thesisPrototype

Government data

US private R&D fixed investment, Q3 2026

What will BEA report as the first-print nominal seasonally adjusted annual-rate value for private fixed investment in research and development, NIPA Table 5.3.5 line 18, in 2026-Q3?

Forecast

$964B
$934B–$994B
80% prediction interval
gpt-5.5Run record ↗
7908679451,0222025-Q12026-Q2Oct 2026$964B
historyforecast path80% interval
Probability distribution

Cumulative probability

Chance that the outcome is at or below each value.

0%25%50%75%100%$889B$964B$1,039BForecast value

Shaded band: 80% interval ($934B–$994B). Dashed line: point estimate ($964B).

Derived from the point estimate and 80% interval; the agent did not report a full distribution.

Analysis

§US private research and development fixed investment, 2026-Q3 first print

Resolver framing: this is BEA NIPA Table 5.3.5 line 18, Research and development, nominal seasonally adjusted annual rate. The registered target unit is usd_billions, so the resolver multiplies BEA's table value in millions by 0.001. The exact source concept is T50305:L18, mirrored by FRED/ALFRED series Y006RC1Q027SBEA; FRED and ALFRED are history mirrors only, and BEA is the resolving source.

▸ Reported tool use: official.lookupmodel report
Checked BEA's 2026 release schedule and the current GDP advance-release page for the release date and current macro context.
↳ BEA's schedule lists GDP (Advance Estimate), 3rd Quarter 2026 on October 29, 2026 at 8:30 AM; the Q2 2026 advance release reported real GDP +1.5 percent SAAR, current-dollar GDP +7.9 percent SAAR, real final sales to private domestic purchasers +3.9 percent, and PCE prices +5.1 percent.

Variant check: all historical anchors are the same nominal seasonally adjusted annual-rate R&D fixed-investment variant as the target, not real growth rates, not annual data, and not government R&D.

▸ Reported tool use: official.lookupmodel report
Fetched the latest FRED public BEA mirror for Y006RC1Q027SBEA, the BEA R&D fixed-investment series corresponding to NIPA Table 5.3.5 line 18.
↳ The fetched recent values were 2025-Q2 829.769, 2025-Q3 855.584, 2025-Q4 881.507, 2026-Q1 909.507, and 2026-Q2 937.772 usd_billions; FRED identifies the source as BEA, release Gross Domestic Product, units billions of dollars SAAR, quarterly.
▸ Reported tool use: official.lookupmodel report
Checked ALFRED's GDP release history for the same BEA/FRED series to extend the recent reference class back one additional distinct quarter.
↳ The ALFRED GDP release page for the nonresidential tag shows Y006RC1Q027SBEA, Gross Private Domestic Investment: Fixed Investment: Nonresidential: Intellectual Property Products: Research and Development, with Q1 2025 at 817.780 usd_billions on the 2025-06-26 release page.
▸ Reported tool use: official.lookupmodel report
Checked BEA's Q2 2026 advance-release technical notes for source-data mechanisms relevant to the next advance print.
↳ BEA reported Q2 2026 real GDP +1.5 percent SAAR and current-dollar GDP +7.9 percent SAAR; Q2 investment increased, intellectual property products increased, and BEA attributed the R&D component to a judgmental trend and BLS Current Employment Statistics.

Base rate/reference class: over the six fetched quarterly levels from 2025-Q1 through 2026-Q2, the five one-quarter changes were +11.989, +25.815, +25.923, +28.000, and +28.265 usd_billions, for a mean base rate of about +23.998 usd_billions per quarter.

Prior/update/interval: persistence prior is the last fetched level, 937.772, plus the 2025-Q1 to 2026-Q2 mean quarterly change of 23.998, giving 961.770. I add about 2.2 for continued nominal momentum, the cited Q2 BEA IPP/R&D strength, and the upward recent run rate in the fetched series, giving a point estimate of 964. For interval sizing, sigma is computed from the fetched successive quarterly level changes: sigma = 6.809 usd_billions, so 1.28*sigma = 8.715. I judgmentally widen to a 30 usd_billions half-width because the Q3 first print follows BEA's September 30, 2026 annual update and the advance estimate uses partial source data plus projections; final bounds are 964-30=934 and 964+30=994.

