What the data says, treating the 10Y-3M spread as the only signal.
2022-24 inversion was historically extreme. 780 days inverted (Oct 2022 to Dec 2024), deepest -1.89pp May 2023. More than 2x longer than the prior longest (315 days, 2006-07) and deeper than 2007. Both depth and duration historically extend the lead time before recession.
Past inversion-to-recession lead times (post-1982):
| Un-inverted | Recession | Lead |
|---|---|---|
| Dec 1989 | Jul 1990 | 7 mo |
| Jan 2001 | Mar 2001 | 2 mo |
| May 2007 | Dec 2007 | 7 mo |
| Oct 2019 | Feb 2020 | 4 mo |
Median lead from un-inversion: ~7 months. The 2022-24 inversion's depth and duration argues for the longer end.
Where we are now. Main inversion un-inverted Dec 2024. Eight short re-inversions through 2025 (longest 26 days). Most recent ended Oct 17 2025. Latest reading +0.62pp.
Prediction (yield curve alone):
1973 caveat. 1973 inversion-to-recession lead was only ~4 months — oil shock acted as the trigger, with the curve as confirmation. If 2026 follows the 1973 template, the curve's own forecast becomes too late.
Integrated read: yield curve says recession Q3 2026 to Q2 2027 (Q = quarter, H = half). Comparing to 1973 specifically argues for earlier than the curve says.
The number of new privately-owned housing units on which construction has begun in a given month. Reported by the US Census Bureau as a seasonally-adjusted annualized rate (SAAR) — i.e. "if this monthly pace continued for a full year, this many homes would be started." Reading 1,500k = pace of 1.5 million starts per year.
Why it leads: housing is the most interest-rate-sensitive sector of the economy. When the Fed raises rates, mortgage costs rise, builders pull permits, and construction slows BEFORE the rate hikes feed through to the broader economy. Construction is also a major employer (~5% of US jobs); when it shrinks, jobs vanish quickly.
Track record: housing starts peaked Jan 2006 — fully 23 months before the 2007-09 recession was officially dated. Per the GFC chronology, this was the earliest macro indicator to turn. The single best real-economy leading signal in the modern record.
Publication lag: US Census normally releases each month's data ~17 business days after month-end, but recent government data-release pauses have stretched this. If Feb/Mar look missing here, they likely haven't been published yet — the next poll will pick them up when FRED ingests them.
A real-time recession indicator developed by economist Claudia Sahm. Computed as: 3-month moving average of the unemployment rate, minus its lowest value in the trailing 12 months.
When the indicator exceeds 0.5pp, the US has already been in recession in every postwar case (1949 through 2020). Triggered Aug 2024 and reset shortly after — one of the few false signals in 70+ years.
It works because labour markets deteriorate non-linearly: once unemployment rises 0.5pp above its recent low, momentum carries it much higher. It's a coincident-to-mildly-leading signal — confirms recession faster than NBER's official dating committee, which usually announces 6-12 months after the fact.
The two benchmarks.
The spread. Historically Brent trades at a $1-5 premium to WTI, reflecting the cost of moving WTI from Cushing to a coast where the rest of the world can buy it (pipeline tariffs + Gulf Coast loading). When the spread is normal, the global market is well-arbitraged: any divergence gets closed by US producers shipping more to the Gulf and onto tankers.
Why a wide spread = seaborne disruption. The arbitrage that holds Brent ≈ WTI + ~$3 requires functioning seaborne export channels. WTI can only "reach" the global market via Gulf Coast → tankers. If something disrupts the global seaborne supply pool but doesn't disrupt US pipeline-delivered crude, then:
The spread is essentially measuring the cost of moving crude across water during the disruption. The March 31 2026 print of $25/bbl says with precision: the disruption is not in production capacity, it's in the channel by which barrels reach the global market.
Leading indicator for de-escalation. If/when Hormuz reopens and tanker insurance normalises, the spread will compress back toward $5 long before headlines confirm it. Watch for sustained compression as the cleanest early signal of resolution.
