Four-quarter US GDP growth above 5%

Where things stand

• Latest data. BEA's latest four-quarter real GDP growth (Table 1.1.11, line 1; FRED series A191RO1Q156NBEA, downloaded 13 Oct 2026) is 2.2% for Q2 2026. Q1 2026 was 2.6% and Q4 2025 was 2.0%.
  • Q2 2026 grew at 2.2% annualised in the third estimate of 30 Sep 2026, which also included the 2026 annual update.
• Q3 2026, the earliest eligible quarter. The advance estimate is due 29 Oct 2026, with later estimates on 25 Nov and 23 Dec. The Atlanta Fed's GDPNow model is at about 3.6% annualised (8 Oct).
  • From the GDPC1 levels, that implies four-quarter growth of only about 2.1%.
  • To clear 5.05%, Q3 would have needed about 16% annualised growth. Q4 2026 would need about 12%, and Q4 2026 plus Q1 2027 would need to average about 7.2%.
  • So the 2026–27 horizons are close to zero.
• Economic backdrop.
  • The Fed raised rates to 3.75–4% on 16 Sep 2026.
  • The 10-year Treasury yield is above 5%, oil is around $100 because of the Iran war, and quarterly annualised PCE inflation was 5% in Q2.
  • Housing and other rate-sensitive sectors are weak.
  • Productivity growth is 2.2% year on year (FRED OPHNFB).
  • Deloitte's forecast (29 Sep) sees the working-age population flat and growth of about 2% through 2028.
  • The Philadelphia Fed's Survey of Professional Forecasters for Q3 2026 has median annual growth of 2.1–2.4% for 2026–29. From its probability file (prob.xlsx), the mean chance of annual-average growth of 5.5% or more is 0.27%, 0.19% and 0.44% for 2027, 2028 and 2029, and the 4.0–5.4% range gets 3–4%.

Base rate (computed)

• Long history. In the FRED series, 60 of 314 quarters since 1948 exceeded 5%. Since 1985, the current vintage shows only 2000Q2 (5.2%) and the COVID rebound (2021Q2–Q4).
• Earlier published estimates. Using the Philadelphia Fed real-time dataset (routputqvqd.xlsx, every published vintage), some vintages also put 1987Q4–88Q1 and 2000Q1–Q3 above 5%; 2000Q2 was 6.1% in early estimates. Because any published estimate counts, this matters.
• Five-year windows. 29% of 21-quarter windows starting 1985–2021 contained a reading above 5%. That figure comes from only two episodes, both in eras when trend growth was higher.
• Distance from trend. From 1985 to 2019, four-quarter growth never ran more than 2.4 points above its lagged 10-year trend in any 21-quarter window (current vintage). Today it would need about 2.6–2.9 points above a trend of roughly 2.2–2.5%.
• Without an AI boom. I put the chance of a non-AI path at about 3% over five years. This covers a COVID-style collapse and rebound, a wartime build-up, or a generous early estimate.

The AI path, which decides the answer

• Capabilities are moving very fast in this world.
  • Anthropic CEO Dario Amodei's essay "We Must Pace the Frontier" (12 Sep 2026) describes AI increasingly building the next generation of AI.
  • The length of tasks AI can finish on its own (measured by the evaluator METR) is doubling roughly every four months.
  • Anthropic's annualised revenue was about $65bn in July 2026.
• Economic scenarios. Anthropic's economic model (Korinek and Jones, Sep 2026) puts growth in the 12 months to 2030 at 5.4% in its "substantial" scenario and 15.4% in its "extreme" scenario.
• Skeptics. Economists Moll and Imas still call 4–5% a reasonable baseline for the next 10–15 years.
• Expert surveys. The Forecasting Research Institute survey (Mar 2026) gives economists about a 14% chance of "rapid" AI progress by 2030. Even then, their median growth for 2025–29 is only 3.3%; the 10th to 90th percentile range is 1.2–5.5%.
• Prediction markets. On Manifold, a market on 15% per-person growth by 2033 trades at about 29%. A very thin market on annual growth above 5% before 2030 is at 64%; I give it little weight.

My estimate: about a 20% chance that AI drives an acceleration to 4–6% trend growth, or more, by 2030–31. Given that, I put the chance of at least one published reading above 5% by Q3 2031 at roughly 45–70%. I add about 4% for explosive-growth scenarios. A Monte Carlo of these assumptions gives about 16% by the end of 2031.

Forecast shape

The hazard is about 0.1% per quarter through 2027, then rises to roughly 0.5–0.7% per quarter in 2028, about 1% in 2029, and 1.3–1.35% in 2030–31. That gives:

By end of	Chance of YES
2027	0.8%
2028	3%
2029	6.7%
2030	11.4%
2031	16%

The last eligible quarter is Q3 2031, whose advance estimate comes out in late October 2031.

Strongest case against my number

• Lower. Mainstream macro forecasters (the professional survey, the Fed, CBO, Deloitte) imply under 5–8%. Labour supply is stagnant, the Fed is leaning against demand booms, there is no sign of AI in productivity data yet, and the "pacing" push could slow AI.
• Higher. If capabilities spread into the economy on Anthropic's "substantial" timeline, 25–35% would be defensible.

I weight the two sides and stay near 16%.

Question. By the last day of each quarter, has BEA reported four-quarter US real GDP growth above 5% for any quarter?

Definitions.

• BEA: the US Bureau of Economic Analysis.
• Four-quarter US real GDP growth: the seasonally adjusted percent change in total US real gross domestic product from the same quarter one year earlier, as reported in National Income and Product Accounts (NIPA) Table 1.1.11, “Real Gross Domestic Product: Percent Change From Quarter One Year Ago,” line 1 (corresponding FRED series A191RO1Q156NBEA). This measures 100 × (real GDP in quarter t / real GDP in quarter t−4 − 1), not annualized quarter-on-quarter growth or growth in annual-average GDP.
• Above 5%: strictly greater than 5.0% using BEA’s published percent-change value at its published precision. A reported 5.0% does not qualify; calculations from rounded GDP levels do not override the published percent-change value.

Resolves YES if an official BEA publication reports a qualifying value for at least one completed quarter by the horizon. Advance, second, and third estimates, and subsequent annual, comprehensive, or other official revisions, all count. Apply the Observation period convention: the measured quarter must end after the opening date, so Q3 2026 is the earliest eligible quarter. A revision that first raises an eligible quarter above the threshold counts on the revision’s publication date. Once a qualifying value has been published, a later downward revision does not undo YES.

Resolves NO if no qualifying publication exists by the horizon. Exclude forecasts and nowcasts, nominal GDP, GDP per capita, individual GDP components, state or regional GDP, annualized quarter-on-quarter rates, and annual-average growth rates. Historical readings or revisions for quarters ending before the opening date do not count. No attribution to AI is required; growth due to a rebound or other non-AI causes counts.

Resolving source. BEA’s official GDP releases, accompanying NIPA tables, and dated BEA data archives, specifically Table 1.1.11, line 1, or its official successor reporting the same measure. FRED/ALFRED series A191RO1Q156NBEA may locate and corroborate vintages but does not independently resolve the question. Apply the Resolution window convention and Resolution date (A3); use the date BEA first publishes the qualifying estimate, not the measured quarter’s end date.