Forecast locked · Awaiting grade · Expected November 9
Will households put higher odds on rising unemployment? Our October call
October 14, 2026 · 1,000 synthetic profiles · Before release, during October fieldwork
Our forecast for the New York Fed's October reading is a mean 47.33% subjective chance that U.S. unemployment will be higher 12 months from now. September's reading was 43.87%. This forecasts a survey mean, not the actual unemployment rate or the share of people answering Yes.
Our forecast
47.33%
Mean subjective probability
September baseline
43.87%
Full precision retained for grading
Official October outcome
Pending
Expected November 9, 11 a.m. Eastern
Why this call
The question is whether households' unemployment expectations have increased since September. That outlook can provide context for consumer-demand and hiring plans; this forecast does not measure changes in spending or hiring. Our call is about 3.5 percentage points above September. The official October answer is expected November 9 at 11 a.m. Eastern / 16:00 UTC.
The frozen grading rule
We fixed what counts as a win before any responses: our absolute error must be smaller than simply carrying forward September, and no greater than 1.07501220703125 percentage points. Both conditions must hold. Ties fail.
abs(F−Y) < abs(B−Y) AND abs(F−Y) ≤ T
F = 47.33; B = 43.865966796875; T = 1.07501220703125 points.
That tolerance is the median absolute change across 23 monthly steps, using October 2024 through September 2026. It is a historical movement threshold, not a confidence interval. For this frozen forecast, October values from 46.25498779296875% through 48.40501220703125%, inclusive, satisfy both conditions. We compare full precision; rounded display values do not determine the grade.
Grading will use the first published full-precision October workbook value on the Unemployment Expectations sheet. The outcome is pending. First-release bytes, timestamp and value will be retained; later revisions will be preserved separately, and a failure will stay visible.
How we got 47.33%
We used 1,000 synthetic profiles grounded in adult ACS household reference-person records across all 50 states and DC, a proxy for SCE household heads. All 1,000 generated a valid numeric answer; none were invalid or missing. We preserved those numbers and used their exact arithmetic mean, with no fitted adjustment.
| Probability | Profiles |
|---|---|
| 35% | 201 |
| 45% | 524 |
| 55% | 116 |
| 65% | 159 |
The answers concentrate at just four probabilities, so the exact mean does not imply finely resolved individual judgments or forecast certainty. Primary weights are equal. Selection uses approximate source-derived sampling; it does not recreate a human probability sample. The 1,000 profiles contain 977 distinct demographic prompts and share one model and a fixed seed. They are not 1,000 human participants, and we claim no human margin of error.
Assigned / attempted / valid / invalid / missing: 1,000 / 1,000 / 1,000 / 0 / 0. There were no retries or unattempted records. No clipping, imputation or calibration was applied.
The question and its limits
We retained the official probability and economy introductions and asked:
What do you think is the percent chance that 12 months from now the unemployment rate in the U.S. will be higher than it is now?
This was a single-item adaptation: we omitted the separate residential-mobility practice question and replaced the graphical response box with a numeric instruction. The frame uses 2020–2024 ACS housing-unit reference persons, excluding group quarters; it does not reproduce SCE recruitment or human survey weights. Cached urbanicity is a PUMA-name heuristic. This is a call before the release, not before October fieldwork.
Even a miss will be informative: this public series gives the forecaster a clear endpoint and a strong last-reading baseline, without changing the rule after the answer appears. One comparison will not establish general forecasting accuracy.
Data, registration and sources
Protocol locked October 14 at 20:47:38 UTC, before any generated response. The exact forecast was frozen at 20:58:42 UTC. Download the frozen forecast, aggregates, exact instrument and grading rule.
- Official Survey of Consumer Expectations workbook
- Official questionnaire
- November release calendar
- Source terms of use
Some survey questions were taken or adapted from the Survey of Consumer Expectations, ©2013-2020 Federal Reserve Bank of New York. The SCE questions are available without charge at https://www.newyorkfed.org/microeconomics/sce and may be used subject to license terms posted there. The New York Fed did not participate in or endorse the Lewsearch October 2026 synthetic unemployment-expectations study, and the New York Fed disclaims any responsibility or legal liability for the administration of the survey and the analysis and interpretation of data collected.
Source: Survey of Consumer Expectations, ©2013-2020 Federal Reserve Bank of New York. The SCE data are available without charge at https://www.newyorkfed.org/microeconomics/sce and may be used subject to license terms posted there. The New York Fed disclaims any responsibility for this analysis and interpretation of Survey of Consumer Expectations data.