Caveat: a negative spread (WTI > Brent) is also abnormal but indicates a different stress — US domestic infrastructure constraint. The 2010-13 Cushing glut produced -$15 to -$25 spreads as US shale outpaced Permian-to-Gulf pipeline capacity. That's not a global supply story.
| date | indicator | type | description |
|---|---|---|---|
| 2026-05-10 | au_home_value_index_28d | band_change | au_home_value_index_28d moved from green to amber at value 0.1000 (-0.3 to 0.2% — stagnation / inflexion zone) |
| 2026-05-10 | au_total_listings | band_change | au_total_listings moved from green to amber at value 80108.0000 (80-95k — above current; watch zone) |
| 2026-05-10 | qld_diesel_pump | band_change | qld_diesel_pump moved from crisis to amber at value 242.9000 (240–270c/L — elevated) |
| 2026-04-30 | westpac_mi_sentiment | band_change | westpac_mi_sentiment moved from unknown to amber at value 80.1000 (80-90 - cautious; consumption flat) |
| 2026-04-27 | us_brent_wti_spread | band_change | us_brent_wti_spread moved from unknown to amber at value 8.3400 ($5-10/bbl - tankers tight) |
| 2026-04-26 | au_auction_volume | band_change | au_auction_volume moved from unknown to amber at value 757.0000 (Below 800 — light week (Easter / Jan slow-down typical)) |
| 2026-04-26 | auction_clearance | band_change | auction_clearance moved from unknown to amber at value 60.7000 (60–70% — normal balanced market) |
| 2026-04-26 | aud_usd | band_change | aud_usd moved from unknown to green at value 0.7153 (0.65–0.72 — normal range) |
| 2026-04-24 | all_ords | band_change | all_ords moved from unknown to green at value 9006.4004 (Above 8,000 — normal/strong) |
| 2026-04-24 | brent_oil | band_change | brent_oil moved from green to amber at value 99.1300 ($60–100 — normal range) |
| 2026-04-24 | iron_ore | band_change | iron_ore moved from unknown to green at value 107.1000 (Above $100 — China demand robust) |
| 2026-04-24 | us_yield_curve | band_change | us_yield_curve moved from green to red at value 0.6200 (Currently inverted) |
| 2026-04-23 | us_hy_credit_spreads | band_change | us_hy_credit_spreads moved from unknown to green at value 2.8600 (2-4% - normal range) |
| 2026-04-19 | au_vacancy_rate | band_change | au_vacancy_rate moved from amber to green at value 1.5000 (1.5-3% — healthy balanced range) |
| 2026-04-19 | qld_diesel_pump | band_change | qld_diesel_pump moved from unknown to crisis at value 296.9000 (270–300c/L — sharp household pain) |
| 2026-04-18 | us_jobless_claims | band_change | us_jobless_claims moved from crisis to green at value 210.7500 (200-280k - normal range) |
| 2026-04-12 | au_home_value_index_28d | band_change | au_home_value_index_28d moved from unknown to green at value 0.4000 (0.2 to 0.8% — normal growth band) |
| 2026-04-12 | au_total_listings | band_change | au_total_listings moved from unknown to green at value 75145.0000 (60-80k — current normal range) |
| 2026-04-12 | au_vacancy_rate | band_change | au_vacancy_rate moved from unknown to amber at value 1.5000 (1.5-3% — healthy balanced range) |
| 2026-04-01 | us_sahm_rule | band_change | us_sahm_rule moved from amber to green at value 0.1300 (Below 0.3 - below trigger threshold) |
| 2026-03-18 | rba_cash_rate | band_change | rba_cash_rate moved from unknown to green at value 4.1000 (3–4.5% — normal range; restrictive but not extreme) |
| 2026-03-01 | unemployment_rate | band_change | unemployment_rate moved from unknown to green at value 4.2552 (3.5–4.5% — near full-employment range) |
| 2026-03-01 | us_copper | band_change | us_copper moved from unknown to green at value 12528.7095 (Above $10000 - strong industrial demand) |
| 2026-03-01 | us_heavy_truck_sales | band_change | us_heavy_truck_sales moved from red to amber at value 367.0000 (350-400k - cooling) |
| 2026-03-01 | us_industrial_production_yoy | band_change | us_industrial_production_yoy moved from unknown to green at value 0.7417 (0-3% - normal expansion) |
| 2026-03-01 | us_sahm_rule | band_change | us_sahm_rule moved from unknown to amber at value 0.2000 (Below 0.3 - below trigger threshold) |
| 2026-03-01 | us_temp_employment_yoy | band_change | us_temp_employment_yoy moved from unknown to red at value -2.1279 (-5 to -2% - recession-grade layoffs) |
| 2026-03-01 | us_umich_sentiment | band_change | us_umich_sentiment moved from unknown to crisis at value 53.3000 (Below 60 - 2008/2022-grade pessimism) |
| 2026-03-01 | us_unemployment | band_change | us_unemployment moved from unknown to green at value 4.3000 (3.5-4.5% - near full-employment) |
| 2026-02-01 | us_m2_yoy | band_change | us_m2_yoy moved from unknown to green at value 4.8771 (3-10% - normal money growth) |