Why Consumer Spending Drives So Much of the Economy

Personal consumption expenditures (PCE), the Bureau of Economic Analysis (BEA) measure of what US households buy, were about 67.5% of gross domestic product (GDP) in the second quarter of 2026, up from roughly 60% in 1970. Spending dominates because it is the largest part of final demand, and the income that funds it comes from the production it pays for. The share is built partly from imputed and third-party-paid items, and it is revised after first publication.

Key takeaways

PCE was $21,993.2 billion of $32,563.0 billion nominal GDP in Q2 2026 (67.5%), against a 1970 average of 60.3%, based on FRED copies of BEA data downloaded on October 20, 2026.
Services are 68.8% of PCE and durable goods 10.9%. Services held up far better than durable goods in 2008–09 but fell harder in 2020.
PCE includes imputed purchases such as owner-occupied housing and care paid for by employers and government programs, so it is wider than what shoppers pay at a till.
Spending is read alongside income, saving and credit: in August 2026 real PCE was up 2.6% from a year earlier while real disposable income was up 1.3%.

How big the spending share is

BEA's methodology handbook describes PCE as the primary measure of consumer spending and says it accounts for about two-thirds of domestic final spending. The exact share is a division of two published series, nominal PCE by nominal GDP, both in current dollars at seasonally adjusted annual rates.

PCE share of GDP = nominal PCE / nominal GDP x 100

Q2 2026: 21,993.2 / 32,563.0 x 100 = 67.5%
Q1 2026: 21,526.0 / 31,906.3 x 100 = 67.5%

The figures come from the St. Louis Fed's FRED copies of BEA's national accounts: personal consumption expenditures (PCEC) and gross domestic product (GDP). They reflect the BEA annual update published on September 30, 2026, and they will move again at the next update.

The share since 1970

The share has not been constant. Averaging the four quarters of 1970 gives 60.3%. It reached 66.2% in the first quarter of 2000 and 68.8% in the third quarter of 2009, the highest quarter in the series, which starts in 1947. Since 2010 it has stayed between about 66% (Q2 2020) and 69%.

Period	PCE share of nominal GDP
1970 (average of four quarters)	60.3%
Q1 1990	63.7%
Q1 2000	66.2%
Q3 2009 (series high)	68.8%
Q2 2026	67.5%

A higher share does not by itself mean households are healthier. GDP also includes investment, government purchases and net exports, so the share can rise when any of those weaken.

What households actually buy

Inside PCE, the mix has shifted toward services. Using the FRED series for durable goods, nondurable goods and services, the split in two quarters looks like this.

Category	Q1 1970, share of PCE	Q2 2026, share of PCE	Q2 2026, $ billions (annual rate)
Durable goods (cars, appliances)	14.2%	10.9%	2,390.8
Nondurable goods (food, clothing, fuel)	35.5%	20.3%	4,470.6
Services (housing, health care, recreation)	50.3%	68.8%	15,131.8
Total PCE	100%	100%	21,993.2

Services alone equal 46.5% of GDP in Q2 2026 (15,131.8 / 32,563.0). That matters for how the data behave. A household can postpone a car purchase, but rent, insurance premiums and medical care are harder to put off. In current dollars, from Q3 2008 to Q2 2009 durable goods spending fell 9.2% while services fell 0.9%. The pattern reversed when services themselves were restricted: from Q4 2019 to Q2 2020 services fell 13.0% and durable goods 5.3%. A large services base steadies spending in a typical downturn, not in every one.

Why the 70% headline needs care

Headlines often say consumers are "70% of the economy", but the series has never gone above 68.8%. More important is what PCE contains. According to the BEA handbook, PCE covers purchases by households and by nonprofit institutions serving households, and it also includes expenditures financed by third-party payers on behalf of households, such as employer-paid health insurance and medical care financed through government programs. BEA's own consumer spending explainer gives the same kind of example: an insurance company paying part of a hospital bill counts.

The handbook also says PCE includes imputed purchases that keep the total unchanged by how an activity is arranged, for example whether housing is rented or owned, or whether employees are paid in cash or in kind. BEA explains why in its note on GDP imputations: an owner-occupied home is valued at what the owner could have earned by renting it out. No cash changes hands, yet the service counts as spending.

PCE is therefore an accounting measure of consumption, not a count of what people paid from their own wallets, and comparing it with a household budget mixes definitions.

How spending feeds back into income

Each dollar a household spends becomes revenue for a business, which pays wages, suppliers and owners, and some of that becomes household income again. The San Francisco Fed defines the marginal propensity to consume (MPC) as the fraction of an extra dollar that a person spends, with the rest going to saving or debt repayment.

Illustrative example (round numbers, not real data): suppose households receive an extra $100 transfer and every household spends 60% of each new dollar, so the MPC is 0.6. The first $60 spent becomes someone's income, of which $36 is spent, then $21.60, then $12.96, and so on. The letter notes that in a simple Keynesian model the multiplier on transfers equals MPC/(1 − MPC).

Extra spending = 100 x (0.6 + 0.6^2 + 0.6^3 + ...) = 100 x 0.6 / (1 - 0.6) = 150
Extra income   = the first 100 + 150 of spending that becomes income = 250

After six rounds of spending: 60 + 36 + 21.6 + 12.96 + 7.78 + 4.67 = 143.0 (about 95% of 150)

With an MPC of 0.6, each transfer dollar ends up as $1.50 of spending. The model's lesson is limited: taxes, imports, prices and policy reactions change how much spending stays in the loop. The same San Francisco Fed letter concludes that the MPC on transfers is particularly high when unemployment is high and liquidity constraints bind, which is a reminder that the loop is not a fixed number.

