CPI vs PCE Inflation: A Visual Comparison

CPI and PCE both measure inflation, but they do not measure the same spending universe in exactly the same way. CPI focuses mainly on prices paid directly by urban consumers. PCE covers a broader range of household consumption, including some spending made on behalf of households. The indexes also use different formulas and different spending weights. Because of those differences, CPI and PCE can show different inflation rates for the same month.

The Federal Reserve sets its longer run inflation objective using the annual change in the PCE price index. That does not make CPI unimportant. CPI remains one of the most widely followed measures of consumer prices.

This visual explainer shows what each index measures, why the numbers can differ, how headline and core measures compare, and which measure fits different questions.

CPI vs PCE at a glance

The simplest way to think about the difference is this:

CPI asks how prices changed for the goods and services bought directly by urban consumers.

PCE asks how prices changed for a broader set of goods and services consumed by people, including some purchases made on their behalf.

That broader PCE scope matters most in areas such as health care. The formula and weight system also matter because they change how category price movements affect the final index.

Source: BLS Consumer Price Index, BEA PCE Price Index, and BEA explanation of CPI and PCE differences.

What is CPI?

The Consumer Price Index is produced by the U.S. Bureau of Labor Statistics.

CPI measures changes in prices paid by consumers for goods and services. The most common national version is CPI for All Urban Consumers, often called CPI U. BLS says this population group represents more than 90 percent of the U.S. population.

Headline CPI includes food and energy.

Core CPI removes food and energy. Analysts often look at core CPI because food and energy prices can move sharply from month to month. Core CPI is useful for studying broader price pressure, but it is not a replacement for headline CPI. Households still pay for food and energy.

A CPI index level is not the same thing as an inflation rate.

For example, an index level tells you where the price index stands relative to its base period. A 12 month inflation rate tells you how much the index changed compared with the same month one year earlier.

For normal 12 month comparisons, this article uses not seasonally adjusted CPI series. For month to month changes, seasonally adjusted CPI is more useful because it removes recurring seasonal patterns.

What is the PCE price index?

The Personal Consumption Expenditures price index is produced by the U.S. Bureau of Economic Analysis.

PCE measures prices for goods and services purchased by people in the United States. Its scope is broader than CPI because it includes some spending made on behalf of households.

Health care is a useful example.

CPI mainly captures household out of pocket medical spending in its covered consumer scope. PCE also includes broader medical spending paid by third parties, such as some government programs and employer related payments.

PCE is part of the National Income and Product Accounts. These accounts connect consumer spending with the wider national economy.

Headline PCE includes food and energy.

Core PCE excludes food and energy.

The Federal Reserve defines its longer run 2 percent inflation objective using the annual change in the PCE price index.

What is the main difference between CPI and PCE?

The difference comes from more than one rule.

BEA groups the main sources of difference into four categories:

1. Formula effect

2. Weight effect

3. Scope effect

4. Other effects

Different formulas

PCE uses a Fisher Ideal formula.

CPI uses a modified Laspeyres approach.

You do not need advanced mathematics to understand the practical result.

The PCE method can respond more quickly when consumers change what they buy. CPI also updates its weights, but the index structure is different.

This does not mean people can always replace one item with another easily. Rent, medicine, electricity, and many services are not simple substitutes.

The key point is that the two indexes process changing spending patterns differently.

Different weights

A category has more influence when it has a larger weight in the index.

Imagine that housing prices rise 4 percent while another category rises 1 percent. If housing has more weight in one index, that 4 percent increase will have more effect on that index.

This is one reason shelter often matters so much for CPI.

PCE gives different relative importance to several categories because its spending data come from a broader national accounts framework.

Different scope

Scope asks which spending belongs inside the index.

CPI focuses mainly on household out of pocket spending for its covered urban population.

PCE measures spending by and on behalf of the personal sector, including households and nonprofit institutions serving households.

This is why medical care can have a different role in PCE.

The price of a hospital service may affect household consumption even when another party pays much of the bill.

Other measurement differences

BEA also identifies other effects.

These can include seasonal adjustment differences, price source differences, and residual measurement differences.

