's PCE weight. Outside the US, stronger Chinese export supply and weaker Chinese import demand have freed goods supply for global markets. Goldman Sachs estimates imports from China lowered goods prices by 0.8% on average in non-US developed markets and by 0.6% in major emerging markets, using data through 2026Q2. It expects this structural disinflationary tailwind to persist under its China team's forecast of a further increase in China's current-account surplus. The report argues that this should more than offset limited global energy-spillover pressures since the Iran war and a modest, correctly measured AI-related goods-price boost worldwide. Tariffs mean the US benefits less directly from Chinese disinflation, but the report still views underlying US core-goods inflation as benign once one-off effects are removed. For non-shelter services, official data show elevated inflation in emerging markets and most developed markets, including especially high readings in US PCE. Goldman Sachs harmonizes the treatment of catering, medical and financial services across developed markets, excluding financial-service prices that are measured differently and can be sensitive to US equity-market movements, while aligning health-insurance treatment. On this basis, US core-services inflation is lower than in other large developed markets. Unit labor-cost growth, which the report identifies as the primary medium-term driver of non-shelter services inflation, is close to its pre-pandemic average in the US but elevated elsewhere. This points to greater risk of sustained service-sector inflation outside the US. Shelter inflation has fully normalized in the US and emerging markets on average, while remaining elevated relative to long-run trends in other developed markets. Market-rent indicators, which are timelier than official measures and avoid some cross-country differences in owner-occupied housing measurement, imply a more benign US rental-inflation outlook than elsewhere. Goldman Sachs links stronger progress in the US, Canada and New Zealand partly to weaker housing demand after sharper pullbacks in immigration; developed markets with less immigration retrenchment have made less rental-inflation progress since 2023. To judge inflation relative to central-bank targets, the report estimates target-consistent inflation rates for each component and country. Since productivity growth is normally faster in goods than services, target-consistent goods inflation tends to be below the overall target and target-consistent services inflation somewhat above it. The resulting country comparisons show non-shelter services generally above target-consistent levels across non-US developed markets, CEEMEA and Latin America. Core goods are more evenly distributed above and below target-consistent levels: US PCE, Brazil and Colombia are the largest upside outliers, while France and Poland are the largest downside outliers. Shelter appears less concerning on this basis, suggesting that elevated shelter readings in non-US developed markets may be less likely to keep overall inflation above target as dynamics normalize. At the aggregate level, inflation is furthest above target in Latin America, CEEMEA and US PCE, while it is more benign in emerging Asia. Services account for most of the global core-inflation overshoot. Brazil, the US on a core-PCE basis, and Australia are exceptions where goods account for a larger share; in the US, Goldman Sachs again attributes this mainly to tariffs and AI-related measurement bias. US PCE non-shelter services are also affected by unusual healthcare and financial-services treatment, while market rents suggest further downside to US rental inflation. Because service inflation is generally stickier than goods inflation, the report concludes that the US faces a less persistent inflation challenge than headline core measures imply.

Analysis framework

Goldman Sachs compares cross-country core inflation by decomposing it into goods, non-shelter services and shelter. It then adjusts service measures for cross-country classification differences, assesses goods-price drivers including tariffs, AI-related measurement and Chinese supply, uses unit labor costs to assess medium-term service inflation, and benchmarks each component against estimated target-consistent rates.

Methodology notes

• Other

  Component-level target-consistent inflation calculation

  Using a Dallas Fed economist methodology, the report calculates the goods, services and shelter inflation rates consistent with each central bank's target while holding historical non-recessionary component spreads constant. This is used to identify which components are driving above-target inflation in each country.

• Industry AnalysisVolume-price decomposition

  Core-inflation basket decomposition

  The report separates core inflation into goods, non-shelter services and shelter so that temporary price effects can be distinguished from potentially stickier service and housing pressures.

• Other

  Harmonized cross-country non-shelter services measure

  Goldman Sachs adjusts the treatment of catering, medical and financial services across developed markets to make the US and international service-inflation readings more comparable.

Key data

• US PCE core-goods inflation versus pre-pandemic trendAround 3pp above trendCompared with 1pp above trend in US CPI core-goods inflation.
• Tariff contribution to US year-over-year core PCE goods inflation2.4ppExpected to mostly vanish by H2 of next year.
• PCE software-and-accessories contribution to year-over-year core-goods inflation1ppAttributed to AI-related memory-price increases and measurement distortions; the comparable effect in other developed markets is less than 0.1pp.
• Estimated Chinese-import effect on goods prices-0.8% in non-US developed markets; -0.6% in major emerging marketsEstimate incorporates trade data through 2026Q2.
• Earlier estimated Chinese-import effect in non-US developed markets-0.7%Estimate covered the two years from 2024Q1 to 2026Q1.

Impact & implications

The report argues that policy-relevant inflation persistence is concentrated more in services outside the US than in US goods inflation. It therefore views the US headline core-inflation overshoot as disproportionately affected by temporary tariffs, AI-related measurement issues and category-specific PCE treatment, while labor-cost and service pressures make inflation elsewhere more durable.

Risks

• The target-consistent inflation method assumes that historical relationships between goods, services and shelter inflation remain stable; it does not capture changes in those relationships over time.

What to watch

• Whether the tariff effect on US core PCE goods inflation fades by H2 of next year.
• Whether AI-related memory-price effects fade in 2027 and whether BEA methodology changes reduce the PCE weight of software and accessories.
• Further Chinese current-account-surplus growth and its effect on global goods supply and disinflation.
• Unit labor-cost growth and non-shelter services inflation outside the US.
• Market-rent trends and the effect of immigration changes on rental inflation across developed markets.