SEP Forecast Monitor
December 8–9, 2026
Next FOMC projection round

The front table,
before the meeting.

A systematic, vintage-aware forecast of the Federal Reserve’s Summary of Economic Projections—anchored to the prior SEP and updated with the latest professional consensus and released macro data.

Forecast front table

Calendar-year projections

Percent
Variable2026202720282029
Real GDP growth2.3CT2.2–2.4Range2.1–2.62.4CT2.2–2.6Range2.0–2.92.2CT2.1–2.3Range1.8–2.62.1CT2.0–2.2Range1.8–2.4
Unemployment rate4.1CT4.1–4.2Range4.0–4.34.1CT4.0–4.2Range4.0–4.54.1CT4.0–4.2Range3.9–4.44.1CT4.0–4.3Range3.8–4.4
PCE inflation3.7CT3.5–3.7Range2.9–3.82.4CT2.3–2.6Range2.1–2.72.1CT2.0–2.2Range2.0–2.22.0CT2.0–2.0Range2.0–2.0
Core PCE inflation3.3CT3.2–3.3Range2.7–3.42.5CT2.3–2.6Range2.1–2.72.2CT2.0–2.2Range2.0–2.32.0CT2.0–2.0Range2.0–2.1
Large figure: medianCT Central tendencyRange Full range
Source freshness check passed

6 of 6 required series have a new eligible vintage since the SPF deadline; all monthly series meet the observation-period floor.

Versus June 2026 SEP

Forecast median changes

Percentage points; after SEP-style rounding.

Variable2026202720282029
Real GDP growthNo changeNo changeNo changeNo change
Unemployment rateNo changeNo changeNo changeNo change
PCE inflationNo change+0.1No changeNo change
Core PCE inflation-0.1No changeNo changeNo change
Information set

What the model has seen

The live run records the exact database row used for every series at both the SPF deadline and forecast origin. Capture timestamps are the admissibility boundary for live FRED observations.

New vintage

Core PCE prices

Aug 2026

Freshness floor met
New vintage

Industrial production

Sep 2026

Freshness floor met
New vintage

Payroll employment

Sep 2026

Freshness floor met
New vintage

PCE prices

Aug 2026

Freshness floor met
New vintage

Real GDP

Sep 2026

Freshness floor met
New vintage

Unemployment rate

Sep 2026

Freshness floor met
Frozen methodology

A disciplined chain from public data to the table.

Each historical prediction is generated only with information that was public at its forecast origin. The same frozen specification produces this live run.

  1. 01

    Prior SEP

    Start from the last published median, central tendency, and full range for each horizon.

  2. 02

    Professional consensus

    Map the 2026Q3 SPF into SEP definitions, including respondent-level Q4/Q4 GDP growth.

  3. 03

    Released-data update

    A small Minnesota-prior BVAR updates that consensus with growth, labor, and inflation news.

  4. 04

    Frozen gate & shape

    A rolling one-standard-error rule decides whether to leave persistence in place; the selected median shift moves the prior distribution endpoints.

0.226Median MAE
0.241Persistence MAE

Strict public-vintage backtest: 476 median cells across 34 forecast origins. For non-median endpoints, MAE is 0.260 versus 0.272 for persistence. These are historical research results, not a guarantee.

Reproducibility

Every forecast leaves a receipt.

Outputs are non-overwriting audit bundles with model, code, source-row, cutoff, and fallback provenance. No post-cutoff release can enter a locked run.

Run ID
20261031T150052066675Z
Model
sep_calendar_year_front_table_v1
Training through
20260916
Macro provenance
12 cutoff-specific records
Model spec SHA-256
188ae75f606ec7449b5bd4a007a2e999828bd759a5679cc966b9d65c04a78ff7
Source code SHA-256
1252a6ea3bebf71881102302e219cbf20e3fe1788e2ee9cdc708fa1bbe40b8f0
Forecast cells audited
16 medians
SPF vintage note
Historical files are cumulative and may include later errata.
Primary public sourcesFederal Reserve SEPPhiladelphia Fed SPFBLS Employment SituationBEA releases