So when CPI and PCE do not match, there is rarely one single reason.

Why CPI and PCE differ

The BLS comparison research describes weight and scope differences directly. It notes that CPI expenditure data are sourced from consumers, while PCE relies more on business information within the national accounts system.

CPI vs PCE latest inflation comparison

As of August 9, 2026, June 2026 is the latest month available for both CPI and PCE.

The July CPI report is scheduled for August 12, 2026. The July PCE report is scheduled for August 26, 2026.

For June 2026, headline CPI rose 3.5 percent over 12 months. Core CPI rose 2.6 percent.

Headline PCE rose 3.7 percent over 12 months. Core PCE rose 3.3 percent.

That means PCE was above CPI for both headline and core inflation in this particular month.

This is a useful reminder that CPI does not always have to run above PCE.

Source: BLS CPI June 2026 release and BEA Personal Income and Outlays, June 2026.

The dumbbell chart makes the June comparison easy to see.

Headline PCE was 0.2 percentage points above headline CPI.

Core PCE was 0.7 percentage points above core CPI.

Those are percentage point differences. They are not percent differences.

What happened from May to June 2026?

The monthly picture also differed.

Seasonally adjusted headline CPI fell 0.4 percent in June.

Core CPI was unchanged.

The PCE price index fell 0.1 percent in June, while core PCE rose 0.1 percent.

A one month move can be strongly affected by energy, travel, or another volatile category.

In June 2026, BLS reported that the energy index fell sharply and was the largest contributor to the monthly decline in headline CPI.

That does not mean the general price level returned to where it was one year earlier. Headline CPI was still 3.5 percent higher than in June 2025.

CPI minus PCE inflation gap

A useful comparison is:

CPI 12 month inflation minus PCE 12 month inflation

A positive value means CPI inflation is higher.

A negative value means PCE inflation is higher.

For June 2026:

Headline gap = 3.5 minus 3.7 = negative 0.2 percentage points.

Core gap = 2.6 minus 3.3 = negative 0.7 percentage points.

The gap changes over time.

Do not build a rule that says CPI must always be higher or PCE must always be lower.

The better approach is to calculate the gap over a long period and see when the sign changes.

Headline CPI vs headline PCE over time

The most useful historical chart compares 12 month headline inflation from both indexes on the same monthly date axis.

The chart should begin in January 2000 and continue through the latest common month.

Because both BLS and BEA revise some data, the reproducible approach is better than freezing an old historical image forever.

Use not seasonally adjusted CPI for the 12 month CPI calculation.

For PCE, use the monthly PCE price index from BEA and label the seasonal status returned by the source.

The formula is:

12 month inflation = 100 × ((index this month / index 12 months ago) minus 1)

The following Python pattern retrieves CPI directly from the BLS API.

import os
import requests
import pandas as pd

BLS_URL = "https://api.bls.gov/publicAPI/v2/timeseries/data/"
BLS_KEY = os.getenv("BLS_API_KEY")

CPI_SERIES = {
    "headline_cpi": "CUUR0000SA0",
    "core_cpi": "CUUR0000SA0L1E",
}

def fetch_bls_block(start_year, end_year):
    payload = {
        "seriesid": list(CPI_SERIES.values()),
        "startyear": str(start_year),
        "endyear": str(end_year),
        "registrationkey": BLS_KEY,
    }

    response = requests.post(
        BLS_URL,
        json=payload,
        timeout=30,
    )
    response.raise_for_status()
    data = response.json()

    if data["status"] != "REQUEST_SUCCEEDED":
        raise RuntimeError(data.get("message"))

    rows = []

    reverse = {
        value: key
        for key, value in CPI_SERIES.items()
    }

    for series in data["Results"]["series"]:
        name = reverse[series["seriesID"]]

        for obs in series["data"]:
            if not obs["period"].startswith("M"):
                continue

            month = int(obs["period"][1:])

            rows.append({
                "series": name,
                "date": pd.Timestamp(
                    year=int(obs["year"]),
                    month=month,
                    day=1,
                ),
                "index": float(obs["value"]),
            })

    return pd.DataFrame(rows)

cpi = pd.concat([
    fetch_bls_block(2000, 2019),
    fetch_bls_block(2020, 2026),
], ignore_index=True)

cpi = (
    cpi
    .sort_values(["series", "date"])
    .drop_duplicates(["series", "date"])
)

cpi["inflation_12m"] = (
    cpi
    .groupby("series")["index"]
    .pct_change(12, fill_method=None)
    * 100
)

The registered BLS API supports requests covering up to 20 years at a time, so the example splits the historical pull into two blocks.

