US real GDP growth

Q6 forecast — September 14, 2026

Central forecast: 3.6% annualized growth in the 2026Q3 advance estimate, 2.2% in 2026Q4, and 2.1% per quarter thereafter. These are marginal medians, not a prediction that realized growth will be smooth. The 80% intervals widen from 0.6%–6.2% for 2026Q3 to -2.2%–6.6% for 2031Q3.

Evidence cutoff: September 14, 2026. The latest dated economic evidence used is the September 11 New York Fed nowcast and BLS inflation release. I also use the Polymarket recession price observed on September 14, with low weight because its resolution differs from this question.

Reference class and calibration

My reference class is U.S. first-release quarterly real GDP growth, not annual growth or subsequently revised GDP. From the Philadelphia Fed's 27-quarter first-release sample, 2018Q4–2025Q2, I calculate approximately 19% negative quarters, a 2.6% median, and empirical 10th/90th percentiles of approximately -1.1%/6.4%. This small, turbulent sample is a scale check, not a sufficient forecasting model. The same publication's longer 1985Q1–2023Q1 evaluation gives SPF first-release RMSEs of 1.89 percentage points for the current quarter and 4.31 points four quarters ahead, versus corresponding mean absolute errors of 1.23 and 1.93 points. That contrast supports retaining extreme tails rather than treating all uncertainty as Gaussian. The evaluation file is dated August 22, 2025.

I therefore start with a distribution centered around ordinary expansion, with meaningful contraction and unusually strong-growth tails. Near-term observations shift and narrow that distribution. Beyond approximately a year, uncertainty increasingly reflects the business cycle and structural change, rather than uncertainty about today's monthly indicators. The eventual 8.8-point interval is somewhat wider than the recent sample's empirical central range. These are 80% intervals separately for each quarter, not an 80% simultaneous band for the entire path.

Current status and the near-term forecast

None of the requested quarters has a published first estimate yet. BEA's schedule lists the 2026Q3 advance release for October 29, 2026, at 8:30 a.m. Eastern. The accessible schedule currently covers 2026, so I retain the question's late-January expectation for the following quarter rather than inventing a confirmed 2027 date.

The principal observations are:

Evidence	Observation	Publication/update
BEA 2026Q2 advance estimate	Real GDP 1.5%; real final sales to private domestic purchasers 3.9%	July 30, 2026
BEA 2026Q2 second estimate	GDP still 1.5%; private domestic final sales revised to 4.2%	August 26, 2026
Atlanta Fed GDPNow, 2026Q3	4.4%, down from 4.7% on September 3	September 10, 2026
New York Fed Staff Nowcast, 2026Q3	2.3%	September 11, 2026

The second estimate is useful information about the economy, but it would not replace an advance-estimate outcome under these rules.

The August 14 Survey of Professional Forecasters supplies the main expert-survey anchor: 2.5% for 2026Q3, 2.3% for Q4, 2.1% for both 2027Q1 and Q2, and 2.2% for Q3. Its mean probabilities of negative growth were 13.3% in 2026Q3 and approximately 20%–22% over the following four quarters.

My Q3 center is approximately a 60% GDPNow / 25% New York Fed / 15% SPF blend: 3.59%, rounded to 3.6%. These are judgmental weights, not a fitted ensemble. GDPNow gets the largest weight because it explicitly tracks the BEA first-release accounting components; the competing nowcast and survey prevent me from treating its unusually strong investment projection as certain. Atlanta's September 10 estimate includes 3.6% consumption growth and 19.1% gross private domestic investment growth. Its published final-nowcast historical RMSE is 1.17 points; today's forecast is still well before the final update, so a much wider interval is appropriate.

The underlying evidence is mixed rather than recessionary. August payrolls rose 162,000, with unemployment 4.1%, although the previous twelve months averaged only 31,000 additional jobs per month. July real consumption was essentially flat, following 0.4% growth in June. The July trade deficit widened to $88.6 billion, with computer-related imports prominent. BEA also warns that gold trade receives a different treatment in GDP, so the nominal trade deficit should not be mechanically translated into a growth contribution.

For Q4, I put much less weight on extrapolating Q3's investment surge. My 2.2% median is close to the SPF's 2.3%, with allowance for financing and purchasing-power headwinds. The September 11 CPI release reported August headline inflation of 0.4% month over month and 3.4% year over year. In July, the FOMC held its target at 3.50%–3.75%, with three dissenters preferring an increase. Thus, rapid monetary easing is not a safe baseline assumption.

Outside forecasts, markets, and the distant path

Three independent longer-term anchors favor modest expansion:

• The June 17 FOMC projections put real growth at 2.3% in 2027, 2.2% in 2028, and 2.0% in the longer run. These are annual Q4/Q4 projections, not quarterly advance-estimate forecasts.
• CBO's February 11 outlook averages 1.8% growth over 2027–2036. Its potential-growth estimates average 2.1% in 2026–2030 and 1.8% in 2031–2036. I checked its outlook page for subsequent updates; the August tariff-budget update is not a replacement full economic baseline.
• Deloitte's July 1 forecast expects 1.8% growth in 2027 and 2.1% in 2030, with stronger expected AI-related productivity than in its previous forecast. Its scenarios explicitly contrast sustained investment with an oil shock and investment retrenchment.

On September 14, the accessible Polymarket “US recession by end of 2026?” contract displayed approximately 9%, with about $1.75 million cumulative volume. That contract allows either a specified sequence of negative GDP estimates or an NBER announcement, and uses a different vintage convention. I treat it as evidence against an imminent deep downturn, not as a 9% probability of any particular requested quarter being negative. I found a matching Kalshi Q3 GDP listing, but could not verify executable threshold prices from the accessible page/API, so I do not invent a market-implied GDP distribution.

My 2.1% distant median is a synthesis of these anchors, not a direct conversion of annual forecasts into quarterly data. I make no claim to know the timing of a recession five years ahead. Aging and slower labor-force growth restrain the center; productivity and capital deepening provide an offset. AI-related investment creates both an upside pathway through productive capacity and a downside pathway if expected returns fail to materialize. The widening tails also accommodate policy changes, energy shocks, and first-release measurement uncertainty. These adjustments are my judgments, informed by the CBO and Deloitte mechanisms.

Forecasting the publication and the correct vintage

BEA's September 30, 2026 annual update is an important intervening event. Its August 17 announcement describes changes to measurement of portfolio-management services, legal services, and software/accessories. Those changes can alter the levels and composition used when the Q3 advance estimate is constructed. Ignoring revisions to the target after publication does not mean freezing the preceding quarter at its own advance vintage. I do not assume a directional effect from the forthcoming methodological changes.

The relevant comparable case is 2025Q3: the shutdown-delayed first estimate was published on December 23, 2025, labeled “Initial Estimate,” at 4.3%. It replaced both the originally scheduled advance and second releases and combined source information normally used in those estimates. Under this question's fallback rule, that first publication would be the target—not a later revised value. The 2025Q4 advance release was also delayed, from January 29 to February 20, 2026.

BEA's November 2025 comparison reports average absolute real-growth revisions of 0.5 points from advance to second, 0.6 to third, and 1.2 to latest. I consequently use first-release evidence where available rather than calibrating solely on today's revised historical series. I also avoid adding all eventual revision variance on top of a model already evaluated against first releases.

The publication mechanism is primarily scheduled statistical production, not discretionary corporate disclosure. BEA's published policy emphasizes simultaneous public dissemination and separation of statistical releases from policy interpretation. The principal publication risks I allow for are source-data delays, shutdowns, and methodological changes; I found no basis for a directional adjustment premised on politically motivated suppression.

Strongest cases against the central estimate

The strongest upward case is that my Q3 shrinkage is excessive: GDPNow's 4.4%, strong private domestic final demand, and the improved labor reading may be better indicators than the lower survey/factor-model estimates. Longer term, AI productivity could raise trend growth much more than the approximately 2% institutional baselines allow. This argues for preserving a substantial upper tail, but not for making a technology-led growth acceleration the median without stronger aggregate evidence.

The strongest downward case is that investment concentration masks fragile consumption, while persistent energy inflation and financing costs produce a sharper slowdown. Deloitte's downside scenario illustrates the combination of an investment reversal and an oil shock. I account for this through negative lower quantiles from Q4 onward rather than scheduling a recession as the central outcome.

Evidence in the next 90 days that would cause a major update

The Q3 interval's half-width is 2.8 points; Q4's is 3.2 points. Changes larger than these require more than an ordinary monthly surprise:

1. A corroborated nowcast collapse or surge: August/September consumption, investment, inventory, and trade data push both major nowcasts below 0% or above 7%, with the accounting contributions explaining the move. That could shift the Q3 median by more than 2.8 points. A large September 30 benchmark change could be part of this evidence.
2. A severe demand/credit shock: repeated monthly real-consumption declines around 1% or more, broad declines in hours and output, sharply rising unemployment, and clear credit disruption would justify moving Q4 toward contraction below -1.5%, a change exceeding its current half-width. A similarly large, verified positive demand surprise would move it upward.
3. A persistent structural break: for distant quarters, a change exceeding the roughly 4–4.4-point half-width would require extraordinary evidence—such as independently corroborated, economy-wide productivity gains consistent with a durable growth rate above roughly 6.5%–7%, or a lasting major loss of productive capacity. A single strong AI investment quarter would not suffice.
4. The resolving release: once BEA publishes the October 29 Q3 advance estimate, that period becomes the reported number, with no subsequent substitution of revised estimates. Ordinary intervening inflation or FOMC surprises would usually lead to smaller updates, not the large changes above.

