thesisPrototype

Government data

US MTS September 2026 deficit

U.S. Department of the Treasury Monthly Treasury Statement Table 1 current-month Total Surplus (+) or Deficit (-) for September 2026, not seasonally adjusted, first print, converted to deficit-positive USD billions

Forecast

$85B
$-180B–$350B
80% prediction interval
gpt-5.5Run record ↗
-298-222545302019-092025-09Oct 2026$85B
historyforecast path80% interval
Probability distribution

Cumulative probability

Chance that the outcome is at or below each value.

0%25%50%75%100%$-577.5B$85B$747.5BForecast value

Shaded band: 80% interval ($-180B–$350B). Dashed line: point estimate ($85B).

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

Analysis

§September 2026 MTS Deficit Forecast

Framing and exact resolver: the target is Treasury Monthly Treasury Statement Table 1, current-month Total Surplus (+) or Deficit (-) for September 2026, not fiscal-year-to-date deficit, receipts, outlays, refunds, or a revised historical table. The FiscalData series is represented by FRED mirror MTSDS133FMS for history, but final resolution is the Treasury FiscalData/MTS first print.

Variant discipline: all anchors below use the not seasonally adjusted monthly MTS surplus/deficit concept. I convert the published Treasury surplus-positive series into the registered deficit-positive usd_billions unit by multiplying by -0.001; the registered sourceBinding transform factor of 1 is treated as the adapter-side binding, while this forecast states the unit conversion explicitly.

Release timing: Treasury Fiscal Service MTS documentation says the statement is normally released on the 8th workday after the reporting month, and the release-calendar mirror lists Monthly Treasury Statement dates including 2026-09-11 at 1:00 pm CT and 2026-10-13 at 1:00 pm CT; the registered ledger nevertheless binds this target to the conservative expected-release-window end, 2026-10-25, which I keep as the resolutionDate.

▸ Reported tool use: official.lookupmodel report
Opened the official FY2025 September Monthly Treasury Statement PDF, Table 1 and Table 2.
↳ Fetched September 2025 Table 1 values: receipts 543,663 million, outlays 345,713 million, and Total Surplus (+) or Deficit (-) +197,950 million, which converts to -197.950 usd_billions deficit-positive. The same PDF says FY2025 full-year deficit was 1,775,357 million.
▸ Reported tool use: official.lookupmodel report
Read the FRED Treasury MTSDS133FMS data table as a Treasury Fiscal Service history mirror for latest available 2026 monthly deficit/surplus prints.
↳ Fetched latest available MTSDS133FMS values: 2026-03 -164,100.33882 million, 2026-04 +215,024.14020 million, 2026-05 -292,648.46208 million, 2026-06 -120,305.27559 million, and 2026-07 -432,307.87462 million; in deficit-positive usd_billions, July 2026 is 432.308.
▸ Reported tool use: official.lookupmodel report
Fetched the September reference class from the FRED Treasury MTSDS133FMS table and converted each print to deficit-positive usd_billions.
↳ Fetched September values: 2019 +82,767.65846 million, 2020 -124,610.55842 million, 2021 -61,544.25121 million, 2022 -429,673.47896 million, 2023 -170,982.20087 million, 2024 +64,263.00887 million, and 2025 +197,949.63036 million; converted deficit-positive values are -82.768, 124.611, 61.544, 429.673, 170.982, -64.263, and -197.950 usd_billions.

Base rate/reference class: the same-calendar-month September reference class is the right base rate because MTS monthly flows are dominated by quarterly tax dates and benefit-payment timing. The 2019-2025 September sample has mean deficit 63.119 usd_billions and median deficit 61.544 usd_billions after converting surpluses to negative values.

Prior/update/interval: persistence prior = September reference-class mean 63.119 usd_billions, historical sample = 2019-2025 September MTSDS133FMS converted to deficit-positive usd_billions, model prior = no separate AR/trend/monthly time-series model because monthly MTS flows are dominated by calendar and tax-payment timing and the only public 2026 observations available in the draft are incomplete pre-September months. Adjustment components = +25 for large recent FY2026 monthly deficits, using directional evidence from May 2026 at 292.648 usd_billions and July 2026 at 432.308 usd_billions rather than a fully cited cumulative through-July comparison, -3 because October 1 2026 is a business day so there is no 2023-style pull-forward of October payments into September, and 0 for one-off policy because tariff refunds and tax-receipt strength point in opposite directions. Final point = 63.119 + 25 - 3 = 85.119, rounded to 85. For this flow series I size the interval from the values themselves: [-82.768,124.611,61.544,429.673,170.982,-64.263,-197.950] gives sigma = 206.34 usd_billions; 1.28*sigma = 1.28*206.34 = 264.11, so rounded 80% bounds are 85 - 265 = -180 and 85 + 265 = 350. The seven-observation sample is small and includes COVID/payment-timing outliers, so I keep the mechanically implied wide interval.