Pull monthly PCE data from BEA

Use the BEA NIPA dataset and discover the current table metadata rather than assuming a line number forever.

The April 2026 BEA API guide documents monthly NIPA requests with DataSetName=NIPA, TableName, Frequency=M, and Year.

A practical table for monthly PCE price indexes is NIPA Table 2.8.4.

import os
import requests
import pandas as pd

BEA_URL = "https://apps.bea.gov/api/data"
BEA_KEY = os.getenv("BEA_API_KEY")

def bea_request(params):
    response = requests.get(
        BEA_URL,
        params={
            "UserID": BEA_KEY,
            "ResultFormat": "JSON",
            **params,
        },
        timeout=30,
    )
    response.raise_for_status()

    payload = response.json()
    results = payload["BEAAPI"]["Results"]

    if "Error" in results:
        raise RuntimeError(results["Error"])

    return results

pce_results = bea_request({
    "method": "GetData",
    "DataSetName": "NIPA",
    "TableName": "T20804",
    "Frequency": "M",
    "Year": ",".join(
        str(year)
        for year in range(2000, 2027)
    ),
})

pce_raw = pd.DataFrame(
    pce_results["Data"]
)

print(
    pce_raw[
        ["LineNumber", "LineDescription"]
    ]
    .drop_duplicates()
    .to_string(index=False)
)

Inspect the current line descriptions before selecting headline and core PCE.

Do not guess a line number from an old article.

After the correct rows are identified, convert DataValue to numeric and calculate the 12 month change.

pce_raw["value"] = pd.to_numeric(
    pce_raw["DataValue"]
    .str.replace(",", "", regex=False),
    errors="coerce",
)

def parse_bea_month(text):
    year = int(text[:4])
    month = int(text.split("M")[1])
    return pd.Timestamp(year=year, month=month, day=1)

pce_raw["date"] = (
    pce_raw["TimePeriod"]
    .map(parse_bea_month)
)

Then join CPI and PCE on the same monthly date.

Never compare a newer CPI month with an older PCE month as though they are the same period.

Core CPI vs core PCE over time

The second historical plot should show core CPI and core PCE from January 2000 through the latest common month.

The calculation is the same.

The difference is the series.

Core CPI excludes food and energy.

Core PCE also excludes food and energy.

That does not make the two core indexes identical because the formula, scope, and weights still differ.

A clear chart should use the same vertical unit for both lines.

Avoid placing all four headline and core series in one chart if the result becomes difficult to read.

Build the historical inflation gap chart

Once both data sources are aligned, calculate:

comparison["headline_gap_pp"] = (
    comparison["headline_cpi_12m"]
    - comparison["headline_pce_12m"]
)

comparison["core_gap_pp"] = (
    comparison["core_cpi_12m"]
    - comparison["core_pce_12m"]
)

Plot both gaps with a zero line.

import matplotlib.pyplot as plt

fig, ax = plt.subplots(figsize=(11, 6))

ax.plot(
    comparison["date"],
    comparison["headline_gap_pp"],
    label="Headline CPI minus PCE",
)

ax.plot(
    comparison["date"],
    comparison["core_gap_pp"],
    label="Core CPI minus PCE",
)

ax.axhline(0, linewidth=1)

ax.set_title(
    "CPI Minus PCE Inflation Gap"
)

ax.set_ylabel("Percentage points")
ax.set_xlabel("Date")
ax.legend()
ax.grid(alpha=0.25)

fig.tight_layout()
plt.show()

Then calculate average gaps over clearly stated periods.

for start in ["2000-01-01", "2010-01-01", "2020-01-01"]:
    sample = comparison[
        comparison["date"] >= start
    ]

    print(
        start,
        sample[
            [
                "headline_gap_pp",
                "core_gap_pp",
            ]
        ].mean()
    )

Generate these averages from live data at publication time.