Source URLs and publication dates

• Philadelphia Fed, first-release forecast evaluation and realizations, August 22, 2025: https://www.philadelphiafed.org/-/media/FRBP/Assets/Surveys-And-Data/survey-of-professional-forecasters/data-files/RGDP/SPF_Error_Statistics_RGDP_3_AICt.txt
• BEA release schedule, checked September 14, 2026: https://www.bea.gov/news/schedule
• BEA Q2 advance, July 30, 2026: https://www.bea.gov/news/2026/gdp-advance-estimate-2nd-quarter-2026
• BEA Q2 second estimate, August 26, 2026: https://www.bea.gov/news/2026/gdp-second-estimate-and-corporate-profits-2nd-quarter-2026
• Atlanta Fed commentary, September 10, 2026, and methodology/accuracy page: https://www.atlantafed.org/research-and-data/data/gdpnow/current-and-past-gdpnow-commentaries and https://www.atlantafed.org/research-and-data/data/gdpnow
• New York Fed nowcast announcement, September 11, 2026: https://www.newyorkfed.org/research/AMEC
• Philadelphia Fed SPF, August 14, 2026: https://www.philadelphiafed.org/surveys-and-data/real-time-data-research/spf-q3-2026
• BLS August employment, September 4, 2026: https://www.bls.gov/news.release/empsit.nr0.htm
• BEA July personal income/outlays, August 26, 2026: https://www.bea.gov/news/2026/personal-income-and-outlays-july-2026
• BEA July trade, September 3, 2026: https://www.bea.gov/news/2026/us-international-trade-goods-and-services-july-2026
• BLS August CPI, September 11, 2026: https://www.bls.gov/news.release/cpi.nr0.htm
• Federal Reserve projections, June 17, 2026: https://www.federalreserve.gov/monetarypolicy/fomcprojtabl20260617.htm
• FOMC statement, July 29, 2026: https://www.federalreserve.gov/newsevents/pressreleases/monetary20260729a.htm
• CBO outlook, February 11, 2026, and update index: https://www.cbo.gov/publication/62105 and https://www.cbo.gov/topics/economy/outlook-budget-and-economy
• Deloitte forecast, July 1, 2026: https://www.deloitte.com/us/en/insights/topics/economy/us-economic-forecast/united-states-outlook-analysis.html
• Polymarket, live observation September 14, 2026: https://polymarket.com/event/us-recession-by-end-of-2026
• Kalshi listing, checked September 14, 2026, without a verified price: https://kalshi.com/markets_by_ticker/KXGDP-26OCT30-T1.0
• BEA annual-update announcement, August 17, 2026: https://www.bea.gov/news/blog/2026-08-17/annual-update-gdp-industry-and-state-stats-publicly-available-starting-sept-30
• BEA delayed Q3 initial release, December 23, 2025: https://www.bea.gov/news/2025/gross-domestic-product-3rd-quarter-2025-initial-estimate-and-corporate-profits
• BEA delayed Q4 advance, February 20, 2026: https://www.bea.gov/news/2026/gdp-advance-estimate-4th-quarter-and-year-2025
• BEA revision comparison, updated November 2025: https://www.bea.gov/sites/default/files/2025-11/relia.pdf
• BEA policies, modified February 5, 2026, checked September 14: https://www.bea.gov/about/policies-and-information

Forecast date: 2026-09-14. Evidence cutoff: 2026-09-11.

Current status against the resolution criteria

The resolving figure is the headline SAAR in each quarter's BEA advance estimate (first published figure if relabeled). The most recent data:

• Q2 2026 advance (2026-07-30): +1.5% vs. consensus 2.1%; consumer spending +3.2% but imports subtracted heavily (Reuters, https://www.reuters.com/world/us/us-economic-growth-slows-second-quarter-domestic-demand-robust-2026-07-30/). Second estimate (2026-08-26) unchanged at 1.5% (https://www.advisorperspectives.com/dshort/updates/2026/08/26/q2-gdp-second-estimate-real-gdp-at-1-5-unchanged).
• Prior advance prints: Q1 2026 2.1%, Q4 2025 1.4%, Q3 2025 4.3% — illustrating how noisy advance headline prints are around a ~2% trend, largely because of trade/inventory swings.

Q3 2026 nowcasts (release scheduled 2026-10-29):

• Atlanta Fed GDPNow 4.4% (Sept 10, 2026), down from 4.7–4.8% early September; components PCE 3.6%, gross private domestic investment ~19%, net exports ≈ −1.5 pp (https://www.atlantafed.org/research-and-data/data/gdpnow/current-and-past-gdpnow-commentaries).
• NY Fed Staff Nowcast 2.26% (Sept 11, 2026) (https://www.marketscreener.com/news/new-york-fed-us-q3-gdp-nowcast-estimate-unrevised-at-2-26-gain-ce785bdfdf80f723).
• Philadelphia Fed SPF Q3 2026 (2026-08-14, 32 forecasters): medians 2026Q3 2.5, 2026Q4 2.3, 2027Q1 2.1, 2027Q2 2.1, 2027Q3 2.2; annual 2026 2.1, 2027 2.2, 2028 2.1, 2029 2.4 (https://www.philadelphiafed.org/surveys-and-data/real-time-data-research/spf-q3-2026).

Policy/calendar context: FOMC meets Sept 15–16; a Reuters poll (2026-09-09) expects a hold for the rest of 2026 but a rising number of analysts see at least one hike, and prediction markets lean toward a hike (https://www.reuters.com/business/fed-hold-rates-steady-rest-2026-rising-number-analysts-see-least-one-hike-2026-09-09/; https://federalnewsnetwork.com/prediction-markets/2026/09/rate-hike-coming-september-prediction-markets-next-fed-decision/). Congress passed a continuing resolution funding the government to Dec 11, 2026 (https://www.theguardian.com/us-news/2026/sep/01/house-bill-prevents-government-shutdown), so the Oct 29 Q3 advance release is not at shutdown risk; a December lapse could delay the Q4 advance (late Jan 2027), but the rule says to use the first published estimate under any name, so this affects timing rather than the number. August payrolls were strong (+162k vs ~53k expected), and a Middle East/Iran energy shock plus higher bond yields and 6.8% mortgage rates are the main near-term headwinds.

Reference class and base rate

The series is the quarterly SAAR advance estimate. Over 2010–2025 (excluding the 2020 pandemic quarters) the advance-estimate headline has a mean of roughly 2.3% and a standard deviation near 1.9–2.0 pp; including recession-type outcomes fattens the left tail. Advance prints deviate from third estimates/annual revisions by ~0.5–0.7 pp RMSE, and the nowcast-to-advance error at this point in the quarter (mid-month 3) is roughly 1.0–1.3 pp RMSE for GDPNow. Potential growth is estimated at ~1.8–2.0% (CBO, slower immigration and aging labor force, offset by capex/AI-driven productivity), so the long-run median should sit near 2.0, slightly below the 2010s average, with a recession probability of roughly 15–20% per year producing a fat left tail.

Building the series

• 2026Q3 (p10 1.4 / p50 3.0 / p90 4.6). The 2.1 pp gap between GDPNow (4.4) and NY Fed (2.3) is unusually wide. GDPNow's 19% investment component and 3.6% PCE look strong relative to Q2's 1.5% print; historically when GDPNow sits far above consensus mid-quarter the advance print lands between them. I weight GDPNow ~40%, NY Fed ~30%, SPF ~30%, giving ~3.0. Band width (~3.2 pp) reflects the nowcast disagreement plus typical nowcast error, and the possibility of another import-driven downside surprise as in Q2.
• 2026Q4 (−0.2 / 1.9 / 3.9). Payback from a Q3 investment/inventory surge, tighter financial conditions (possible Fed hike, higher long yields), energy shock drag; SPF median 2.3 adjusted down for mean reversion after a strong Q3.
• 2027 (p50 ≈ 2.0–2.1, p10 ≈ −0.6 to −0.9, p90 ≈ 4.3–4.6). Anchored on SPF medians (2.1–2.2) shaded toward potential (~2.0); bands widen toward the unconditional distribution as nowcast information vanishes.
• 2028–2031 (p50 2.0, p10 −1.1 → −1.3, p90 4.7 → 4.9). Unconditional distribution around potential growth, with the left tail widened to reflect recession risk accumulating over five years and the tendency of advance prints to overshoot in both directions on trade/inventory swings. Median held at 2.0 (SPF 2028 2.1, 2029 2.4; CBO potential ~1.8; AI-capex upside vs. demographic drag roughly offset).

Strongest consideration against the central estimates

For Q3 2026, the strongest case against a ~3% median is that the NY Fed nowcast (2.3) and SPF (2.5) agree, while GDPNow's mid-quarter investment component has a track record of being revised down sharply once trade and inventory data arrive; a repeat of Q2's import-driven surprise could deliver a sub-2% print. Conversely, GDPNow real final sales still imply >2.5% underlying demand and August payrolls were strong, so a 4%+ print is well within reach. For the long horizons, the strongest argument against a 2.0 median is either a productivity acceleration from AI investment (which would push the median toward 2.5) or a slower-growth regime from restrictive immigration and higher real rates (pushing toward 1.5); I treat these as roughly offsetting and reflect them in the bands.

