A plain-English guide to the advance GDP estimate: three vintages, the line that matters, and three inflation windows.
Written Tuesday November 3, accompanying the October 29 analysis: real GDP grew 3.4 percent at an annual rate, final sales to private domestic purchasers 4.9, the quarterly core PCE price index 2.5 annualised while September's monthly core rose 0.3 and the yearly rate was 3.1 (BEA). Why BEA publishes three estimates, where the revisions come from, why final sales to private domestic purchasers is the line to read first, and how to keep quarterly, monthly and twelve-month inflation rates apart.
Written Tuesday November 3, 2026, accompanying the Thursday October 29 analysis note, which recorded the third-quarter advance GDP estimate and the September personal income report, released in the same half hour. BEA reported that real GDP grew at a 3.4 percent annual rate in the third quarter, after 2.2 in the second; real final sales to private domestic purchasers rose 4.9 percent; the quarterly PCE price index rose at a 2.1 percent annual rate, core 2.5. Twenty minutes of reading later, the same agency's September release said monthly PCE prices rose 0.5 percent, core 0.3, and 3.6 percent on the year. The 5-year yield fell 7 basis points on a strong growth print. This guide explains what the advance estimate is, why it is revised, which lines inside it carry the signal, and how the quarterly and monthly price figures fit together.
Three estimates, one quarter
The Bureau of Economic Analysis publishes each quarter's GDP three times: an advance estimate about four weeks after the quarter ends, a second estimate a month later, and a third a month after that. Each one replaces missing source data with actual data. The advance estimate for the third quarter of 2026 came out on October 29; the second is due in late November and the third in late December (BEA release schedule; confirm the dates there). Every summer, the annual update revises several years at once as benchmark surveys and tax data arrive.
The advance estimate is the one markets trade, and it is the least complete. BEA has to assume values for the parts of the quarter it has not yet seen: the third month's trade data, inventories, and much of services spending are partly estimated. BEA documents which components are assumed in a "key source data and assumptions" table published with each advance release (BEA GDP page). The nowcasting guide covers how the Atlanta and New York Fed models try to anticipate it.
Why it gets revised
Three reasons. Late source data is the obvious one. Seasonal adjustment is the second: each quarter's factors are re-estimated as new data arrive. The third is methodological: BEA periodically redefines what counts, as it did when software and research spending were reclassified as investment. The practical consequence for anyone grading a forecast or a plan against the advance figure is that "Q3 grew 3.4 percent" is a provisional statement that will be rewritten at least twice, and the number in the history books may differ by tenths, occasionally by more. BEA publishes its own revision statistics on the same site; read those rather than trusting any rule of thumb, including ours.
The lines that carry the signal
| Line | Q3 2026 advance | Q2 2026 | What it is for |
|---|---|---|---|
| Real GDP | +3.4% | +2.2% | The headline; includes inventories, trade and government |
| Real final sales to private domestic purchasers | +4.9% | +4.6% | Consumer spending plus private fixed investment; strips out inventories, trade and government |
| PCE price index (quarterly, annualised) | +2.1% | +5.0% | The quarter's inflation at an annual rate |
| Core PCE price index (quarterly, annualised) | +2.5% | +3.3% | Same, excluding food and energy |
Source: BEA, GDP advance estimate, third quarter 2026. All figures are seasonally adjusted annual rates.
Final sales to private domestic purchasers is the line Fed staff and most forecasters watch, because the three things it excludes are the three noisiest: inventory swings reverse, net exports move with import timing, and government spending is set by budgets rather than by the economy. In the third quarter it rose faster than headline GDP, 4.9 against 3.4, which says the domestic private economy was stronger than the headline and that inventories, trade or government subtracted. The analysis note did not break that down, and this guide does not either; the point is that the composition line is the one to read before the headline.
Quarterly annualised vs monthly vs yearly
The same morning produced three inflation numbers that look inconsistent and are not. The quarterly core PCE price index rose at a 2.5 percent annual rate; the September monthly core index rose 0.3 percent; the twelve-month core rate was 3.1 percent (BEA personal income and outlays, September 2026). Each is a different window on the same index:
- Quarterly annualised compares the average index level of July to September with the average of April to June, then compounds to a year. Three months of slow monthly prints in the middle of the quarter can pull it down even if the last month was firm.
- Monthly is September against August. A 0.3 percent month compounds to about 3.7 percent a year, well above the quarterly 2.5. That gap tells you September was hotter than the quarter's average.
- Twelve-month is September 2026 against September 2025, which still contains the hot months from late 2025 that the quarterly figure has dropped.
For reading the trend, the six-month annualised rate sits between the monthly noise and the yearly lag. For the Fed's 2 percent target, the yearly rate is the one in the mandate. For the next FOMC statement, the quarterly figure barely matters: it is a weighted average of monthly prints the committee has already seen. How PCE differs from CPI is in PCE vs CPI.
Why strong growth did not lift yields
The 5-year closed 7.0 basis points lower at 4.801 percent and the 10-year 5.6 lower at 4.989 (Yahoo, from the analysis note), with the move mostly in real yields and mostly after the 8:30 window, which was small. The note offered two untested candidates: the soft quarterly price figures and a give-back of the previous day's post-FOMC sell-off. There is a third, structural one. The advance GDP estimate rarely moves bonds much on its own, because by release day the monthly data underneath it (retail sales, trade, construction, the first two months of PCE) are already public. The new information is the price deflators and the third month, and on October 29 the price side was the softer side. That is a reading, not a finding; with no plan there was nothing to test it against.
A checklist for the next release
- Read final sales to private domestic purchasers before the headline.
- Note which components BEA assumed rather than measured; those are where the second estimate will move.
- Keep the three inflation windows separate; never compare a quarterly annualised rate with a twelve-month one as if they were the same thing.
- Check the prior quarter's figure in the new release: Q2's 2.2 percent is itself a revised number.
- Expect the bond reaction to come from the deflators and the same-morning monthly PCE, not from the growth headline.
Where this shows up in practice
- October 29 (Thursday): GDP 3.4 percent, core PCE 0.3 on the month, the 5-year fell 7 basis points (the live example)
- October 30 (Friday): the employment cost index, the next morning
Related reading
Nothing on this site is investment advice or a recommendation to trade. Most retail traders lose money.