The Federal Open Market Committee's September 16, 2026 statement said domestic spending had been resilient. BEA's schedule puts the advance estimate of third-quarter 2026 GDP and the September personal income and outlays report on October 29, 2026.

Revisions, nominal and real figures

How the estimates are revised

BEA's Survey of Current Business article on revisions describes the advance estimate arriving about one month after a quarter ends, the second about two months after and the third about three months after. An annual update, held in September in the years after 2018, can revise up to five years. For real GDP growth over 1999 to 2022, the mean absolute revision was 0.45 percentage points from advance to second estimate, 0.58 from advance to third and 0.28 from second to third.

The latest quarter shows the scale. BEA's Q2 2026 third estimate, released September 30, 2026, put real GDP growth at an annual rate of 2.2%, up 0.7 percentage point from 1.5% in the second estimate, mainly on higher investment, consumer spending and government spending. The second-to-third revision of 0.7 point is two and a half times the 0.28-point average for that step, so a single early print deserves caution.

Contribution to growth

BEA reports how many percentage points of real growth each component adds. The FRED series for PCE's contribution to real GDP growth shows 2.51 percentage points (at an annual rate) in Q2 2026 and 0.49 in Q1 2026, against total real GDP growth of 2.2% and 2.5%. In Q2, then, PCE added more than the whole of GDP growth, because other components netted to a negative. Because the contribution depends on both the growth of PCE and its weight, the same spending growth contributes less when the share is lower.

Nominal versus real

Nominal spending includes price changes; real spending removes them. Monthly PCE in August 2026 was $22,308.5 billion at an annual rate, up 6.1% from $21,029.5 billion a year earlier. BEA reported the PCE price index up 3.4% over the same period, and real PCE (chained 2017 dollars) rose 2.6%, from $16,529.7 billion to $16,955.3 billion. Roughly 6.1% nominal growth contained about 3.4% of price change and 2.6% of extra quantity (1.034 x 1.0257 = 1.0606). The purchasing power calculator applies the same real-versus-nominal logic to a dollar amount over time.

Retail sales are a different measure

The Census Bureau's monthly retail survey covers firms in retail trade (NAICS 44–45) and food services (722), per its survey description. Most services in PCE (68.8% of the total), such as housing, health care and insurance, fall outside that scope; restaurants and bars are the main overlap, so retail sales are a partial guide to PCE. The advance estimate for September 2026, released October 15, was $745.0 billion, up 0.8% from August and 6.3% from September 2025, not adjusted for price changes. The monthly rise carries a margin of sampling error of plus or minus 0.4 percentage point, a 90% confidence range in Census's release.

Credit, saving and household debt

Spending can outrun income when households save less or borrow more. In BEA's August 2026 personal income and outlays release, real disposable personal income was unchanged on the month while real PCE rose 0.6%. The saving rate was 4.1% of disposable income. Using FRED's personal saving rate and real disposable personal income series, the saving rate was 5.2% in August 2025, and real disposable income rose 1.3% over the 12 months to August 2026, half the 2.6% growth in real PCE.

The saving rate is itself revised. BEA's July 2026 release first showed 3.0%; the current FRED series, after the September 30 annual update, shows 4.6% for July.

Two lender-side sources give the borrowing side. The Fed's July 2026 Senior Loan Officer Opinion Survey, covering the second quarter, found a modest net share of banks tightening standards on credit card loans, with auto and other consumer loan standards basically unchanged and weaker demand for auto loans. The New York Fed's Q2 2026 household debt report, released August 11, put total household debt at $18.771 trillion, down $13 billion in the quarter, with credit card balances at $1.263 trillion and 4.7% of outstanding debt in some stage of delinquency.

Frequently asked questions

Is consumer spending the same as retail sales?

No. Retail sales cover retailers, including online sellers, and restaurants and bars, while PCE also includes services such as health care and housing, and imputed items. The two series use different sources, so they can move in different directions in a given month.

Why does the spending share in the news differ from 67.5%?

Outlets use nominal or real series, quarterly or annual data, and different vintages, and many round to "70%". The share shifts slightly with each annual update, so cite the quarter and the vintage.

What does a quarterly growth rate of 2.2% mean?

BEA expresses quarterly growth as an annual rate, so the figure shows what growth would be if the quarter's pace lasted a year. For Q2 2026, BEA's release puts the quarterly rate at 0.6%, which corresponds to the 2.2% annual rate.

Does buying a new house count as consumer spending?

No. In BEA's treatment, the purchase of a new house is investment. What enters PCE is the housing service that flows from the home over its life: rent paid by tenants, or, for owners, an imputed value based on what the home could earn if rented out.

Where can I check the latest release dates?

Each BEA release states the next date; the August 2026 income and outlays release lists the September report for October 29, 2026. Census and the Fed publish separate schedules for retail sales and the SLOOS.

A checklist for each new release

Note the vintage: advance, second or third estimate, or post-annual-update, and the release date.
Check whether the figure is nominal or real, and the annualization used.
Split spending into durable goods, nondurable goods and services before drawing a conclusion.
Compare real PCE growth with real disposable income and the saving rate.
Check credit evidence: bank standards in the SLOOS and delinquency in the New York Fed report.
Save the series ID and download date, because the same quarter can read differently after the next update.