Evidence in the next 90 days that would move the forecast by more than half the 10–90 band

• GDPNow after the Sept 16 retail sales and Sept 26 durable goods/trade releases moving below ~3.0% or above ~5.0%, or the NY Fed nowcast converging toward 3%+ (Q3 band).
• The Sept 15–16 FOMC decision and dot plot: a hike plus hawkish guidance would lower Q4 2026–2027 medians by ~0.3 pp; a dovish hold would raise them slightly.
• A sharp escalation of the Middle East energy shock (oil sustained above ~$110) or a major credit/equity drawdown would shift the 2026Q4–2027 distribution down by more than 1 pp.
• The Q3 advance print itself (Oct 29) — a print ≥4% raises the odds of Q4 payback (lower Q4 median), a print ≤2% would lower the trend read for 2027.
• A December 2026 shutdown delaying/relabeling the Q4 advance (timing only, per resolution rule).

Current status vs resolution

Resolving figure is BEA headline SAAR in GDP Advance release. Q1 2026 = 2.1%, Q2 2026 advance = 1.5% on Jul 30, unrevised in second estimate Aug 26 (BEA). Next: Q2 third estimate Sep 30, then Q3 2026 advance Oct 29 8:30am (BEA schedule). Q4 advance expected late Jan 2027.

• BEA: "Real GDP increased at an annual rate of 1.5% in Q2 2026" advance and second; "In Q1, real GDP increased 2.1%" (bea.gov 2026-07-30, 2026-08-26).
• BEA schedule confirms Oct 29 GDP Advance Q3 2026 (bea.gov/news/schedule).

Near quarters (2026Q3-Q4)

H1 2026 annualized ~1.8% (avg of 2.1, 1.5), 2024 was 2.8% so slowdown.

Q3 tracking as of early-mid Sep:

• Atlanta GDPNow 4.4% on Sep 10 (down from 4.7% Sep 3, 4.8% Sep 1, 4.0-4.3% mid-Aug; opened ~5-6.2%) (Newsquawk 2026-09-10, AtlantaFed X 2026-09-03).
• NY Fed Staff Nowcast Q3 2.26% (Sep 11, unrevised) with 50% interval [1.1,3.4]%, 80% [0.0,4.5]%, and Q4 2.5% (MarketScreener 2026-09-11, MTS Insights 2026-09-04).
• SPF / Bloomberg economists: 2.5% for Q3 (up from 2.0-2.2%), 2.3% next quarter (Bloomberg 2026-08-21; Instagram SPF 2026-08-27).
• Atlanta is systematically high early: Q1 started ~3% -> ~1.2%, Q2 started >4% -> ~1.6% vs actual 1.5%/2.1% (LinkedIn tracker). Discounts 4.4% heavily.

Compromise median 2.8% for Q3 advance: above NY/SPF consensus to reflect strong incoming consumption data captured by GDPNow, below GDPNow for bias and further drift before Oct 29. 80% band 1.0-4.3 approx matches NY 80% [0,4.5] tightened for 5 weeks of data already in.

Q4: only NY 2.5% and SPF 2.3% available; median 2.1% (slight payback, tighter policy with fed funds 3.50-3.75% and hawkish talk of hikes, PCE 3.7% Jul). Band 0.0-3.9.

Base rate / long-run distribution

• Historical US SAAR advance: mean ~3.18% since 1947 (TradingEconomics), but modern mean ~2-2.5%, SD ~1.3-1.8pp ex-COVID (COVID ±30pp excluded). 2010-19 range roughly -1 to +5. 10th pct ~0.5%, 90th ~3.5% in expansions.
• Institutional trend: EY 2.0% in 2026, 1.9% in 2027 (2026-09-09); Conference Board 1.9% in 2026 and 1.9% in 2027 (2026-08-20); Deloitte 2.0% 2026, 1.8% 2027; CBO 1.8%/yr 2027-28 and 1.8% avg 2027-31 (CBO 2026-01, 2026-02); FOMC Jun SEP median 2.2% 2026, 2.3% 2027, 2.2% 2028, LR 2.0% (reported Sep 2026 preview).
• So unconditional median ~1.9-2.0% with Q1 seasonal shave to 1.8% (residual seasonality: Q1s average ~0.5pp weaker).
• Recession risk (~20-30% in any 2-yr window) creates left tail to -1 to -2% quarterly prints; AI capex boom + fiscal (OBBBA) creates right tail to 4%+. Hence widen p10 to -1.0, p90 to 4.2 by 2030-31. Width ~3.3pp in 2027Q1 growing to ~5.2pp far out = ~1.3x unconditional SD plus parameter uncertainty, calibrated to contain outcome ~80%.

Strongest against

Atlanta 4%+ could persist to advance (consumption + defense spending anecdotes, 2.8% Q3 chatter) making 2.8% too low; conversely restrictive Fed (pricing hikes to 4.1% by Dec, 4.6% by Sep 2027, Reuters 2026-09-09) and tariff drag could pull Q4-2027 below 1.5% and trigger 2028 contraction (Deloitte -1% 2028 scenario).

What would move >10pp / halve interval

• Sep 30 Q2 third + annual update revising H1 materially; Sep CPI/PCE/jobs shifting GDPNow/NY nowcast by >1pp; Sep 15-16 FOMC SEP/dot plot signaling hikes vs holds.

Current status. The resolving series is the BEA advance-estimate SAAR real GDP growth per quarter. Recent history is volatile and tariff-distorted: Q3 2025 +4.4%, Q4 2025 +0.5%, Q1 2026 +2.1%, Q2 2026 +1.5% (advance, unrevised in the second estimate; BEA release Aug 26, 2026). The Q3 2026 advance is scheduled for Oct 29, 2026, after the current CR funds the government through Dec 11, 2026, so no shutdown threatens that release; a Dec-11 funding cliff could, however, delay or relabel the Q4 2026 advance (handled by the resolution rule: use the first published estimate).

Near-horizon anchoring (2026Q3). Atlanta Fed GDPNow (Sep 10, 2026 tracking slides) puts Q3 2026 at 4.4%, with contributions of roughly: PCE +2.4pp, equipment investment +0.95pp (AI capex boom — equipment growing 17% SAAR), net exports −1.44pp (exports falling), inventories +1.99pp ($115bn SAAR build). The nowcast has ranged 4.0–6.2 this quarter. Key uncertainties for the advance print: (a) the inventory contribution is extrapolated in advance estimates and has historically been the largest source of advance-vs-later noise (±1.5pp swings are common); (b) tariff-era trade data has produced large nowcast misses before (e.g., 2025 import front-running/gold-related noise); (c) seven weeks of data remain. GDPNow is roughly unbiased vs the advance, with final-nowcast MAE ~0.6pp and larger errors further out. I center at 3.9 (slight haircut for inventory payback within the quarter and soft export momentum) with p10 1.6 (inventories surprise low / PCE fades) and p90 5.7 (inventory extrapolation overshoots, capex surge continues; GDPNow peaked at 6.2 this quarter).

2026Q4. The Q3 inventory build implies payback risk; the Fed is expected to hike Sept 15–16 (markets >80% priced; UBS expects a second hike in December; Reuters poll economists lean hold — genuine two-sided policy uncertainty), and the Dec 11 funding cliff falls mid-quarter. Against that, the AI capex boom accelerates into 2027 (~$800bn in 2026 → ~$1T in 2027 per Kobeissi/Reuters Breakingviews, ~3% of GDP annually 2027–29), jobs are solid (Aug +162k, U3 4.1%). Median 1.8, p10 −1.4, p90 4.5.

2027–2028: the boom-vs-bust window. The regime rhymes with 1999–2000: a historic tech investment boom, a Fed restarting hikes into above-target inflation (Aug CPI 3.4% y/y; core PCE ≈3.4% — tariff pass-through plus oil near $100 on Middle East tensions), and long rates at multiyear highs on fiscal concerns. Two paths dominate: (i) soft-landing/boom-continuation — capex and consumption keep growth near 2–3% (SPF panel: 2.1–2.4% annual growth 2026–2029; IMF sees the US outperforming); (ii) a 2001-style capex bust and/or Fed overtightening producing recession (Polymarket: ~34% recession by end-2027; NY Fed spread model 13.9% for 2027, likely understated because the curve is steep for fiscal reasons; prominent investors e.g. Cooperman publicly forecast a 2027 recession). I set medians at 1.6–1.8 for 2027 quarters with fat left tails (p10 −1.9 to −2.6), which prices a recession-adjacent outcome below the consensus ~2% without fully embracing it, since the labor market is still firm and the investment pipeline is contracted/not easily cancelled.

2028 and beyond. If a recession arrives, 2027H2–2028H1 is the most likely window, so 2028 medians (1.9–2.0) blend a recovery scenario with an ongoing-contraction scenario; p10s of −2.8 reflect average recession-quarter prints (recession quarters average roughly −2 to −3 SAAR, with 2008-style tails below the 10th percentile). From 2029–2031 I revert toward the unconditional distribution of quarterly advance estimates (post-1990: mean ≈2.3, SD ≈2.2, plus advance-release measurement noise), slightly lifted by possible AI-driven productivity gains: medians 2.1–2.2, p10 −1.8 to −2.2, p90 5.4–5.6. Bands are wide deliberately: the last four quarters alone spanned 0.5–4.4, and the instruction warns that too-narrow bands are the usual failure.

Strongest case against my central path: the boom-continuation view — the AI capex cycle is contracted through 2027–2029, the labor market is rebalancing without breaking, and the Fed is hiking from only ~3.6%, which is not deeply restrictive; on that view 2027 quarters print 2.5–3% and my medians are 80–100bp too low. I partially accommodate this with right tails at 4.5–4.8 for 2027, but keep medians below consensus because the Fed is tightening into 3.4% core inflation with fiscal/energy shocks layered on, the historical base rate for a recession within any 2-year window is ~30–40%, and the investment share of growth is unusually concentrated (half of growth from AI capex), which raises the amplitude of any bust.

