Direct Answer

CPI and PCE are both measures of consumer price inflation but use different baskets, different weighting methodologies, and capture different populations of spending. CPI measures the price change of a fixed basket of goods and services for urban consumers, as defined by the Bureau of Labor Statistics. PCE, Personal Consumption Expenditures, published by the Bureau of Economic Analysis: uses a chain-weighted formula that allows the basket to shift over time as consumers substitute between goods, and also captures healthcare spending paid by employers or government on consumers' behalf. PCE typically runs 0.3-0.5 percentage points per year lower than CPI. The Federal Reserve's official inflation target is 2% PCE; it uses PCE because it is less susceptible to substitution bias and better represents the full consumption basket.

Despite the Fed's PCE mandate, financial markets react more intensely to CPI because it is released earlier in the month and its fixed basket creates sharper month-to-month price swings that are more immediately visible. Core measures, which strip out food and energy, exist because food and energy prices are highly volatile and driven by supply shocks rather than monetary policy. "Sticky" inflation components (shelter, healthcare, insurance) change slowly and are more resistant to Fed tightening; "flexible" components (gasoline, airfares, used cars) can reverse quickly. Traders and Fed watchers monitor sticky vs. flexible decompositions to assess whether underlying inflation pressure is durable or transitory.

Key Takeaways

  • PCE is the Fed's target, CPI moves markets: The Fed's 2% target is PCE, but CPI is released 2-3 weeks earlier and generates larger short-term market reactions because of its fixed-basket volatility.
  • Core strips volatile components: Core CPI and Core PCE exclude food and energy, isolating structural demand-driven price pressure from supply-side noise. The Fed focuses heavily on core PCE as its policy guide.
  • Sticky vs. flexible components matter more than headline: Shelter (owners' equivalent rent), medical services, and insurance are sticky and slow to turn; gasoline, airfares, and apparel are flexible and can reverse quickly. A high headline with flexible drivers is less durably inflationary than a high headline with sticky drivers.
  • Month-over-month is more policy-relevant than year-over-year: Annualized 3-month trimmed-mean and supercore measures have become the Fed's preferred real-time gauges because year-over-year comparisons are distorted by base effects from high/low months 12 months prior.
  • Supercore = services ex-shelter: The Fed popularized "supercore" CPI/PCE in 2022-2023, services inflation excluding shelter, as the most direct measure of demand-driven inflation tied to the labor market and wage growth.
  • Positive CPI surprise in hiking cycle = bonds sell, equities sell: When inflation beats consensus during a period of active Fed tightening, both bonds (higher rates) and equities (higher discount rate, lower multiples) typically sell off simultaneously.
  • Base effects create predictable calendar seasonality: YoY CPI comparisons are distorted by the month 12 periods ago. A favorable base (a high-print month rolling off) artificially compresses YoY readings; an unfavorable base (a low-print rolling off) inflates them.
  • Owners' equivalent rent lags actual rent by ~12-18 months: The BLS methodology for shelter inflation surveys what homeowners estimate they would charge to rent their home, creating a significant lag versus real-time asking-rent data from sources like Zillow or Apartment List.

Core Concepts

CPI Construction and Its Bias Toward Overstating Inflation

The BLS computes CPI by tracking the prices of a fixed basket of goods and services selected based on the Consumer Expenditure Survey. The basket weights are updated every two years (more recently annually), but between updates they remain fixed. This creates substitution bias: when beef becomes expensive and consumers buy more chicken, the fixed basket keeps a high weight on beef, overstating how much the average consumer's cost of living has risen because it doesn't capture the substitution.

CPI also uses a geometric mean for lower-level aggregations (below the category level) but an arithmetic mean at higher levels, creating an additional upward bias. The Boskin Commission in 1996 estimated CPI overstated true inflation by approximately 1.1 percentage points annually from multiple sources of bias. BLS has addressed some but not all of these methodologically since then.

The practical implication for traders: CPI readings are systematically higher than PCE readings for structural reasons, not because of genuine differences in underlying inflation. A CPI reading of 3.5% is roughly comparable to a PCE reading of 3.0-3.2% in terms of actual price pressure. Market participants who compare CPI directly to the Fed's 2% PCE target without this adjustment will consistently overestimate how far inflation is from target.

To verify this in the data, download the Federal Reserve's FRED database series for CPI (CPIAUCSL) and PCE (PCEPI) and compute the rolling 12-month difference. The PCE-CPI gap typically ranges from 0.2% to 0.7% per year, with PCE running lower. This gap has been remarkably stable across time despite major changes in the composition of the baskets.

PCE's Advantages and Why the Fed Chose It

PCE uses a Fisher Ideal chain-weighted price index, which allows the basket weights to evolve as consumer spending patterns change. This directly addresses the substitution bias problem in CPI. PCE also covers a broader population, not just urban consumers but all US households, and captures healthcare spending paid by employers and government on behalf of consumers, which is a major and growing share of total consumption that CPI misses.

