Is the Yield Curve Inverted?
As of 2026-08-20, the 10-year minus 2-year Treasury spread is +50bp — not inverted. Below: today's full par curve, and every sustained inversion since 1990-01-02, computed from Treasury's own daily series rather than recalled.
Par yield curve, 2026-08-20
Maturities are spaced logarithmically — on a linear axis the eleven tenors under two years collapse into the left edge and the short end, which is the part that inverts, becomes unreadable.
| 1 Mo | 1.5 Month | 2 Mo | 3 Mo | 4 Mo | 6 Mo | 1 Yr | 2 Yr | 3 Yr | 5 Yr | 7 Yr | 10 Yr | 20 Yr | 30 Yr |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 3.80 | 3.77 | 3.79 | 3.87 | 3.88 | 3.94 | 3.99 | 4.19 | 4.26 | 4.39 | 4.53 | 4.69 | 5.20 | 5.23 |
10-year minus 2-year spread, 1990-01-02 to 2026-08-20
Shaded bands are the periods the spread spent below zero. Values are percentage points.
Every sustained inversion since 1990-01-02
9 episodes of the 10y−2y spread lasting at least 10 calendar days. Each row is computed from the daily series: the run of consecutive published observations with a negative spread, and the single deepest reading within it.
| Start | End | Calendar days | Trading days | Deepest | Deepest on |
|---|---|---|---|---|---|
| 2022-07-06 | 2024-08-26 | 783 | 537 | -108bp | 2023-07-03 |
| 2007-05-03 | 2007-05-21 | 19 | 13 | -6bp | 2007-05-09 |
| 2006-08-17 | 2007-03-20 | 216 | 147 | -19bp | 2006-11-15 |
| 2006-06-30 | 2006-07-26 | 27 | 18 | -7bp | 2006-07-11 |
| 2006-06-08 | 2006-06-28 | 21 | 15 | -6bp | 2006-06-14 |
| 2006-01-31 | 2006-03-07 | 36 | 25 | -16bp | 2006-02-23 |
| 2000-02-11 | 2000-12-26 | 320 | 220 | -52bp | 2000-04-07 |
| 1998-06-15 | 1998-07-09 | 25 | 18 | -7bp | 1998-06-25 |
| 1990-03-08 | 1990-03-29 | 22 | 16 | -14bp | 1990-03-20 |
A further 16 dips below zero lasted under 10 days, totalling 53 days. They are excluded from the table above and counted here instead: when the spread sits near zero it crosses repeatedly, and treating each crossing as a separate inversion inflates the count without adding information.
How an episode is defined
A maximal run of consecutive published observations with a negative spread. Weekends and holidays are not publication days and so do not break a run; a genuine return to a positive spread does.
Where either leg of a spread was not published on a date, that date has no spread rather than a carried-forward one.
This matters more than it sounds. Published inversion dates disagree with each other largely because of these two choices — whether a weekend counts as a break, and whether a one-day flicker back above zero ends an episode. Both choices are stated here and applied uniformly, so the numbers are reproducible from the source data by anyone who wants to check them.
What this does not tell you
The association between inversions and subsequent recessions rests on a handful of episodes — nine in this record — which is far too few to support a confident probability, however often the relationship is quoted. Lead times in past episodes have varied from months to more than two years, and an average taken over nine observations with that much spread carries almost no information about the next one. This page reports what the curve has done. It does not forecast, and nothing here is investment advice.
Frequently asked questions
Is the yield curve inverted right now?+
As of 2026-08-20, the 10-year minus 2-year spread is +50bp, so it is not inverted. The 10-year minus 3-month spread is +82bp, also not inverted. Both are computed from the Treasury par yields published for that date.
What does an inverted yield curve mean?+
It means investors accept a lower yield to lend for ten years than for two, which is the reverse of the usual compensation for tying money up longer. The common reading is that the market expects short-term rates to fall, which usually happens when growth is expected to weaken. It is a description of market prices, not a forecast, and the association with recessions is a historical statistical relationship over a modest number of episodes.
How many times has the curve inverted?+
Since 1990-01-02, the 10-year minus 2-year spread has been negative on 1,052 of 9,165 published business days, across 9 sustained episodes lasting at least 10 calendar days. There were a further 16 briefer dips totalling 53 days, listed separately because the spread crosses zero repeatedly when it sits near the line and counting each crossing separately inflates the total.
What was the longest inversion?+
2022-07-06 to 2024-08-26 — 783 calendar days (537 trading days), reaching -108bp at its deepest on 2023-07-03. The deepest of any episode in this record was -108bp on 2023-07-03.
Which spread should I look at, 10y-2y or 10y-3m?+
Both are shown because they do not always agree, and the disagreement is informative. The 10y-3m spread is the one used in most of the academic recession-prediction literature (Estrella and Mishkin), while 10y-2y is the one financial media quote most often. The 3-month leg tracks policy rates closely, so 10y-3m tends to invert later and more sharply.
Where does this data come from?+
Directly from the US Treasury's daily par yield curve rates — 9,165 published business days from 1990-01-02 to 2026-08-20. These are par yields, not zero-coupon yields. Treasury data is a US government work in the public domain, so it can be republished here freely.
Get the data
The full artefact — 9,165 daily observations, both spread series, and every episode with its provenance — is served as static JSON.
Download treasury-curve.jsonGenerated 2026-08-20T23:37:57+00:00. Source: US Department of the Treasury — Daily Treasury Par Yield Curve Rates. Par yields, not zero-coupon. Treasury publishes on business days only. US government work, public domain (17 U.S.C. 105).
Related tools
Source: US Department of the Treasury, Daily Treasury Par Yield Curve Rates. This page is not affiliated with or endorsed by the Treasury. Nothing here is investment advice or a forecast of interest rates, growth, or recession.