What would move me >10pp / materially:

• Sept 15–16 FOMC: a 50bp hike or a hawkish dot plot implying ~4.5%+ terminal rate → shift 2027 medians down ~0.5pp and widen left tails; a surprise hold on soft inflation → raise 2026Q4–2027 medians toward 2.5.
• October GDPNow prints: sustained readings below ~3% (inventory unwinding faster) → cut Q3 p50 toward 3%; above 5.5% → raise toward 4.5.
• Big-tech Q3 earnings capex guidance for 2027 (late Oct): any wavering → raise 2027 recession weight.
• A shutdown extending past mid-January 2027 (Q4 release relabeled/delayed; direct activity hit to Q4/Q1).
• Oil sustained >$110 or a payroll trend flipping to sustained losses → shift 2027 quarters down ~0.5pp.

Sources: BEA Q2 2026 GDP second estimate (bea.gov, Aug 26, 2026); Atlanta Fed GDPNow Q3 2026 tracking slides (Sep 10, 2026, atlantafed.org); Reuters CPI report Sep 11, 2026; WSJ core-PCE estimate; Reuters Fed poll Sep 9, 2026; UBS Fed call (Sep 7, 2026); The Hill CR passage (Sep 1, 2026, funding through Dec 11); NY Fed recession-probability model (Aug 2026, 13.9% for 2027); Polymarket recession odds (Aug 2026); SPF panel 2026–2029 growth (Aug 2026); Kobeissi Letter / Reuters Breakingviews on AI capex (~3% of GDP 2027–29); BLS August jobs report (Sep 4, 2026).

Current status. As of 2026-09-14 the latest published BEA figures are Q1 2026 real GDP +2.1% SAAR and Q2 2026 +1.5% SAAR (advance and second estimate unchanged). Q2 private domestic final purchases were much stronger at +4.2% SAAR; the headline was dragged by inventories (about −0.7 pp) and net exports (about −1.1 pp). Q2 PCE inflation printed 5.3% SAAR (core 3.6%). Calendar: Q2 third estimate plus the 2026 annual NIPA revision is 2026-09-30; the Q3 2026 advance is scheduled 2026-10-29.

Recent history of the resolving series (advance/first prints where noted): 2025Q2 +3.8%, 2025Q3 +4.4%, 2025Q4 +1.4% (advance), 2026Q1 +2.1%, 2026Q2 +1.5%. Full-year 2025 real GDP was about +2.1–2.2%.

Nowcasts and professional forecasts for Q3 2026 (the only quarter with a live nowcast).

• Atlanta Fed GDPNow 4.4% as of 2026-09-10 (next update 09-16). Path this quarter: started 5.0% (7/30), peaked 6.2% (8/3), low 4.0% (8/18), now 4.4%. Final sales 2.4%; PDFP 4.2%; inventories contribute ~+2.0 pp; net exports ~−1.4 pp; PCE growth 3.6%. GDPNow is built to track the advance estimate (full-sample RMSE 1.17 pp, MAE 0.77 pp; late-quarter RMSE closer to ~0.6–0.8 pp).
• NY Fed Staff Nowcast: 2.26–2.3% for Q3 and 2.51% for Q4 (early September).
• Blue Chip consensus: about +2.5% and rising (was ~2.2–2.25% in August).
• Philadelphia Fed SPF Q3 2026 (released 2026-08-14): Q3 2.5%, Q4 2.3%, 2027Q1–Q2 2.1%, 2027Q3 2.2%. Annual-average GDP 2.1% (2026), 2.2% (2027), 2.1% (2028), 2.4% (2029). Mean probability of a negative quarter: 13.3% (Q3), 20.0% (Q4), ~21–22% through 2027Q3.
• FOMC June 2026 SEP (Q4/Q4): 2026 2.2%, 2027 2.3%, 2028 2.2%, longer run 2.0%. Central tendency 2.0–2.3% for 2026. New SEP due 2026-09-16.
• CBO (Feb 2026): +2.2% in 2026, then 1.8% average 2027–2036.
• Conference Board (Sept 2026): ~1.9% for 2026 and 2027; now expects three consecutive funds-rate hikes (Sep/Oct/Dec) and a deeper consumer pullback.

Macro backdrop that shapes 2026Q4 onward. Unemployment 4.1% (August), payrolls strong; funds rate 3.50–3.75% with a hawkish-hold debate at the 15–16 Sep FOMC. CPI +3.4% YoY in August, but energy +16.3% YoY and gasoline +27.4% YoY. Brent ~$105 (Fortune/ICE, 11–13 Sep 2026). This is a live oil shock: the US–Iran / Hormuz conflict is ~six months old (BNEF, Reuters Aug 2026); Brent averaged ~$90 YTD vs ~$70 in 2025 and spiked above $120 in April. That is a stagflationary supply shock of 1990/2022 scale, not fully in the mid-August SPF. Real consumer purchasing power is the transmission channel; Q2 PDFP strength can fade as gasoline and rates bite.

Reference class and base rate (named before case-specific adjustment).

1. Near-term (Q3 2026): late-quarter nowcast error vs the BEA advance. GDPNow’s own RMSE plus the current 2 pp GDPNow–Blue Chip gap. When that gap is inventory-driven, the advance often lands between GDPNow and consensus, because the advance still has only two months of inventory source data.
2. Subsequent quarters: mixture of (a) post-1990 expansion-quarter SAAR (mean ~2.3%, SD ~1.5–1.8 pp) and (b) a ~12–20% chance a given quarter is a recession/contraction print. The unconditional 1990–2019 distribution of quarterly SAAR (revised) has roughly p10 ≈ −1.5%, p50 ≈ 2.5%, p90 ≈ 5%. Advance prints are a bit noisier than finals but less extreme in deep recessions (2008Q4 advance −3.8% vs later −8.4%). Distant 10–90 bands are set to that historical 80% spread, not to SPF point-forecast dispersion.
3. Long-run median: FOMC longer-run 2.0% and CBO 1.8%, with a modest AI/productivity upside by 2030. Recession risk pulls the mean down more than the median.

Q3 2026 (advance 29 Oct). Blend: ~40% GDPNow 4.4, ~40% consensus/SPF/Blue Chip ~2.5, ~20% NY Fed 2.3 → ~3.3. Discount GDPNow ~0.8–1.0 pp for the well-known inventory overshoot when CIPI is the residual. Remaining September source data (retail sales 16 Sep, housing, M3, advance economic indicators 28 Oct) can still move GDPNow 0.5–1.5 pp. p10 = 1.1, p50 = 3.3, p90 = 5.1. The 10–90 is wider than a 0.8 pp RMSE Gaussian (±1.0 pp) because of the 2 pp model disagreement.

Q4 2026. Inventory payback after a ~2 pp Q3 add, oil at $105 hitting real PCE, possible additional tightening. SPF 2.3 / NY Fed 2.5 are the anchors; shade down for the post-SPF oil spike and payback. p10 = −0.5, p50 = 1.9, p90 = 3.8.

2027. Elevated (~20% per quarter SPF) contraction risk, energy shock persistence, and a Fed that may hike rather than cut. Medians 1.7–1.8, below the June SEP 2.3% Q4/Q4. Bands widen toward the historical mix.

2028–2031Q3. Median glides to 2.0, then 2.1 from 2030H2 on mild AI/potential-growth upside. p10 settles near −2.0 (covers typical recession-quarter advance prints; 2008Q4/2020Q2 remain outside p10, as they should). p90 settles near 4.4–4.5 (2025Q3’s 4.4% is a ~p90 expansion quarter). Width ~6.5 pp matches the 1990–2019 80% historical spread of quarterly SAAR, plus a bit of advance-estimate noise.

Strongest case against the central path. (1) GDPNow is designed to track the advance and is already late-quarter at 4.4% — Q3 p50 of 3.3 could be 0.5–1.0 pp too low if inventories really add 2 pp in the first print. (2) PDFP running 4% argues domestic demand is not rolling over, so 2027 medians near 1.7 could be too weak if the oil shock fades. (3) Conversely, a Hormuz-style shock plus three hikes could put 2027 p50 closer to 0–1% and pull p10 to −4. I left Q3 below GDPNow and 2027 below the SEP, but did not put a recession in the median.

Sources (accessed 2026-09-14):

• BEA Q2 2026 second estimate, 2026-08-26: https://www.bea.gov/news/2026/gdp-second-estimate-and-corporate-profits-2nd-quarter-2026
• BEA release schedule (Q3 advance 2026-10-29): https://www.bea.gov/news/schedule
• Atlanta Fed GDPNow slides, 2026-09-10: https://www.atlantafed.org/-/media/Project/Atlanta/FRBA/Documents/cqer/researchcq/gdpnow/RealGDPTrackingSlides.pdf and https://www.atlantafed.org/research-and-data/data/gdpnow
• FOMC SEP, 2026-06-17: https://www.federalreserve.gov/monetarypolicy/fomcprojtabl20260617.htm
• SPF Q2 2026 (15 May) and Q3 2026 (14 Aug): https://www.philadelphiafed.org/surveys-and-data/real-time-data-research/spf-q2-2026 and https://www.philadelphiafed.org/surveys-and-data/real-time-data-research/spf-q3-2026
• BLS CPI August 2026, 2026-09-11: https://www.bls.gov/news.release/PDF/cpi.PDF
• CBO Budget and Economic Outlook 2026–2036, Feb 2026: https://www.cbo.gov/publication/62105
• Conference Board US forecast page / 11 Sep 2026 outlook: https://www.conference-board.org/research/us-forecast
• Oil: Fortune 11 Sep 2026 (~$105.82 Brent); ICE Brent ~$104; Reuters 27 Aug 2026 on six months of war.
• 2025Q4 advance +1.4%, 2025Q3 +4.4%: https://www.bea.gov/news/2026/gdp-advance-estimate-4th-quarter-and-year-2025

Current status (as of 2026-09-14)

Recent prints (BEA advance estimates): Q1 2026: +2.1%; Q2 2026: +1.5% (advance, unrevised at second estimate on Aug 26, 2026 — BEA).

