Tools chosen: kalshi_direct kalshi_related polymarket_related fred claude_news gdelt_news code_execution

Sub-questions (Fermi decomposition)

Planner reasoning

This is a nominal-dollar year-over-year comparison of IRS collections, which historically rises almost every year due to nominal GDP growth and inflation; the main risks are the OBBBA tax cuts taking effect for tax year 2025/2026 (no tax on tips/overtime, bonus depreciation, expanded deductions), a recession, or IRS enforcement capacity collapse from staffing cuts. Key research is the Kalshi price anchor, historical IRS gross collections/Treasury receipts base rate, current monthly receipts trend, and news on the revenue impact of the 2025 tax law.

# Event Will IRS/federal tax collections in calendar/fiscal 2026 exceed 2025 collections? # Outcomes to forecast - Yes (2026 collections > 2025) - No (2026 collections ≤ 2025) # Kalshi market anchor KXIRSCOLLECT-26 currently trades at **89% YES**, up from a 62% low and +9pts over the last 30 days (7-day change flat). Avg daily volume is thin (~76 contracts/day over 91 days of data), so the price is directional but not deeply liquid. This 89% anchor is well above the model-derived base-rate/Monte-Carlo estimate (see below), suggesting the market is pricing in already-observed 2026 receipts data, not just historical base rates. # Sub-question answers 1. **Kalshi price/volume/trend** — 89% YES, +9pts in 30 days, range 62-90%, low volume (~76 contracts/day). [kalshi_direct] 2. **Historical decline frequency (FY1980-2025)** — Nominal federal receipts declined YoY in 8 of 45 years (17.8%), almost always recession-linked (FY1983, 2001-03, 2008-09, 2020, 2023). Non-recession years saw declines only 2.9% of the time (1/35, the 2023 post-surge normalization). [code_execution] 3. **Current receipt trend** — FRED FGRECPT (federal govt current tax receipts, quarterly annualized) shows continued nominal growth: Q1 2026 = $5,872.5B vs Q1 2025 = $5,392.7B (+8.9% YoY); Q4 2025 ($5,898.5B) also up sharply vs Q4 2024 ($5,313.2B). Individual income tax component (W006RC1Q027SBEA) shows similar ~12% YoY growth into Q1 2026. Trend is clearly rising, not falling. [FRED] 4. **OBBBA drag estimate** — CBO's post-OBBBA baseline shows individual income tax revenue $2.8T lower cumulatively FY2026-2035 vs pre-OBBBA baseline; corporate receipts lower FY2026-2029 but net +$38B over 2026-2035. IRS staffing cuts (~28,000 FTE reduction) projected to reduce revenue ~$600B over 2026-2035 (Budget Lab/Yale). These are relative-to-counterfactual reductions, not necessarily YoY declines. [claude_news] 5. **CBO/Treasury FY2026 vs FY2025 baseline level** — CBO's own baseline (cited via JEC/Senate) projects **total net receipts rising from $5.163T (FY25) to $5.580T (FY26) to $5.935T (FY27)** — i.e., CBO's baseline itself shows nominal growth, aided heavily by tariffs (projected ~7.5% of FY2026 receipts, near record). [claude_news, CBO] 6. **Recession probability 2026** — Not directly quantified in research; code_execution model assumes ~15% chance of a "growth-relevant recession" in FY2026 as an overlay. No sourced forecaster consensus found in this research pass. Some soft-data warning signs (US retail sales posted first decline in 9 months, July 2026, per gdelt/Yahoo Finance) but no recession call. 7. **Resolution data source/timing** — Not explicitly specified in rules; likely uses IRS Data Book gross collections or Treasury/CBO annual receipts, both of which publish well before the Dec 2027 close (IRS Data Book ~spring following year; Treasury MTS monthly). No source-mismatch risk flagged. # Key facts (high-confidence, factual) 1. [Treasury/fiscal.treasury.gov] FY2025 total federal receipts = $5,234.6B, +6.4% ($316.5B) vs FY2024. 2. [claude_news/CBO] FY2025 individual income taxes +9.5% YoY; corporate income taxes -14.7% YoY. 3. [FRED] Q1 2026 annualized current tax receipts (FGRECPT) already +8.9% YoY vs Q1 2025 — real-time data trending up, not down. 4. [CBO via JEC] CBO's own multi-year baseline (post-OBBBA) projects FY2026 net receipts ($5.580T) above FY2025 ($5.163T). 5. [code_execution] Historical base rate: only ~18% of years see nominal YoY declines, nearly all recession-linked; non-recession decline rate is just ~3%. # Cross-market signals - Kalshi related: No direct KXIRSCOLLECT peer markets; unrelated series only (IPO markets, tax-credit reinstatement markets) — no arbitrage signal. - Polymarket: No matching markets found (0/100 scanned). - Sportsbook implied: N/A (not a sports event). # Analyst opinions and speculation - Tax Foundation: OBBBA reduces revenue by $5.2T cumulative 2025-2034 but boosts GDP growth (+0.8% by 2026), partially offsetting via base growth. - Budget Lab/Yale, PGPF, CAP: IRS staffing cuts (30%+ revenue-agent attrition) will erode enforcement-driven collections, though effect builds over years rather than causing an outright 2026 dip. - Fox Business (Aug 2026): FY2026 deficit tracking toward $2.1T, implying spending is outpacing revenue growth but not necessarily that revenue itself is falling. # Directional lean per outcome - **Yes (collects more in 2026)**: Strongly supported — actual FRED data already shows ~9-12% YoY receipt growth into Q1 2026; CBO's own baseline (which bakes in OBBBA) still projects nominal growth FY25→FY26; record tariff revenue is a major structural tailwind; base rate for non-recession years favors increases (~97%). Kalshi at 89% reflects this. - **No (collects less)**: Supported only by OBBBA-driven individual/corporate tax relief, IRS enforcement erosion, and tail recession risk. No confirmed 2026 recession; soft retail-sales dip (July 2026) is the only warning sign. Code_execution's abstract Monte Carlo (ignoring realized 2026 data) implies a much closer race (~53-57% Yes), but this pre-dates/ignores the actual FRED receipts prints showing strong YoY growth already realized. # Gaps / unknowns - Exact resolution source (IRS Data Book vs Treasury MTS vs CBO) not specified in rules — could affect fiscal vs calendar year framing. - No explicit sourced 2026 recession probability from a named forecaster. - Tariff revenue durability (legal challenges to tariff authority) not addressed in research. - Full-year FY2026 data incomplete; only partial-year (Q1) actuals available. # Calibration anchors - Kalshi current YES price: **89%** (anchor, upward trending, low but consistent volume). - Historical base rate for nominal YoY revenue increase (non-recession years): ~97%. - Model-only Monte Carlo (pre-real-data): ~53-57% Yes — likely stale relative to observed Q1 2026 FRED prints showing strong growth, which should pull true probability toward Kalshi's higher estimate.