The practical consequence of PCE's broader healthcare coverage is that it tends to show lower healthcare inflation than CPI because employer and government negotiated rates grow more slowly than out-of-pocket consumer prices. This is a real economic phenomenon worth tracking separately: the gap between what people actually pay for healthcare out of pocket versus what the economy as a whole pays for healthcare is important for both policy and portfolio positioning in healthcare sectors.

The Fed switched from CPI to PCE as its official mandate measure in 2000, following the Boskin Commission's findings. The 2% PCE target has been in place since the Fed formally stated the target in 2012. Understanding this history matters for reading Fed communications: when Fed officials say "inflation is at 2%," they mean PCE, not CPI.

Sticky vs. Flexible Inflation and the Durability Question

The Atlanta Fed publishes a decomposition of CPI into sticky and flexible components based on the frequency with which prices change in BLS data. Flexible components, gasoline, airfares, used vehicles, apparel, hotel rates, change price frequently (monthly or more). Their prices respond quickly to supply and demand shocks and can reverse sharply. Sticky components, owners' equivalent rent, medical services, education, insurance, change price infrequently and tend to persist once they change.

During the 2021-2023 inflation episode, flexible CPI (which had driven the initial surge in goods prices) peaked and reversed by mid-2022 as supply chains normalized. Sticky CPI remained elevated through 2023-2024 and was significantly harder to reduce. This decomposition explained why the Fed maintained restrictive rates well into 2024 even as headline inflation fell, the sticky components had not returned to pre-pandemic levels and were the last mile of the inflation problem.

Traders who monitor sticky vs. flexible decomposition have a forward-looking tool: when sticky inflation is falling but flexible is rising, the overall trend is still disinflationary. When sticky inflation is rising, the Fed is unlikely to pivot regardless of what happens to the volatile components. The Atlanta Fed Sticky-Price CPI series is freely available on FRED.

Supercore and the Labor Market Inflation Link

During Chair Powell's November 2022 Brookings speech, he introduced "supercore" inflation, services inflation excluding shelter, as the Fed's preferred real-time gauge of demand-driven inflationary pressure. The logic: shelter inflation lags actual market rents by 12-18 months due to the OER methodology, distorting the real-time read. Commodity goods prices had already rolled over. The remaining inflationary pressure was concentrated in services, which is tightly linked to labor costs.

Supercore PCE is calculated by taking core PCE (ex-food and energy) and further excluding housing services. Historically, supercore has been the most persistent component of inflation and the hardest to reduce without a significant increase in unemployment. Its correlation with nominal wage growth is among the highest of any inflation sub-component, which is why the Fed explicitly linked the labor market (payrolls, wages) to the inflation outlook rather than treating them as independent datasets.

For traders, a supercore reading above the equivalent of 2% annualized rate signals continued tightening pressure on the Fed regardless of what headline or core does. A sustained decline in supercore, driven by decelerating wage growth, is a more reliable disinflation signal than a fall in flexible components alone.

Worked Scenario

  1. Setup: The BLS is releasing the February CPI report. Prior month core CPI was +0.3% MoM. Bloomberg consensus for February core CPI is +0.3% MoM. Year-over-year consensus is 3.1%. The Fed is in a data-dependent pause, with markets pricing one 25bp cut by December.
  2. The print: Core CPI prints +0.4% MoM. Year-over-year rises to 3.2%. Shelter was +0.6% MoM, services ex-shelter was +0.3%. Flexible components were mixed, with used vehicles down but airfares up.
  3. Immediate reaction: 2-year Treasury yield spikes 12 basis points (markets reprice the near-term rate path). 10-year yield rises 9 basis points. S&P 500 futures fall 0.8-1.2%. Dollar index strengthens 0.4%.
  4. Secondary analysis: Traders who decompose the report note: shelter contributed +0.22% to the +0.4% MoM, consistent with the lagged OER effect that has been well-telegraphed. Supercore (services ex-shelter) was +0.3% MoM, annualizing to ~3.6%, still elevated but not reaccelerating. This nuance emerges in commentary within the hour.
  5. Markets recalibrate: After the initial shock, equities recover half the decline as analysts note the shelter-driven nature of the miss and the in-line supercore reading. December cut pricing collapses from 75% to 40%. The 2-year yield stays elevated but 10-year yields narrow slightly as the market sees less second-round risk from the miss.
  6. PCE context: When the PCE report for the same month is released two weeks later, it prints +0.3% core MoM, in line with consensus. Markets barely react because PCE was anticipated to be softer than CPI given the healthcare and basket differences, confirming the two-index relationship. The Fed's preferred gauge remains closer to target than the CPI print suggested.