Do not type them from memory.

CPI vs PCE inflation race

A visual animation can make the comparison easier to understand.

The recommended animation has two rows:

Headline inflation

Core inflation

Each row has one CPI point and one PCE point.

Each frame represents one real month.

Use January 2020 through the latest common month.

The axis must stay fixed across every frame.

Do not rescale each month because that can make small movements look much larger than they are.

The data should come from the same aligned table used for the static charts.

import plotly.express as px

long = comparison.melt(
    id_vars=["date"],
    value_vars=[
        "headline_cpi_12m",
        "headline_pce_12m",
        "core_cpi_12m",
        "core_pce_12m",
    ],
    var_name="measure",
    value_name="inflation",
)

long["group"] = long["measure"].map({
    "headline_cpi_12m": "Headline",
    "headline_pce_12m": "Headline",
    "core_cpi_12m": "Core",
    "core_pce_12m": "Core",
})

long["index"] = long["measure"].map({
    "headline_cpi_12m": "CPI",
    "headline_pce_12m": "PCE",
    "core_cpi_12m": "CPI",
    "core_pce_12m": "PCE",
})

long["month"] = (
    long["date"]
    .dt.strftime("%b %Y")
)

animation = px.scatter(
    long[
        long["date"] >= "2020-01-01"
    ],
    x="inflation",
    y="group",
    color="index",
    animation_frame="month",
    animation_group="measure",
    range_x=[
        long["inflation"].min() - 1,
        long["inflation"].max() + 1,
    ],
    title="CPI vs PCE Inflation Race",
)

animation.write_html(
    "cpi-vs-pce-inflation-race.html"
)

animation.show()

The article HTML can also use a short current month animation as a visual preview.

The full historical animation should be regenerated from actual monthly data before publication.

Why does the Federal Reserve use PCE?

The Federal Reserve says that inflation of 2 percent over the longer run, measured by the annual change in the PCE price index, is most consistent with its mandate for maximum employment and price stability.

PCE offers broader spending coverage.

Its formula also responds more readily to changing expenditure patterns.

These features make it useful for broad monetary policy analysis.

This does not mean the Federal Reserve ignores CPI.

CPI is released earlier in the month and provides an important view of consumer prices.

Markets often react strongly to CPI because it is timely and widely followed.

A careful policy analysis normally looks at CPI, PCE, wage growth, inflation expectations, labor market conditions, and other price measures together.

Which inflation measure should you use?

There is no single correct answer for every question.

If the question is about the prices consumers directly face at the store, in rent, or in other household purchases, CPI is often the natural starting point.

If the question is about the inflation measure used for the Federal Reserve target, use PCE.

If the question is about the broad inflation environment, compare both.

Headline CPI vs headline PCE

Headline measures include food and energy.

These categories matter because households buy them frequently and can notice their price changes quickly.

Food and energy can also move sharply.

That can create large monthly moves in headline inflation.

Do not treat this as bad data.

Headline inflation answers a real question about the full price basket.

The June 2026 CPI report is a clear example.

Headline CPI fell on a monthly basis, largely because energy prices dropped sharply, even though the index remained higher than one year earlier.

Core CPI vs core PCE

Core CPI and core PCE remove food and energy.

Analysts use them to study price pressure outside two categories that can swing sharply.

Core inflation can be more stable than headline inflation over short periods.

But core inflation is not automatically a forecast.

It is one view of the inflation process.

Core CPI and core PCE can also differ from each other because their category weights and spending scope are not the same.

Housing and health care explain part of the difference

Housing and health care are two useful examples because they show why the indexes can react differently even when they are describing the same economy.

Housing

Shelter has a large role in CPI.

CPI includes rent paid by tenants and owners equivalent rent, which estimates the rental value of owner occupied housing.