Upside risk: stronger Q3 nominal GDP, continued IPP/R&D acceleration, or an upward annual-update level revision would land above the interval. Downside risk: weaker R&D source inputs, slower corporate R&D capitalization, or a downward annual-update rebasing would land below the interval. An outside the interval outcome would most likely require both a large annual-update level shift and an unusually large Q3 R&D source-data surprise.

Review disposition: accepted the critique by tying the +2.2 update to cited BEA/FRED evidence rather than uncited AI/software/payroll claims, explicitly stating FRED/ALFRED are mirrors, and clarifying that the 30 billion half-width is judgmental widening beyond recent realized volatility.

Key drivers

  • recent nominal R&D investment level rose in each listed quarter
  • Q2 advance release reported current-dollar GDP up 7.9 percent SAAR
  • BEA said Q2 intellectual property products growth included research and development
  • September 2026 annual update can rebase the level before the Q3 advance print
  • advance GDP estimates use partial source data and BEA projections

Tool evidence

This run has no captured tool responses. “Reported tool use” in the analysis is the model’s account, not an archived tool response.

Sources and resolution

Official source
U.S. Bureau of Economic Analysis GDP advance release, NIPA Table 5.3.5 line 18
Resolution date
October 29, 2026· outcome not recorded
Resolution rule
Resolve to the first official BEA GDP advance-release print for 2026-Q3, NIPA Table 5.3.5, line 18 Research and development, nominal seasonally adjusted annual rate. Use the published table value transformed by multiplying the table's millions-of-dollars value by 0.001 to usd_billions, retaining the precision implied by the official table; do not use later second, third, annual-update, or comprehensive-revision values except insofar as BEA has already incorporated them into the first Q3 advance-release table on 2026-10-29.

Forecast history

Select a version to read its estimate and analysis.

Forecast versions, estimates, and intervals
VersionDateEstimate80% interval
gpt-5.5Sep 21, 2026$962B$934B–$988.4B
gpt-5.5 · selectedAug 17, 2026$964B$934B–$994B
Run details

The analysis is the model’s written report. Tool-use descriptions in that report are model claims; the activity artifacts contain the execution record.

thesis.analyst · gpt-5.5 · fast · v2.5.11

pre-submit review · completed

Draft is publishable with no blocking issues; resolver, unit, first-print rule, prior/update/interval structure, and JSON fields are coherent.

  • warning update: The +2.2 inside-view adjustment cites software, AI, and professional payroll trends, but the shown evidence only directly supports recent BEA R&D/IPP strength and Q2 macro context.
  • info optional_suggestion: State explicitly that FRED/ALFRED are mirrors only and BEA is the resolving source.
  • info optional_suggestion: Clarify whether the 30 billion half-width is judgmental widening beyond realized recent-quarter volatility rather than mechanically implied by the five-change sample.

disposition accepted: Review disposition: accepted the critique by tying the +2.2 update to cited BEA/FRED evidence rather than uncited AI/software/payroll claims, explicitly stating FRED/ALFRED are mirrors, and clarifying that the 30 billion half-width is judgmental widening beyond recent realized volatility.

disposition not applicable: Review disposition: accepted the critique by tying the +2.2 update to cited BEA/FRED evidence rather than uncited AI/software/payroll claims, explicitly stating FRED/ALFRED are mirrors, and clarifying that the 30 billion half-width is judgmental widening beyond recent realized volatility.

disposition not applicable: Review disposition: accepted the critique by tying the +2.2 update to cited BEA/FRED evidence rather than uncited AI/software/payroll claims, explicitly stating FRED/ALFRED are mirrors, and clarifying that the 30 billion half-width is judgmental widening beyond recent realized volatility.