Measurement Framework

MeasurementQuestion to Answer
Core CPI MoM (month-over-month)What is the current momentum of underlying inflation, stripping volatile food and energy?
Core PCE MoMWhat is the Fed's preferred real-time inflation signal?
Supercore PCE (services ex-shelter)Is demand-driven, labor-market-linked inflation decelerating?
Atlanta Fed Sticky CPI vs. Flexible CPIIs the remaining inflation driven by durable or transient components?
Owners' Equivalent Rent MoM vs. Zillow/Apartment List asking rentsDoes real-time rent data suggest OER will lead or lag the official shelter reading?
3-month annualized core PCEIs the underlying trend accelerating or decelerating versus the year-over-year reading?

Common Failure Modes

Comparing CPI to the Fed's 2% PCE Target Directly

CPI and PCE are not directly comparable because of structural differences in basket, weighting, and scope. CPI at 3.0% is roughly equivalent to PCE at 2.5-2.6%. Traders who watch CPI touch 2.5% and declare "inflation is at the Fed's target" are making a systematic error, PCE is still likely above 2.0% at that CPI reading, and the Fed won't cut based on CPI touching 2.5%.

Yellow daisies placed on a financial newspaper on stone pavement during sunset.
Photo by Elif via Pexels

Always translate CPI readings to their PCE equivalent before assessing proximity to the Fed's target. The historical PCE-CPI gap is approximately 0.3-0.5pp annually, PCE below CPI.

Anchoring on Year-Over-Year Readings During Base Effect Distortions

Year-over-year inflation readings are highly distorted by the "base effect", the comparison to the same month 12 periods prior. If last June had an unusually high monthly print (e.g., +0.7% MoM), June of the following year will show a favorable base effect that mechanically reduces the year-over-year reading even if nothing has changed in underlying inflation momentum.

Professional inflation analysts track month-over-month readings and 3-month annualized rates as the policy-relevant signal. Base effects are calendar artifacts. Markets sometimes react to YoY misses or beats that are entirely attributable to a known base effect, creating a brief mispricing that quickly corrects when the base effect is recognized.

Treating All Shelter Inflation as Durable

Shelter inflation in CPI (OER + rent of primary residence) was the primary source of elevated sticky inflation in 2023-2024. However, real-time rent data from Zillow, Apartment List, and CoStar showed new lease asking rents turning negative or flat from mid-2022 onward, a full 12-18 months before OER began to decline in CPI. Traders who assumed OER would stay elevated indefinitely misread the forward trajectory.

The leading indicator for OER is new lease rent data, not OER itself. The BLS methodology captures the average of all existing leases (new and renewal), which turns with a long lag. When new lease rents have been falling for 12+ months, OER's eventual decline is nearly certain, the only question is timing.

Underestimating Services Inflation Persistence

Services inflation, particularly supercore, is driven primarily by labor costs, which are stickier than goods prices. When a services business sets prices (a barber, an insurance company, a restaurant). It is primarily responding to labor cost pressures that only change with wage cycles, not to short-term supply shocks. Expecting services inflation to decline quickly when wage growth remains elevated is a consistent error in inflation forecasting.

Services ex-shelter will not normalize until wage growth (measured by average hourly earnings, Employment Cost Index, or Atlanta Fed Wage Growth Tracker) normalizes. This linkage is the mechanistic reason the Fed cares deeply about the labor market even when discussing inflation.

Frequently Asked Questions

Why does the Fed use PCE instead of CPI?

The Federal Reserve adopted PCE as its official inflation target measure in 2000, following the Boskin Commission's finding that CPI overstated true inflation due to substitution bias, new product bias, and quality adjustment issues. PCE's chain-weighted methodology avoids substitution bias, its broader consumption coverage is more representative of the full economy, and its healthcare scope captures employer and government healthcare spending that affects economic outcomes but is invisible in CPI. The Fed formalized the 2% PCE target in its January 2012 statement of longer-run goals.

What is "supercore" inflation and why does it matter?

Supercore inflation is services inflation excluding shelter (housing services). It was highlighted by Federal Reserve Chair Jerome Powell in November 2022 as the Fed's preferred real-time measure of demand-driven inflation, because it is the component most directly tied to labor market conditions and wage growth. Shelter inflation was excluded because OER lags actual market rents by 12-18 months, distorting the real-time read. Goods inflation was already falling. Supercore is the "last mile" of inflation that requires a cooler labor market to normalize. Available on FRED as "PCE: Services Less Energy and Housing" (DPCCRV1Q225SBEA or monthly series).

How large is the typical gap between CPI and PCE?