1. Neither forecaster addresses the ambiguity flagged in the brief about resolution source/timing (IRS Data Book vs Treasury MTS vs CBO, fiscal vs calendar year) — this is a real risk since "collections" could be defined differently and affect whether tariff revenue or refund timing shifts the comparison, yet both treat it as a minor footnote rather than adjusting uncertainty upward. 2. Both forecasts lean heavily on Q1 2026 data extrapolated to the full year, but the brief explicitly notes "Full-year FY2026 data incomplete; only partial-year (Q1) actuals available" — three more quarters of unknown performance (including the OBBBA individual tax relief phase-in and refund season effects) could meaningfully alter the growth trajectory, and neither forecast discusses how much weight a single quarter should carry. 3. Forecast 1's upward deviation from the Kalshi anchor (92% vs 89%) is not well justified given the market's own thin liquidity and volatility (62-90% range, +9pts in 30 days) — if the market is already incorporating the same FRED data forecaster 1 cites, unilaterally overriding it by 3pts on "stronger realized prints" is a weak edge claim without new information the market lacks. 4. Neither forecaster meaningfully engages with the tariff revenue durability risk mentioned in the brief ("legal challenges to tariff authority... not addressed in research") despite CBO's growth baseline explicitly being "aided heavily by tariffs (projected ~7.5% of FY2026 receipts, near record)" — if tariff revenue is enjoined mid-year, this is a plausible non-recession pathway to a shortfall that both forecasts implicitly dismiss. 5. The 15% recession-overlay assumption from the code_execution Monte Carlo (53-57% Yes) is dismissed by both forecasters as "stale," but neither explains why the current soft-data warning sign (first retail sales decline in 9 months, July 2026) shouldn't at least modestly raise recession-tail risk mid-fiscal-year, especially since a recession-linked decline is the single largest historical driver of the "No" outcome (8 of 8 non-2023 declines were recession years). 6. Both forecasts converge very close to each other (89-92%) and to the Kalshi price without much independent daylight, suggesting anchoring bias rather than independent synthesis — given the brief highlights a notable gap between the market (89%) and a model-only estimate (53-57%), a wider band of scenario analysis (e.g., what specifically would need to happen for No) is underexplored in both rationales.

90% (70% confidence)

Yes: 90% No: 10%

Realized data dominates: FRED shows Q1 2026 federal current tax receipts +8.9% YoY (individual income tax ~+12%), and CBO's own post-OBBBA baseline projects FY2026 receipts of $5.58T vs $5.16T in FY2025, so the growth path is already partly banked rather than merely forecast. The historical base rate reinforces this — nominal receipts decline in only ~18% of years and ~3% of non-recession years — and no 2026 recession has been called. The critique's points are real but second-order: resolution-source ambiguity (IRS Data Book gross collections would exclude tariffs but also exclude refund netting), tariff litigation risk, OBBBA refund-season drag, and mid-year recession tail risk collectively justify keeping ~10% on No rather than pushing toward the 95%+ that pure momentum would imply. I therefore sit essentially at the thin-but-consistent Kalshi anchor of 89-90%, declining to claim an edge over the market on data the market has plainly already absorbed.

Total pipeline time: 270.4s