Q3 2026 (advance due Oct 29, 2026):

• Atlanta Fed GDPNow (Sep 10): 4.4% — but this includes a +2.09pp inventory contribution (FRED CHNGNETINVENTCONTRIBNOW) after a -0.45pp drag in Q2; ex-inventories it implies ~2.3% final sales growth. Real PCE nowcast +3.5%, real investment +21.8% (AI capex boom) per the GDPNow commentary.
• NY Fed Nowcast: 2.26% for Q3, 2.51% for Q4 (LiveSquawk).
• Philadelphia Fed SPF (Aug 14, 2026): 2.5% for Q3, 2.3% for Q4 (staffingtoday summary); annual-average 2.1–2.4% for 2026–2029.
• Private consensus ~2.5% for Q3.

Policy/macro backdrop: FOMC meets Sep 16; KPMG and BBH expect the first rate hike since July 2023 (to 3.75–4.00%) with CPI at 3.4% y/y (KPMG, BBH, BLS CPI). June 2026 SEP medians (Q4/Q4): real GDP 2.2/2.3/2.2 for 2026/27/28, longer-run 2.0 (Fed SEP); the Sep SEP may shave 2027–28 by 0.1pp per TheStreet. Long-end yields are near 5% (10y) / 5.37% (30y) amid ~6%-of-GDP deficits (Wolf Street). Government is funded via CR only through Dec 11, 2026 (EducationCounsel), creating shutdown risk around the Q4 2026 measurement window; the resolution rule covers renamed/delayed first estimates. Market-implied recession-or-major-downturn probability by end-2027 is ~25–41% (summary).

Reference class and calibration

Base rate: quarterly BEA advance-estimate SAAR real GDP growth. 2010–2019: mean ≈ 2.3%, std ≈ 1.4pp, 10th–90th ≈ [0.5, 4.2]. Unconditional 2000–2019 distribution (including recessions): mean ≈ 2.0%, ~12% negative prints, 10th percentile ≈ -0.3. Recession quarters print roughly -1 to -5 in advance estimates (2008Q4: -3.8; 2009Q1: -3.8 advance). I forecast each period as a mixture: a trend state (~2.0–2.2%, std ~1.3) plus a recession-state probability rising from ~5–8% for 2027 quarters to ~15–18% per quarter by 2030–31, plus a boom-state upside from AI-driven productivity (Oxford Economics: +3.5% productivity over a decade; AI investment ≈ half of current GDP growth).

Near-term anchoring: For 2026Q3, the advance releases in 6 weeks. GDPNow's headline (4.4%) and the survey consensus (2.5%) bracket the truth; the gap is almost entirely the volatile inventory component, where nowcast errors are largest. I center at 3.2 (slightly above the SPF, crediting strong PCE/investment and partial survival of the inventory build) with an 80% band of [1.2, 5.2], consistent with GDPNow's historical ~1.3–1.5pp RMSE at this horizon. For 2026Q4 I center on SPF/NY Fed at 2.3, with a lower p10 (0.0) reflecting shutdown risk after Dec 11 and tightening financial conditions, and p90 4.5 reflecting momentum carryover.

Medium/long term: medians glide from 2.2 (2027) to 2.1 (2028) to 2.0 (2029–2031), tracking the SEP central tendency, SPF long-run, and CBO potential, with a small AI-productivity offset to slowing labor-force growth. Bands widen steadily: p10 from -0.3 (2027Q1) to -2.3 (2031), p90 from 4.7 to 6.0. A 4–5-year-ahead 80% band of roughly [-2.3, 6.0] is deliberately wide — the usual failure mode is bands that are too narrow; any single quarter in 2030–31 has a meaningful chance of landing in a recession print (< -1) or a boom print (> 5).

Strongest consideration against my central estimate

The GDPNow 4.4% for Q3 could be right — the inventory swing from -0.45pp to +2.09pp is exactly what post-payback quarters look like, PCE at 3.5% is genuinely strong, and the AI investment boom has repeatedly beaten consensus in 2025–26. If so, my Q3 p50 (3.2) is ~1pp low and my Q4+ medians may also be low if productivity-led trend growth has stepped up toward 2.5%. Against that: mid-quarter GDPNow overshoots are common, the NY Fed and SPF are independent corroboration of ~2.3–2.5%, and the Fed is hiking into 3.4% inflation with 5%+ long yields — a combination that historically precedes slower, not faster, growth 4–6 quarters out.

What would move the forecast >10pp-equivalents (≈ half the 10–90 interval)

• Sep 16 FOMC: a hike plus SEP downgrades (or a surprise hold with dovish dots) — shifts 2026Q4–2027 medians ±0.3–0.5.
• September trade/inventory data (Oct 6–8) and the final GDPNow (Oct 28): resolution of the inventory question moves the Q3 median by ±0.5–1.0.
• A December shutdown: Q4 2026 p50 down ~0.3–0.5, Q1 2027 up symmetrically (payback).
• AI capex retrenchment or credit stress in AI-related issuance: fattens the 2027+ left tail materially; conversely, sustained 3%+ prints would shift 2027–28 medians up 0.3–0.5.
• 10-year yield sustainably >5.5%: raises recession odds embedded in the 2028+ p10s.

Forecast Analysis: US Real GDP Growth (BEA Advance Estimate)

1. Resolution Criteria and Metric Definition

The question asks for the annualized real GDP growth rate (percent) reported in the Bureau of Economic Analysis (BEA) "GDP (Advance Estimate)" news release for each quarter from 2026Q3 through 2031Q3 (21 quarters).

• Resolution turns strictly on the headline seasonally adjusted annualized rate (SAAR) from the initial advance estimate release.
• Second and third estimates and all subsequent comprehensive/benchmark revisions are ignored.
• The release for 2026Q3 is scheduled for October 29, 2026. The 2026Q4 release is expected in late January 2027. Subsequent quarters are released approximately four weeks after the end of the respective quarter.

2. Current Status and Real-Time Data (as of September 14, 2026)

• Recent Advance Estimates (BEA):
  • 2025Q4 Advance: +1.4%
  • 2026Q1 Advance: +2.0%
  • 2026Q2 Advance: +1.5% (with real final sales to private domestic purchasers expanding by 3.9%, offset by inventories and federal nondefense adjustments).
• Q3 2026 High-Frequency Tracking:
  • Atlanta Fed GDPNow: As of September 10, 2026, the GDPNow model estimates Q3 2026 real GDP growth at 4.4% SAAR. Component tracking shows real personal consumption expenditures (PCE) growing at 3.6% and real gross private domestic investment growing at 19.1%, partially offset by a net export drag of -1.46 percentage points.
  • Survey of Professional Forecasters (Philadelphia Fed, August 14, 2026): Panelists projected a median real GDP growth of 2.5% for 2026Q3 and 2.3% for 2026Q4, with the risk of a negative quarter at 13.3% in Q3 and 20.0% in Q4.
  • Labor Market and Inflation Data (BLS): The August 2026 employment report (released September 4, 2026) showed nonfarm payroll additions of +162,000 (well above the +53,000 consensus) with unemployment holding at 4.1%. The August CPI report (released September 11, 2026) showed headline CPI up 0.4% m/m (3.4% y/y) and core CPI up 0.3% m/m, indicating robust economic demand.

3. Base Rates and Reference Class

• Reference Class: The historical time series of BEA advance estimates of quarterly annualized real GDP growth.
• Historical Distribution (Post-1985 / Great Moderation):
  • Mean / Median: ~2.2% to 2.5% annualized.
  • Standard Deviation of quarterly annualized growth: ~2.0% to 2.2% percentage points (reflecting high quarterly noise from inventories, net trade, defense outlays, and initial data imputation).
  • Unconditional 10th percentile: ~ -1.0% to -1.2%.
  • Unconditional 90th percentile: ~ +5.0% to +5.2%.
  • Frequency of negative quarters: Historically ~15%–20% of all quarters (consistent with the SPF’s assessment of ~20%–22% recession/contraction risk in distant quarters).
• Revision Properties (Fixler et al., BEA, 2024): Advance estimates have an average absolute revision of ~1.2 percentage points relative to latest benchmarks, but the advance estimate itself is an unbiased first read with roughly the same unconditional variance as revised figures.

4. Synthesis and Quarter-by-Quarter Forecast Path

Near Term (2026Q3 and 2026Q4)

• 2026Q3: With two months of strong consumer spending and labor data in hand, and GDPNow tracking 4.4%, Q3 is positioned to be significantly stronger than Q2 (1.5%). However, GDPNow historically overshoots late-quarter prints when tracking above 4%, as September source data on inventory accumulation and international trade balances often moderate initial estimates. Weighting the GDPNow tracking (~4.4%) against institutional and SPF consensus (~2.5%–3.0%) yields a median expectation of 3.0%. The 10th percentile is set at 1.3% and 90th percentile at 4.6% (spread: 3.3 pp).
• 2026Q4: The quarter has not yet begun. Restrictive policy rates (Fed funds rate ~3.8%) and normalizing consumption suggest growth will moderate back toward potential. SPF median is 2.3%; we set the median at 2.2%, with an 80% band from 0.2% to 3.9% (spread: 3.7 pp).

