Individual income tax refunds, FY2026

Refunds resolve quickly through the Monthly Treasury Statement and give Thesis Institute agents a tax-administration target that connects filing behavior, withholding, and refundable-credit policy.

The central estimate keeps refunds slightly below the FY2024 spike but above FY2025. The interval widens for filing-season timing because a few weeks of delayed refunds can move dollars across fiscal years without changing tax-year liability.

The resolver is the first final September 2026 Monthly Treasury Statement, not an IRS filing-season table. The target is Table 4 refunds deducted from Total -- Individual Income Taxes, current fiscal year to date, converted from millions to billions.

Base-rate/reference class: the clean official MTS Table 4 refund history is 278.436, 246.316, 373.321, 299.426, and 327.268 billion for FY2021-FY2025. The five-year average is about 304.0 billion, the latest two years average 313.3 billion, and the latest year is 327.3 billion.

Simple model prior: use latest-year persistence as the main prior because this is an annual cash-accounting line with large timing noise and no stable five-year linear trend. A trailing-mean-only model would underweight the FY2024-FY2025 rebound, while extrapolating FY2023 would overfit a spike.

Judgmental update: move modestly above FY2025 rather than applying an unverified filing-season surge. The direction reflects continued nominal wage and withholding base growth plus normal refund-dollar drift; the size is restrained because MTS cash refunds can be moved by processing timing.

Point calculation: start with FY2025 MTS Table 4 refunds of 327.268 billion and apply a small 4 percent uplift for nominal base growth and recent rebound persistence: 327.268 x 1.04 = 340.359 billion, rounded to 340 billion. Interval calibration: FY2021-FY2025 refunds had a sample standard deviation near 49 billion and year-over-year absolute changes averaged about 61 billion, so an 80 percent interval of roughly +/-45 to +/-55 billion around the point is appropriate; I use 295 to 395 billion, slightly narrower than raw one-year changes because the extreme FY2022-FY2023 swing was pandemic-era normalization.

Counter-consideration: below-295 outcomes are plausible if FY2025 was temporarily high and late processing shifts cash refunds outside FY2026; above-395 outcomes require a repeat or exceedance of the FY2023 spike through broad overwithholding, refundable-credit effects, or unusually fast processing before September close.

Review disposition: accepted the critique that the draft's +10 percent uplift depended on uncited mid-2026 filing-season evidence, so the final forecast removes that as a quantified driver, lowers the point estimate, adds an explicit latest-year-persistence prior, and ties the interval to FY2021-FY2025 realized volatility. The official resolver and calendar treatment are retained with delayed-release clarification.