Housing changes can move slowly and stay persistent.

That can keep CPI inflation elevated even when some goods prices are cooling.

PCE also includes housing services, but the relative weight is different.

Health care

Health care shows the scope difference clearly.

CPI mainly measures medical spending paid directly by households within its consumer scope.

PCE has broader medical consumption coverage and includes some spending paid on behalf of households.

That can give medical care more influence in PCE.

Do not compare category percentages unless the categories are mapped carefully.

CPI and PCE categories do not always line up one for one.

Are CPI and PCE revised?

Yes, but the revision process differs.

PCE is part of the national economic accounts.

BEA can revise historical estimates when new source data or updated methods become available.

BEA also conducts annual updates.

CPI has a different revision pattern.

Not seasonally adjusted CPI indexes are generally not revised after publication, apart from certain exceptional corrections and preliminary chained CPI treatment.

Seasonally adjusted CPI data can change.

BLS recalculates seasonal factors each year. The current BLS guidance says the new factors are used to revise the previous five years of seasonally adjusted data.

This matters when you calculate month to month inflation from seasonally adjusted CPI.

Save your retrieval date if you need reproducible research.

A reproducible CPI vs PCE data record

Save the exact BEA line descriptions and line numbers used when the data are downloaded.

That makes later updates easier to audit.

Validation before publishing

Run these checks before any chart or table is published.

1. Confirm there are no duplicate month records.

2. Sort all observations by date before calculating changes.

3. Confirm every 12 month inflation rate uses exactly 12 months of lag.

4. Confirm CPI and PCE are joined on the same reference month.

5. Check that headline and core labels are not mixed.

6. Label seasonally adjusted and not seasonally adjusted data.

7. Use percentage points for the difference between two inflation rates.

8. Confirm the latest values against the official BLS and BEA release pages.

9. Keep chart units consistent.

10. Keep animation axis limits fixed across frames.

11. Save retrieval dates.

12. Record revision notes.

A simple merge check can catch timing mistakes.

common = cpi_frame.merge(
    pce_frame,
    on="date",
    how="inner",
    validate="one_to_one",
)

if common.empty:
    raise ValueError(
        "No common CPI and PCE months found."
    )

latest_common = common["date"].max()
print(latest_common)

Common mistakes in CPI vs PCE comparisons

Comparing different months

CPI is normally released before PCE for the same reference month.

If July CPI is available but July PCE is not, the latest common comparison is still June.

Label any newer CPI value separately.

Comparing index levels

CPI and PCE index levels use different base systems.

Do not subtract raw index levels.

Compare inflation rates.

Mixing monthly and 12 month rates

A monthly change of 0.3 percent is not the same thing as 3.6 percent annual inflation.

Keep the time period in every label.

Mixing seasonal status

Month to month CPI analysis should normally use seasonally adjusted values.

Standard 12 month CPI comparisons can use not seasonally adjusted values.

State the choice.

Calling a percentage point gap a percent gap

If CPI inflation is 3.5 percent and PCE inflation is 3.7 percent, the gap is negative 0.2 percentage points.

Saying one measure is always higher

The June 2026 data show why this is unsafe.

Both headline and core PCE inflation were above their CPI counterparts.

Treating lower inflation as lower prices

If inflation falls from 5 percent to 3 percent, prices are still rising on average.

They are rising more slowly.

Prices fall on average only when the inflation rate is negative.

Frequently asked questions

What is the main difference between CPI and PCE?

CPI focuses mainly on prices paid directly by urban consumers. PCE covers a broader set of household consumption, including some purchases made on behalf of households. The two indexes also use different formulas and spending weights.

Why does the Federal Reserve prefer PCE?

The Federal Reserve defines its longer run 2 percent inflation objective using the PCE price index. PCE has broader consumption coverage and a formula that responds more readily to changing spending patterns.

Is PCE usually lower than CPI?

Not always.

The average relationship depends on the period, and the gap can change sign.

In June 2026, headline and core PCE inflation were both higher than the comparable CPI rates.

What is core CPI?

Core CPI is CPI excluding food and energy.