Medium Term (2027Q1 through 2028Q4)

• Forecasters and policymakers project US potential output growth at 1.8%–2.3% (June 2026 FOMC SEP medians: 2.3% in 2027, 2.2% in 2028; CBO baseline: 1.8%–2.2%).
• As the horizon extends beyond 6 months, specific quarter-to-quarter visibility vanishes, and the forecast converges toward potential growth (median ~2.0%–2.1%) while the uncertainty bands widen to account for business cycle fluctuations and quarterly noise.
• Across 2027, the 10th percentile dips into negative territory (-0.2% to -0.5%), consistent with the ~20% historical probability of experiencing a quarterly contraction. By 2028, the 80% interval expands to [-0.7%, 4.6%].

Long Term (2029Q1 through 2031Q3)

• Anchoring to the long-run potential growth rate of 2.0% (with a minor seasonal pattern assigning 2.1% to Q3).
• While structural factors such as demographic aging exert a drag on labor hours, enterprise adoption of artificial intelligence and automation provides an upside tail to total factor productivity.
• The 10–90 percentile interval expands smoothly to reflect full unconditional quarterly volatility, reaching [-1.2%, 5.2%] by 2031. Historically, ~10% of quarters fall below -1.2% (recessionary quarters) and ~10% exceed +5.2% (cyclical rebounds / tech-driven booms).

5. Counter-Arguments and Key Risks

• Strongest case for higher numbers: Accelerated diffusion of generative AI, robotics, and workflow automation into mainstream enterprise operations could boost labor productivity growth by 0.5%–1.5% annually, permanently lifting quarterly median GDP growth to 2.8%–3.5%.
• Strongest case for lower numbers: Persistent fiscal imbalances, rising sovereign debt service costs, potential trade protectionism/tariffs, or an overtightening by monetary authorities could induce an extended period of stagnation or recession, pushing median growth below 1.5% with multiple contractionary quarters.

6. What Would Move the Forecast in the Next 90 Days

• A dramatic divergence in the final September 2026 economic releases (e.g. an unexpected plunge or spike in retail sales, industrial production, or August trade deficit figures) that swings the Q3 advance estimate outside the 1.5%–4.5% range.
• The actual Q3 2026 BEA Advance Estimate publication on October 29, 2026.
• A major unexpected monetary policy shift at the September or November FOMC meetings.

---

Sources Cited

1. U.S. Bureau of Economic Analysis (BEA), GDP (Advance Estimate), 2nd Quarter 2026, published July 30, 2026: https://www.bea.gov/news/2026/gdp-advance-estimate-2nd-quarter-2026
2. Federal Reserve Bank of Atlanta, Current and Past GDPNow Commentaries (Q3 2026), updated September 10, 2026: https://www.atlantafed.org/research-and-data/data/gdpnow/current-and-past-gdpnow-commentaries
3. Federal Reserve Bank of Philadelphia, Third Quarter 2026 Survey of Professional Forecasters, published August 14, 2026: https://www.philadelphiafed.org/surveys-and-data/real-time-data-research/spf-q3-2026
4. Federal Open Market Committee (FOMC), Summary of Economic Projections, published June 17, 2026: https://www.federalreserve.gov/monetarypolicy/files/fomcprojtabl20260617.pdf
5. Dennis J. Fixler, Eva de Francisco, and Ian Schaaf, Revisions to Gross Domestic Product, Gross Domestic Income, and Their Major Components, Survey of Current Business / BEA, October 2024: https://www.bea.gov/sites/default/files/2024-10/relia.pdf
6. U.S. Bureau of Labor Statistics, The Employment Situation - August 2026, published September 4, 2026: https://www.bls.gov/news.Release/pdf/empsit.Pdf
7. Congressional Budget Office (CBO), The Budget and Economic Outlook: 2026 to 2036, published February 11, 2026: https://www.cbo.gov/publication/62105

1. What the resolving series actually is

The question resolves on the headline SAAR real GDP percent change in the BEA "GDP (Advance Estimate)" release, not on the eventual (revised) figure. That distinction is not academic right now:

• 2025Q4: advance (published 2026‑02‑20, rescheduled from 2026‑01‑29 because of the Oct–Nov 2025 shutdown) = 1.4%; the value now in FRED after revisions = 0.5%. A 0.9pp gap.
• 2026Q1: second estimate 1.6% → third estimate 2.1% (0.5pp revision).
• 2026Q2: advance 1.5% → second estimate 1.5% (unchanged).

Sources: BEA, GDP (Advance Estimate), 4th Quarter and Year 2025, 2026‑02‑20 (https://www.bea.gov/news/2026/gdp-advance-estimate-4th-quarter-and-year-2025); BEA, GDP (Second Estimate) and Corporate Profits, 2nd Quarter 2026, 2026‑08‑26 (https://www.bea.gov/news/2026/gdp-second-estimate-and-corporate-profits-2nd-quarter-2026); FRED A191RL1Q225SBEA.

Release calendar (BEA schedule page, retrieved 2026‑09‑14): Q3 2026 advance = 2026‑10‑29; second 2026‑11‑25; third 2026‑12‑23. 2027 dates are not yet published.

2. The macro state of play as of 2026‑09‑14

I did not trust memory here; the environment has changed substantially. Verified facts:

Growth so far in 2026. Q1 +2.1%, Q2 +1.5% (advance = second estimate). Q2 real final sales to private domestic purchasers +3.9% advance / +4.2% second estimate — i.e. underlying domestic demand is strong; the weak headline came from a −0.72pp inventory drag, a −1.14pp net-export drag, and a very hot deflator (GDP price index +6.4%, gross domestic purchases +5.8%, PCE +5.3%, core PCE +3.6% SAAR). Real GDI was +2.2% vs GDP +1.5%.

Q3 2026 nowcasts (the key inputs for the 2026‑10‑29 release).

Source	Date	Q3 2026 SAAR
Atlanta Fed GDPNow	2026‑09‑10	4.4% (peaked 6.2% on 8/3, initial 5.0% on 7/30)
NY Fed nowcast	2026‑09‑11	2.26%
Blue Chip consensus	early Sept	~2.5%
Philly Fed SPF median (forecasts received by 8/11)	2026‑08‑14	2.5%
S&P Global US Composite PMI 56.0 (52‑month high) implies	Aug	~3.0%

GDPNow's component detail (Atlanta Fed tracking slides, 2026‑09‑10) shows the 4.4% is ~2.0pp inventory-driven (change in private inventories +$115bn SAAR) on top of PCE +3.6%, equipment +17.4%, IPP +6.4%, nonres structures +2.2%, residential −4.5%, government +1.3%, net exports −1.44pp (exports −2.9%, imports +7.7%). Sources: https://www.atlantafed.org/research-and-data/data/gdpnow and https://www.atlantafed.org/-/media/Project/Atlanta/FRBA/Documents/cqer/researchcq/gdpnow/RealGDPTrackingSlides.pdf.

Energy shock. The 2026 Iran war (initiated by the US and Israel on 2026‑02‑28 per Britannica) has escalated sharply in the last two weeks: US strikes on Iranian tankers, Iranian strikes on US Navy ships and on jets parked in Jordan, Houthis seizing Perim Island and closing Bab el‑Mandeb, and — on 2026‑09‑12/13 — Saudi Arabia shutting its East–West pipeline (the main Hormuz bypass, ~4% of global supply) after a drone attack from Iraq, and Riyadh pulling out of the Oman de‑escalation talks. Brent hit $109 (FT, 2026‑09‑13); oil is +16% since end‑August (TD Economics, 2026‑09‑11). US average gasoline $4.31/gal. Sources: https://www.ft.com/content/4845a503-3ddd-4bb7-a469-c51f275bc873; https://www.cnbc.com/2026-09-13/oil-price-iran-war-strait-hormuz-saudi-pipeline.html; https://economics.td.com/us-weekly-bottom-line.

Inflation. August CPI +0.4% m/m, 3.4% y/y; core +0.3% m/m (2.4% y/y); PPI +0.4% m/m, 5.4% y/y (highest of 2026). Gasoline +3.9% m/m, +16.3% y/y. July PCE 3.7% headline / 3.3% core, but trimmed‑mean PCE only 2.3% — the gap says the shock is energy-concentrated, not broad-based. UMich 1‑year inflation expectations jumped to 4.6% from 4.0%. Sources: https://www.cnbc.com/2026-09-11/cpi-inflation-report-august-2026.html; https://www.houseofchimera.com/blog/energy-led-inflation-global-macro-september-2026.

Policy. The Warsh Fed meets 2026‑09‑15/16 and is ~90% priced to hike 25bp to 3.75–4.00% — the first hike since July 2023 — with Reuters' poll (2026‑09‑14) showing economists expect at least one more. Fed funds futures price ~4.1%. The 10‑year is at 4.97%, 30‑year yields at post‑2007 highs, mortgages ~7%. The July FOMC had three dissents for an immediate hike — the biggest split since 2016. Sources: https://www.reuters.com/business/fed-rate-hike-wednesday-now-likely-say-economists-least-one-more-follow-2026-09-14/; https://www.bbh.com/us/en/insights/blog/mind-on-the-markets/Drivers-for-the-Week-of-September-14-2026.html.