Upside risk: a weak September tax-payment month, continued customs-duty refunds, unusually high net interest, or delayed agency spending that lands in September would push the deficit above 350 usd_billions. Downside risk: stronger individual/corporate estimated tax receipts, lower-than-expected refund payments, or September outlays suppressed by fiscal-year-end accounting would land below -180 usd_billions as a large surplus. Outside the interval would require a 2022-like special outlay shock on the high side or a repeat of the unusually low-outlay September 2025 configuration on the low side.

Review disposition: accepted the request to state why no separate AR/trend model was used and softened the +25 update to directional evidence from cited 2026 monthly prints rather than an uncited through-July cumulative comparison; accepted the clarification about adapter/unit conversion and the small-sample interval caveat.

Key drivers

  • same-month September seasonality and quarterly tax receipts
  • large recent FY2026 monthly deficits through July
  • September 1 2026 is a business day, unlike Labor Day 2025
  • October 1 2026 is a business day, limiting pull-forward outlays
  • higher interest and entitlement outlays lift the deficit baseline
  • tariff refund uncertainty can reduce net receipts

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. Department of the Treasury, Bureau of the Fiscal Service, Monthly Treasury Statement Table 1
Resolution date
October 25, 2026· outcome not recorded
Resolution rule
Resolve from the first official Monthly Treasury Statement Table 1 value for Total Surplus (+) or Deficit (-) for September 2026, current-month amount in USD millions. Convert to usd_billions as deficit-positive by multiplying the published surplus/deficit value by -0.001, so a published surplus is negative and a published deficit is positive. Use the first print only; ignore later revisions, corrected historical tables, and same-day correction grace unless Treasury replaces the first-print table before it is observable. The registered target binds resolutionDate to the expected-release-window end, 2026-10-25.
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 mostly publishable, but it should explicitly address the time-series/model-prior choice and tighten one evidence-to-update claim before publication.

  • warning model_prior: The draft uses a September same-month persistence prior, but it does not explicitly say whether a broader time-series/model prior was used or ruled out.
  • warning update: The +25 adjustment relies on FY2026 being about USD 170B worse through July, but the draft does not show the supporting arithmetic or cite the compared FY2025/FY2026 cumulative deficit values in the reasoning.
  • info optional_suggestion: Clarify whether the registered adapter already returns deficit-positive usd_billions despite the Table 1 published value being surplus-positive USD millions, since the sourceBinding transform says factor 1 while the reasoning applies -0.001.
  • info optional_suggestion: The interval method is acceptable, but note that the 2019-2025 sample is only seven observations and includes COVID/payment-timing outliers.

disposition accepted: Review disposition: accepted the request to state why no separate AR/trend model was used and softened the +25 update to directional evidence from cited 2026 monthly prints rather than an uncited through-July cumulative comparison; accepted the clarification about adapter/unit conversion and the small-sample interval caveat.

disposition accepted: Review disposition: accepted the request to state why no separate AR/trend model was used and softened the +25 update to directional evidence from cited 2026 monthly prints rather than an uncited through-July cumulative comparison; accepted the clarification about adapter/unit conversion and the small-sample interval caveat.

disposition not applicable: Review disposition: accepted the request to state why no separate AR/trend model was used and softened the +25 update to directional evidence from cited 2026 monthly prints rather than an uncited through-July cumulative comparison; accepted the clarification about adapter/unit conversion and the small-sample interval caveat.

disposition not applicable: Review disposition: accepted the request to state why no separate AR/trend model was used and softened the +25 update to directional evidence from cited 2026 monthly prints rather than an uncited through-July cumulative comparison; accepted the clarification about adapter/unit conversion and the small-sample interval caveat.

Activity artifacts

Complete original trace

§September 2026 MTS Deficit Forecast

Framing and exact resolver: the target is Treasury Monthly Treasury Statement Table 1, current-month Total Surplus (+) or Deficit (-) for September 2026, not fiscal-year-to-date deficit, receipts, outlays, refunds, or a revised historical table. The FiscalData series is represented by FRED mirror MTSDS133FMS for history, but final resolution is the Treasury FiscalData/MTS first print.