Activity artifacts

Complete original trace

§US private research and development fixed investment, 2026-Q3 first print

Resolver framing: this is BEA NIPA Table 5.3.5 line 18, Research and development, nominal seasonally adjusted annual rate. The registered target unit is usd_billions, so the resolver multiplies BEA's table value in millions by 0.001. The exact source concept is T50305:L18, mirrored by FRED/ALFRED series Y006RC1Q027SBEA; FRED and ALFRED are history mirrors only, and BEA is the resolving source.

▸ Reported tool use: official.lookupmodel report
Checked BEA's 2026 release schedule and the current GDP advance-release page for the release date and current macro context.
↳ BEA's schedule lists GDP (Advance Estimate), 3rd Quarter 2026 on October 29, 2026 at 8:30 AM; the Q2 2026 advance release reported real GDP +1.5 percent SAAR, current-dollar GDP +7.9 percent SAAR, real final sales to private domestic purchasers +3.9 percent, and PCE prices +5.1 percent.

Variant check: all historical anchors are the same nominal seasonally adjusted annual-rate R&D fixed-investment variant as the target, not real growth rates, not annual data, and not government R&D.

▸ Reported tool use: official.lookupmodel report
Fetched the latest FRED public BEA mirror for Y006RC1Q027SBEA, the BEA R&D fixed-investment series corresponding to NIPA Table 5.3.5 line 18.
↳ The fetched recent values were 2025-Q2 829.769, 2025-Q3 855.584, 2025-Q4 881.507, 2026-Q1 909.507, and 2026-Q2 937.772 usd_billions; FRED identifies the source as BEA, release Gross Domestic Product, units billions of dollars SAAR, quarterly.
▸ Reported tool use: official.lookupmodel report
Checked ALFRED's GDP release history for the same BEA/FRED series to extend the recent reference class back one additional distinct quarter.
↳ The ALFRED GDP release page for the nonresidential tag shows Y006RC1Q027SBEA, Gross Private Domestic Investment: Fixed Investment: Nonresidential: Intellectual Property Products: Research and Development, with Q1 2025 at 817.780 usd_billions on the 2025-06-26 release page.
▸ Reported tool use: official.lookupmodel report
Checked BEA's Q2 2026 advance-release technical notes for source-data mechanisms relevant to the next advance print.
↳ BEA reported Q2 2026 real GDP +1.5 percent SAAR and current-dollar GDP +7.9 percent SAAR; Q2 investment increased, intellectual property products increased, and BEA attributed the R&D component to a judgmental trend and BLS Current Employment Statistics.

Base rate/reference class: over the six fetched quarterly levels from 2025-Q1 through 2026-Q2, the five one-quarter changes were +11.989, +25.815, +25.923, +28.000, and +28.265 usd_billions, for a mean base rate of about +23.998 usd_billions per quarter.

Prior/update/interval: persistence prior is the last fetched level, 937.772, plus the 2025-Q1 to 2026-Q2 mean quarterly change of 23.998, giving 961.770. I add about 2.2 for continued nominal momentum, the cited Q2 BEA IPP/R&D strength, and the upward recent run rate in the fetched series, giving a point estimate of 964. For interval sizing, sigma is computed from the fetched successive quarterly level changes: sigma = 6.809 usd_billions, so 1.28*sigma = 8.715. I judgmentally widen to a 30 usd_billions half-width because the Q3 first print follows BEA's September 30, 2026 annual update and the advance estimate uses partial source data plus projections; final bounds are 964-30=934 and 964+30=994.

Upside risk: stronger Q3 nominal GDP, continued IPP/R&D acceleration, or an upward annual-update level revision would land above the interval. Downside risk: weaker R&D source inputs, slower corporate R&D capitalization, or a downward annual-update rebasing would land below the interval. An outside the interval outcome would most likely require both a large annual-update level shift and an unusually large Q3 R&D source-data surprise.

Review disposition: accepted the critique by tying the +2.2 update to cited BEA/FRED evidence rather than uncited AI/software/payroll claims, explicitly stating FRED/ALFRED are mirrors, and clarifying that the 30 billion half-width is judgmental widening beyond recent realized volatility.

calibrated forecast · 80% CI
$964B[$934B · $994B]
Target metadata

Data point: bea.research_and_development_fixed_investment.2026_q3.first_print

More government data forecasts