Trade policy. SCOTUS held 6–3 on 2026‑02‑20 that IEEPA does not authorize tariffs, voiding the reciprocal/fentanyl tariffs. The administration pivoted: a temporary Section 122 15% global tariff from 2026‑02‑24, Section 338 tariffs on Canada from 2026‑08‑22, plus new 10–12.5% forced‑labor tariffs. The US effective tariff rate is ~10.6% (Economist tracker). Tariff swings produce large import/net-export swings, which are a first-order source of quarterly GDP noise. Sources: https://taxpolicycenter.org/features/tracking-trump-tariffs (2026‑08‑31); https://www.politifact.com/article/2026/aug/25/trump-trade-tariffs-supreme-court/; https://www.economist.com/interactive/trump-tariff-costs-tracker.

AI capex is load-bearing. The St. Louis Fed (2026‑01‑12) measured AI-adjacent categories (IPE + software + R&D + data centers) contributing 1.30pp in 2025Q1 and 1.16pp in 2025Q2, i.e. above the 0.97pp average of the 2000 dot-com peak. Analysis of the Q1‑2026 third estimate puts the AI contribution at ~1.55pp of the 2.1% headline (~74%), with consumer spending contributing less than a fifth and residential investment contracting for several straight quarters. Epoch AI puts AI capital formation at ~1.5% of GDP, matching or exceeding the late‑1990s telecom peak. Sources: https://www.stlouisfed.org/on-the-economy/2026/jan/tracking-ai-contribution-gdp-growth; https://www.tftc.io/ai-capex-gdp-growth-q1-2026-bea-third-estimate (2026‑06‑25). There are now visible cracks: AI stocks sold off on 2026‑09‑14 after industry leaders "called for a slowdown," and Michael Burry is comparing the cycle to the dot-com buildout.

Labor market. August payrolls +162k (vs 53k expected), unemployment steady at 4.1%, labor force +683k, June/July revised +55k. But >60% of the August gain was food services and local-government education; real average hourly earnings fell 0.1% for the second straight month; quits 1.9%; July retail sales −0.6% (steepest since May 2025); UMich sentiment 47.8, second-lowest ever. Sources: https://www.reuters.com/business/us-nonfarm-payrolls-surge-august-unemployment-rate-steady-41-2026-09-04/; https://www.houseofchimera.com/blog/energy-led-inflation-global-macro-september-2026.

Fiscal/political. Deficit ~$2T FY2026 (5.8–6% of GDP), debt >$40T and >100% of GDP. A CR funds the government only through 2026‑12‑11, six weeks after the 2026‑11‑03 midterms; Trump's approval is in the low-to-mid 30s and he has promised a $5,000 per-adult "dividend" if Republicans hold both chambers (met with immediate bipartisan pushback; Tax Foundation says tariffs cannot fund it). Sources: https://www.reuters.com/legal/government/us-house-has-enough-votes-pass-stopgap-funds-prevent-shutdowns-oct-1-2026-09-01/; https://apnews.com/article/trump-dividend-5k-5000-check-republicans-cc80644e3168acd31c892129fb849436; https://taxfoundation.org/blog/trump-dividend-5000-check-tariff-revenue/.

3. Reference class and base rates

(a) Long-run distribution of the quarterly SAAR series. From FRED A191RL1Q225SBEA, 2000Q1–2026Q2 excluding the 2020 pandemic quarters (n=106): mean 2.20%, median ≈2.45%, SD ≈2.1–2.4%, p10 ≈ −0.65%, p90 ≈ +4.7%. Over 1990–2026 the p10/p90 are ≈ −0.6/+5.6. This is my anchor for distant-horizon bands.

(b) Nowcast/consensus error. Atlanta Fed's own statistics: final GDPNow RMSE 1.17pp, MAE 0.77pp. Mid-quarter (~6–7 weeks before the advance release, where we are now) the error is materially larger — roughly RMSE 1.8–2.0pp. SPF one-quarter-ahead consensus RMSE is ~1.8–2.0pp; four-quarters-ahead ~2.3pp. I widen these slightly because the question resolves on the advance print, which itself carries ~0.4pp of first-release noise (the 2025Q4 advance-to-final gap was 0.9pp).

(c) Official multi-year paths. CBO February 2026 baseline: 2.2% in 2026, then 1.8%/yr (2027 onward); CBO's January 2026 "Current View of the Economy" gives 1.8% average for 2027–2028. Fed long-run ≈2.0%. IMF July 2026 WEO: US 2.3% (2026), 2.2% (2027). Philly Fed SPF (2026‑08‑14): quarterly Q3 2.5, Q4 2.3, 2027Q1 2.1, 2027Q2 2.1, 2027Q3 2.2, annual 2026 2.1, 2027 2.2, 2028 2.1, 2029 2.4; risk of a negative quarter 13.3% / 20.0% / 21.5% / 22.0% / 20.8%. OMB is much higher (2.7% FY26, 3.2% FY27, 2.9% long run) and CEA higher still (~4% through 2028) — I treat both as political rather than forecast-grade. Sources: https://www.cbo.gov/publication/61882; https://www.cbo.gov/system/files/2026-01/61831-Economy.pdf; https://www.philadelphiafed.org/-/media/frbp/assets/surveys-and-data/survey-of-professional-forecasters/2026/spfq326.pdf; https://www.crfb.org/blogs/trump-cea-projections-tracker (updated 2026‑08‑25).

4. Path construction

2026Q3 (median 2.9). Two-thirds of the quarter is already in the data. I average the three live signals — GDPNow 4.4 (weight ~1/3, discounted because ~2.0pp of it is the least-reliable inventory component), NY Fed 2.26 (~1/3), Blue Chip/SPF consensus 2.5 (~1/3) — giving ~3.05, then shade −0.15 for the September energy spike that is not yet in the price data GDPNow uses: oil is +16% since end-August, gasoline $4.31, and BEA will deflate September nominal spending with a much hotter September CPI/PPI (published mid-October, feeding the 10/28 Advance Economic Indicators). A ~20–25% quarterly jump in energy prices adds roughly 1pp annualized to the Q3 PCE deflator at unchanged nominal spending, which is a direct ~1pp hit to real PCE. Result: 2.9, with a tightish band (0.9 / 4.9) reflecting 6.5 weeks of realized data. The band is asymmetric-wide on the upside because GDPNow's inventory signal, if validated, produces ~4.4%.

2026Q4 (median 1.3). This is my largest departure from consensus (SPF said 2.3 on 8/11). Four adjustments: (i) the oil shock now hits a full quarter rather than one month — historically a 25–30% oil price increase subtracts 0.5–1.0pp from real growth via the deflator and consumption; (ii) the Fed is hiking (Sept 16 plus at least one more per Reuters), with a 5% 10-year and 7% mortgages; (iii) inventory mean reversion — if Q3 carries a +2.0pp inventory contribution as GDPNow tracks, Q4 mechanically gets a drag of similar order (this is exactly what happened from 2025Q3 +4.4% to 2025Q4 advance 1.4%); (iv) a ~25–30% chance of a December shutdown after the CR expires on 12/11, which cost ~1.0pp in the 2025 analogue. Offsets (AI capex, resilient payrolls, big deficits, possible war-ending oil collapse) keep the median positive and the p90 near 3.9.

2027 (medians 1.5 → 1.9 across the year). Monetary policy bites with 3–5 quarter lags, so the late‑2026 hikes land hardest in mid‑2027. The base rate for "Fed hikes ≥75–100bp into a supply-side oil shock" is a meaningful recession probability (1974, 1980, 1990, 2000, 2008), though AI capex is an unusual offset that did not exist in those episodes. SPF's ~21–22% per-quarter probability of a negative quarter (as of 8/11, pre-escalation) is, in my view, now closer to 25–28%, which is what my p10 values of −1.3/−1.4 encode. I add a modest upward slope through the year on the assumption that (a) oil normalizes at some point — Trump himself says the war ends "immediately" after the midterms, and an oil collapse produces a large positive real-GDP print, which is my main p90 pathway; (b) the Fed pivots once energy-driven inflation fades; (c) post-election fiscal action. I also apply a −0.2pp Q1 seasonal shading: Q1 advance prints have averaged 1.49% over 2010–2026 vs 2.74–3.03% for Q2–Q4 (2.13% excluding the 2020 and 2014 one-offs), evidence of residual Q1 softness in the published series.

2028–2031 (medians 1.8–2.2, drifting gently up). Anchor: CBO 1.8% from 2028, Fed 2.0% steady state, SPF 2.1% (2028) / 2.4% (2029). I blend to ~2.0% and let it drift to ~2.2% by 2031Q3, reflecting (i) AI capital deepening at ~1.5% of GDP eventually showing up as multifactor productivity (Brynjolfsson-style J-curve, lag of several years), against (ii) slower labor-force growth from immigration restriction and (iii) a rising fiscal drag from >100% debt/GDP and ~6% deficits. I deliberately do not extrapolate either the AI-boom optimism (CEA/OMB 2.9–4%) or the AI-bust pessimism into the median; both live in the tails. Bands widen with distance to ±3.3/±2.8 by 2031, i.e. roughly the unconditional 2000–2026 p10/p90 spread around my (lower) median, plus ~0.6pp on each side for uncertainty about the trend itself.