Variant discipline: all anchors below use the not seasonally adjusted monthly MTS surplus/deficit concept. I convert the published Treasury surplus-positive series into the registered deficit-positive usd_billions unit by multiplying by -0.001; the registered sourceBinding transform factor of 1 is treated as the adapter-side binding, while this forecast states the unit conversion explicitly.

Release timing: Treasury Fiscal Service MTS documentation says the statement is normally released on the 8th workday after the reporting month, and the release-calendar mirror lists Monthly Treasury Statement dates including 2026-09-11 at 1:00 pm CT and 2026-10-13 at 1:00 pm CT; the registered ledger nevertheless binds this target to the conservative expected-release-window end, 2026-10-25, which I keep as the resolutionDate.

▸ Reported tool use: official.lookupmodel report
Opened the official FY2025 September Monthly Treasury Statement PDF, Table 1 and Table 2.
↳ Fetched September 2025 Table 1 values: receipts 543,663 million, outlays 345,713 million, and Total Surplus (+) or Deficit (-) +197,950 million, which converts to -197.950 usd_billions deficit-positive. The same PDF says FY2025 full-year deficit was 1,775,357 million.
▸ Reported tool use: official.lookupmodel report
Read the FRED Treasury MTSDS133FMS data table as a Treasury Fiscal Service history mirror for latest available 2026 monthly deficit/surplus prints.
↳ Fetched latest available MTSDS133FMS values: 2026-03 -164,100.33882 million, 2026-04 +215,024.14020 million, 2026-05 -292,648.46208 million, 2026-06 -120,305.27559 million, and 2026-07 -432,307.87462 million; in deficit-positive usd_billions, July 2026 is 432.308.
▸ Reported tool use: official.lookupmodel report
Fetched the September reference class from the FRED Treasury MTSDS133FMS table and converted each print to deficit-positive usd_billions.
↳ Fetched September values: 2019 +82,767.65846 million, 2020 -124,610.55842 million, 2021 -61,544.25121 million, 2022 -429,673.47896 million, 2023 -170,982.20087 million, 2024 +64,263.00887 million, and 2025 +197,949.63036 million; converted deficit-positive values are -82.768, 124.611, 61.544, 429.673, 170.982, -64.263, and -197.950 usd_billions.

Base rate/reference class: the same-calendar-month September reference class is the right base rate because MTS monthly flows are dominated by quarterly tax dates and benefit-payment timing. The 2019-2025 September sample has mean deficit 63.119 usd_billions and median deficit 61.544 usd_billions after converting surpluses to negative values.

Prior/update/interval: persistence prior = September reference-class mean 63.119 usd_billions, historical sample = 2019-2025 September MTSDS133FMS converted to deficit-positive usd_billions, model prior = no separate AR/trend/monthly time-series model because monthly MTS flows are dominated by calendar and tax-payment timing and the only public 2026 observations available in the draft are incomplete pre-September months. Adjustment components = +25 for large recent FY2026 monthly deficits, using directional evidence from May 2026 at 292.648 usd_billions and July 2026 at 432.308 usd_billions rather than a fully cited cumulative through-July comparison, -3 because October 1 2026 is a business day so there is no 2023-style pull-forward of October payments into September, and 0 for one-off policy because tariff refunds and tax-receipt strength point in opposite directions. Final point = 63.119 + 25 - 3 = 85.119, rounded to 85. For this flow series I size the interval from the values themselves: [-82.768,124.611,61.544,429.673,170.982,-64.263,-197.950] gives sigma = 206.34 usd_billions; 1.28*sigma = 1.28*206.34 = 264.11, so rounded 80% bounds are 85 - 265 = -180 and 85 + 265 = 350. The seven-observation sample is small and includes COVID/payment-timing outliers, so I keep the mechanically implied wide interval.

Upside risk: a weak September tax-payment month, continued customs-duty refunds, unusually high net interest, or delayed agency spending that lands in September would push the deficit above 350 usd_billions. Downside risk: stronger individual/corporate estimated tax receipts, lower-than-expected refund payments, or September outlays suppressed by fiscal-year-end accounting would land below -180 usd_billions as a large surplus. Outside the interval would require a 2022-like special outlay shock on the high side or a repeat of the unusually low-outlay September 2025 configuration on the low side.

Review disposition: accepted the request to state why no separate AR/trend model was used and softened the +25 update to directional evidence from cited 2026 monthly prints rather than an uncited through-July cumulative comparison; accepted the clarification about adapter/unit conversion and the small-sample interval caveat.

calibrated forecast · 80% CI
$85B[$-180B · $350B]
Target metadata

Data point: treasury.mts.monthly_deficit.september_2026.first_print

More government data forecasts