It is used to study price changes outside two categories that can move sharply.

What is core PCE?

Core PCE is the PCE price index excluding food and energy.

It is closely watched in monetary policy analysis.

Why do housing costs affect CPI and PCE differently?

Housing has different relative weight in the two indexes.

CPI shelter measures also have an especially important role in CPI.

Why does health care have a different impact?

PCE includes broader health care spending, including some spending paid on behalf of households. CPI focuses more on consumer out of pocket medical spending within its scope.

Which measure is better for consumers?

CPI is often the better starting point for direct consumer price experience.

That does not mean it matches every household.

Every household has a different spending pattern.

Which inflation measure does the Federal Reserve target?

The Federal Reserve uses the annual change in the PCE price index for its longer run 2 percent inflation objective.

Can CPI and PCE move in opposite directions?

Yes.

Monthly changes can differ because of weights, scope, formulas, and other measurement differences.

Are CPI and PCE revised?

PCE estimates can be revised as part of the national accounts.

Seasonally adjusted CPI can also be revised when BLS updates seasonal factors.

When are CPI and PCE released?

Both are monthly releases.

CPI normally comes earlier.

PCE is released later as part of the Personal Income and Outlays report.

Should investors watch CPI or PCE?

Both can matter.

CPI provides an earlier view of consumer inflation.

PCE is the measure tied to the Federal Reserve inflation objective.

Why should economists compare both?

The gap itself can reveal how category weights, spending scope, and price patterns are changing.

Using both gives a broader view than using only one number.

Methodology and data notes

This article uses official BLS and BEA definitions as the primary authority.

Headline CPI is CPI for All Urban Consumers, All Items.

Core CPI is CPI for All Urban Consumers, All Items Less Food and Energy.

For 12 month CPI comparisons, the recommended series are CUUR0000SA0 and CUUR0000SA0L1E, which are not seasonally adjusted.

For month to month CPI comparisons, the seasonally adjusted counterparts are appropriate.

PCE data should be retrieved from the BEA NIPA monthly tables. The code inspects current line descriptions instead of trusting a line number copied from an old tutorial.

The 12 month inflation formula is:

100 × ((index this month / index 12 months ago) minus 1)

The monthly change formula is:

100 × ((index this month / index previous month) minus 1)

The CPI minus PCE gap is reported in percentage points.

The latest common month is determined after joining CPI and PCE on their actual reference month.

Missing values remain missing. They are not replaced with zero.

Retrieval date for the current snapshot: August 9, 2026.

Limitations

CPI and PCE are national inflation measures.

They do not measure the personal inflation rate of every household.

A household that spends a large share of income on rent may experience price pressure differently from a homeowner with no mortgage.

A family with large medical expenses may have a different experience from a household with very little medical spending.

Category definitions are also not perfectly identical across CPI and PCE.

That makes detailed category comparisons harder than the top line comparison.

Historical PCE estimates can be revised.

Seasonally adjusted CPI can also be revised through updated seasonal factors.

Finally, a lower inflation rate does not mean prices returned to old levels.

It means the rate of increase slowed, unless inflation became negative.

Final answer

CPI and PCE are not competing versions of the same exact basket.

They answer related but different questions.

CPI is especially useful for understanding prices paid directly by urban consumers.

PCE covers a broader range of household consumption and is the inflation measure used for the Federal Reserve longer run 2 percent objective.

The best approach is not to choose one number and ignore the other.

Compare both, align the reference month, keep the methodology clear, and use the gap as information rather than treating it as an error.

Official sources

1. U.S. Bureau of Labor Statistics, Consumer Price Index

2. BLS CPI June 2026 release

3. BLS CPI seasonal adjustment guidance

4. BLS Public Data API Version 2

5. BLS research comparing PCE and CPI

6. U.S. Bureau of Economic Analysis, PCE Price Index

7. BEA Personal Income and Outlays, June 2026

8. BEA explanation of CPI and PCE differences

9. BEA API for Data Retrieval User Guide

10. Federal Reserve explanation of the 2 percent PCE inflation objective

Last updated: August 9, 2026