Over the past 25 years, core PCE has run approximately 0.3-0.5 percentage points per year below core CPI. During the 2021-2023 inflation episode, this gap widened slightly because the fixed-basket CPI was more exposed to goods price surges. At the peak in mid-2022, headline CPI reached 9.1% YoY while headline PCE peaked at approximately 7.0% YoY, a gap of over 2 percentage points, driven partly by the energy and food basket differences and partly by the PCE chain-weighting effect on goods. The gap is structural rather than cyclical, but its width varies somewhat with the composition of inflation at any given time.

What is the Atlanta Fed Sticky CPI and where can I find it?

The Atlanta Fed publishes a decomposition of CPI into "sticky" prices (those that change infrequently, less than once a year on average) and "flexible" prices (those that change frequently). Sticky CPI includes items like shelter, healthcare, tuition, and household services. Flexible CPI includes energy, food, used vehicles, apparel, and airfares. The data is available free on the Atlanta Fed website (atlantafed.org/research/inflationproject) and on FRED. The Atlanta Fed also publishes a trimmed-mean CPI that removes the highest and lowest outlier components to isolate central tendency.

Why does OER lag real-time rents by so long?

Owners' Equivalent Rent represents what homeowners estimate they would charge to rent their own home, collected by the BLS Consumer Expenditure Survey. Because most housing leases are 12 months in length, the rent that the average occupied unit pays reflects a weighted average of leases signed over the past 12 months, not today's market rate. New leases represent only about 1/12th of the monthly sample weight. This means that when market rents (measured by new lease signings) peak and turn, OER continues rising for 12-18 months as the cohort of higher-rent leases signed near the peak works through the sample. The BLS is aware of this limitation and has experimented with new-tenant-only rent indexes, but OER remains the official measure.

What is a "base effect" and how does it distort year-over-year readings?

A base effect occurs when the year-over-year comparison is influenced by an unusually high or low reading 12 months prior, independent of current inflation momentum. For example, if energy prices surged in March of year one (+8% MoM), then even if energy prices are flat in March of year two, the year-over-year energy component will show a -7.4% contribution, artificially depressing headline YoY inflation. Base effects are calendar artifacts, they do not reflect a change in current price dynamics. To adjust for base effects, focus on month-over-month readings and 3-month or 6-month annualized rates, which measure current momentum rather than comparison distortion.

How does CPI affect the yield curve?

A CPI surprise above consensus typically steepens the front end of the yield curve by pushing short-term yields (2-year) higher as markets price additional Fed hikes or delay rate cut expectations. Longer-term yields (10-year, 30-year) also rise but by somewhat less, because the long end reflects both near-term rate expectations (which go up with an inflation surprise) and longer-term growth and inflation equilibrium (which may not change as much from a single data point). If inflation surprises persistently to the upside across multiple releases, the entire curve reprices higher, often with a bear-flattening (the front rising more than the back) that is characteristic of tightening cycles.

Is there a way to forecast CPI before the release?

Yes, imperfectly. Several methods exist: commodity price trackers (gasoline prices published daily by AAA or EIA), real-time retailer scanner data (reflected in the Cleveland Fed's Inflation Nowcasting model), Truflation and PriceStats (private real-time price indexes), Zillow and Apartment List for shelter, and the BLS import price index (released 2 weeks before CPI) for goods categories. The Cleveland Fed Inflation Nowcasting model is freely available and has historically outperformed the survey consensus in forecasting accuracy, particularly for months with large energy swings.

How does seasonal adjustment affect a monthly inflation print?

Monthly figures are adjusted to remove recurring calendar patterns, so the published change is not the raw price change consumers experienced. The adjustment factors are re-estimated periodically using recent history, which means a period of unusual price behavior can feed back into the factors and shift subsequent adjusted readings. Comparing an adjusted monthly change with an unadjusted year-over-year change mixes two conventions, and the difference between them can be large enough to change the interpretation of a single month.

References

  • Bureau of Labor Statistics. Consumer Price Index Overview: Official methodology, release schedule, and data for CPI, core CPI, and component series.
  • Bureau of Economic Analysis. PCE Price Index: Official source for PCE and core PCE, the Fed's preferred inflation measure.
  • Federal Reserve Bank of Atlanta. Inflation Project: Sticky-Price CPI, Flexible-Price CPI, and Trimmed-Mean CPI series.
  • Federal Reserve Bank of Cleveland. Inflation Nowcasting: Real-time CPI and PCE forecasts ahead of official releases.
  • Powell, J. (2022, November 30). "Inflation and the Labor Market." Brookings Institution., Speech introducing supercore as the Fed's preferred inflation gauge.

Educational Disclaimer

This guide is for educational purposes only. Inflation data interpretation involves uncertainty and professional judgment. Do not make investment decisions based solely on this content. Trading involves risk of loss. Consult a qualified financial professional for personalized advice.