5. Strongest case against my central estimate

The bear case for my Q4-2026/Q1-2027 downgrade: the hard data are simply not cooperating with a slowdown. S&P Global US Composite PMI hit 56.0 in August, a 52‑month high; payrolls beat by 3×; GDPNow has tracked 4–6% for Q3 since late July; AI capex is compounding at double-digit rates and hyperscaler spend is projected at ~$800bn–$1.4trn; the fiscal deficit is ~6% of GDP; and Goldman cut its 12‑month recession probability to 20% from 30% as recently as 2026‑08‑17. Polymarket prices US recession by end‑2026 at only 7%. If the Iran–GCC–Oman track produces a Hormuz deal in the next few weeks (Iran and Oman have already agreed a temporary corridor and shipping map), oil could fall 25–30%, the Fed's hiking rationale evaporates, and Q4‑2026/Q1‑2027 could print 3–4% instead of my 1.3–1.5%. That pathway is why my Q4 p90 is 3.9 and 2027Q1 p90 is 4.2 rather than something tighter. I have kept the medians low anyway because the asymmetry of the mechanical deflator channel favors me: an oil shock that persists suppresses measured real GDP whether or not nominal demand holds up, and inventory mean reversion after a +2pp Q3 build is close to arithmetic.

The bull case for my distant medians: if AI multifactor productivity gains arrive as the hyperscalers claim, 2029–2031 trend growth could be 2.5–3%, making my 2.0–2.2% medians systematically ~0.5pp low across eight quarters. I have partially hedged this with p90 values of 4.7–5.0 in those periods, but I have not moved the median because no such productivity acceleration is yet visible in the BLS data, and because the same AI cycle carries an offsetting 2001-style bust risk that would push several of those quarters negative.

6. Calibration check

My 10–90 widths run from 4.0pp (2026Q3, where 6.5 weeks are already observed) to 6.4pp (2031Q3, ~5 years out). Against the historical series, a 6.4pp width is slightly wider than the unconditional 2000–2026 p10–p90 spread of 5.35pp — appropriate, because that spread reflects realized history around a known trend, whereas I am also uncertain about the trend. Across the 21 periods my medians average ~1.95%, versus a 1990–2026 realized mean of ~2.5% and a 2000–2026 mean of 2.20%; the discount reflects the current combination of a live oil shock, a hiking Fed, a 5% 10-year, a record-low consumer sentiment index, falling real wages, and CBO/Fed trend estimates (1.8–2.0%) that sit below the historical realized average.

Reading adopted

I forecast the first-published ("advance") BEA estimate of quarterly real GDP growth, SAAR, in percent for each of the 21 quarter-ends 2026Q3–2031Q3. Where the advance release is delayed, skipped or relabelled (e.g. after a government shutdown), I assume the first published figure for that quarter under any name is used, which is what the resolution rules specify.

Current status (as of 2026-09-14)

What is already public. The last advance releases are 2026Q1 (≈2.1% SAAR per the later BEA tables; the advance was in the low-2s) and 2026Q2 = 1.5% SAAR, released 2026-07-30 and unchanged in the second estimate on 2026-08-26 (Fox Business, 2026-07-30; Advisor Perspectives, 2026-08-26; NerdWallet). Reuters characterised Q2 as "growth slows… but domestic demand robust" with consumer spending up 3.2% and imports the drag.

Nowcasts for 2026Q3 (advance release due 2026-10-29):

• Atlanta Fed GDPNow: 4.4% (updated 2026-09-10), down from an initial ~5.0% on 7/30, a 6.2% peak, 4.8% on 9/1 and 4.7% the following week (Atlanta Fed GDPNow; TradingView/MNI).
• New York Fed Nowcast: 2.26% (2026-09-04) (Moomoo/MT Newswires).
• Bloomberg economist survey (Aug 14–19): 2.5%, up from 2.0%, with projections "through the end of 2027… confined to a narrow 2 to 2.2 per cent range" (Business Times/Bloomberg, ~2026-08-21).

So the three headline trackers bracket 2.3–4.4%, a wider-than-usual spread. GDPNow is the outlier, and its recent track record is mixed (it badly overstated 2025Q1, which printed −0.3%).

Macro backdrop. The US–Israel–Iran war began 2026-02-28; Brent/WTI has pushed above $100–105/bbl in September, adding ~$100bn of household energy costs (CNN, 2026-09-09; Axios, 2026-09-07). Inflation is sticky (August CPI driven by gasoline; core PCE ~3.2% for 2026 per the Bloomberg survey) and the new Fed chair Kevin Warsh is explicitly hawkish — markets and most banks expect the first rate hike in three years at the 2026-09-15/16 FOMC, though a Reuters poll of economists still has the Fed on hold through year-end (Politico, 2026-09-11; Reuters poll, 2026-09-09; CBS, 2026-09-12). Labour data are solid: +162k payrolls in August, unemployment 4.1% (NYT, 2026-09-04).

Two structural forces are pulling in opposite directions: AI capex (record capital-goods imports in July — computer imports +25% m/m, semis +10%; total AI capex may exceed $1trn in 2026) is supporting growth, while trade is a persistent drag — the July goods-and-services deficit ballooned 24% to $88.6bn after the Supreme Court struck down the IEEPA tariffs (refunds ≈$100bn) and Trump imposed replacement tariffs on 2026-07-24 (Yahoo Finance, 2026-09-03; Reuters, 2026-08-05). Oxford Economics explicitly warned of a "sizeable drag on Q3 GDP" from net trade.

Reference class and base rate

Reference class: advance-estimate quarterly real GDP SAAR prints, 1985–2025. Core PCE-era mean ≈2.5%, median ≈2.3%; standard deviation ≈3.5% including 2020 and ≈1.9% excluding it. Roughly 10–15% of quarters are negative; quarters above +5.5% SAAR are ~7–8% of the sample. A symmetric 80% band for a "generic" quarter is therefore roughly −2% to +6%, centred on ~2.0–2.3%. That is the anchor for the distant horizons.

For the near horizons I use the reference class of one-to-two-quarter-ahead advance-estimate errors: final GDPNow nowcasts have ~1.0–1.1pp MAE against the advance figure, NY Fed Nowcast ~1.2–1.4pp, and the consensus ~1.0pp; these errors are not independent, so pooling three trackers at 2.3/2.5/4.4 gives a central estimate near 2.9 and an 80% band of roughly 1.2–4.7.

Forecast logic by horizon

• 2026Q3 (p50 = 2.9): weight the trackers. GDPNow is 1.9pp above consensus — a large but not unheard-of gap; I shade the consensus up modestly to reflect the strong August payroll report, the 3.2% Q2 consumption print, AI capex momentum, and GDPNow's informative pull, but not all the way to GDPNow's 4.4. Band 1.2–4.7 covers both the consensus-side and GDPNow-side outcomes.
• 2026Q4 (p50 = 2.3): no nowcast yet; the drags (oil above $100, a Fed hike into an energy shock, tariff/trade drag) and the supports (AI capex, solid labour market, tax-cut stimulus) roughly cancel. Band 0.2–4.5.
• 2027 (p50 = 2.0–2.1): Bloomberg's panel has quarterly growth "confined to a narrow 2–2.2% range" through end-2027; I mark my median to ~2.0–2.1 with a band widening to about −1.0 to +4.8, reflecting rising recession chatter (Cooperman's 2027 recession call) and the stagflationary oil/tariff impulse on one side, AI-productivity upside on the other.
• 2028–2031 (p50 = 2.0): converge to the unconditional distribution of the series, mildly above CBO/Fed longer-run potential (~1.8%) to allow for the AI-investment boom persisting, with bands widening monotonically to −2.0 to +5.7. Over five years at least one recession is more likely than not, which is what sets the lower tail; the upper tail allows for an AI-driven productivity/investment boom.

Strongest case against my central estimate

The strongest counter-argument is that I am too low near-term: GDPNow at 4.4% has already absorbed the July trade report that others may be under-weighting, the labour market re-accelerated (162k, with +55k revisions), and there is no sign of the demand collapse that recession-probability pricing implies. In the 2025 analogue GDPNow's high readings were eventually vindicated for Q2 (+3.0 advance vs ~2.6 nowcast) — a Q3 print of 4%+ is entirely plausible, which is why my p90 is 4.7 rather than ~4.0. Conversely, I could be too high in the 2027–2028 window: an energy shock plus a hiking Fed plus an AI-capex payback are precisely the ingredients of a growth recession, and if one begins in 2027 my 2027 medians of ~2.0 will look optimistic (though the ~20–22% mass below zero in each band partly covers that).

Evidence that would move me >10pp (or >half the 10–90 band)

• The 2026-09-15/16 FOMC decision and, more importantly, the dot plot / Warsh press conference signalling a hiking cycle vs. a one-and-done — a signalling of multiple hikes would cut my 2027 medians by ~0.5pp and lower the p10 by ~1pp.
• The 2026-09-16 retail sales/inventories, 9/17 housing starts, 9/25 advance manufacturing, 9/30 Q2 third estimate and 10/6 trade prints, and the next two GDPNow updates: a break of GDPNow below ~2% or above ~5% would move my 2026Q3 median by >1pp.
• Any ceasefire/Oil below $75 (or a further escalation taking oil >$130) — this is the single largest swing factor for 2027 growth.
• Q3/Q4 AI-capex data (data-centre capex guidance from hyperscalers, semiconductor imports) — a stall in the AI build-out would knock ~0.5–1pp off 2027–2028 medians, exactly the "AI investment correction" risk cited by 59% of economists.
• A government shutdown starting 2026-10-01 that delays the advance releases, changing which figure resolves the question.