US Inflation Breakeven Term Structure: The 5Y/10Y Curve, 2003–2026

Chart: Five-Year and Ten-Year Breakeven Inflation, 2003–2026

US Treasury breakeven inflation: 5-year and 10-year, monthly 2003–2026

The two breakeven series tracked each other within roughly 50 basis points for two decades. Since 2021, the 5-year has moved decisively above the 10-year and has stayed there.

Key Takeaway

The spread band in the lower panel is almost entirely below zero from 2010 to early 2021, then mostly above zero from April 2021 onwards. The November 2008 negative outlier (−2.04pp) reflects TIPS market liquidity dysfunction during the Lehman aftermath, not a regime-comparable observation.

Sources: Federal Reserve Bank of St. Louis (T5YIEM, T10YIEM). Chart: Eco3min Research.

PNG
SVG
PDF
Free to use with attribution · Cite

How to Read This Chart

The upper panel plots the 5-year (red) and 10-year (navy) breakeven inflation rates as reported by FRED, monthly closing values. The horizontal dashed line marks the 2% target the Federal Reserve formally adopted in January 2012. Background shading divides the sample into six analytical phases: Pre-GFC (2003-01 to 2008-08), GFC Dysfunction (2008-09 to 2009-03, a brief episode of TIPS market illiquidity discussed below), Recovery/QE (2009-04 to 2015-11), Liftoff (2015-12 to 2020-02), COVID Shock (2020-03 to 2021-03), and Post-COVID (2021-04 onwards).

The lower panel plots the spread between the two — T5YIE minus T10YIE. When the spread is below zero (navy fill), the term structure is in its conventional configuration: investors price more inflation in the distant future than in the near future. When the spread is above zero (red fill), the configuration has reversed: near-term inflation is priced higher than long-term inflation. The visual pattern is the page’s central empirical finding. For methodology, including the regime classification thresholds and the treatment of the GFC dysfunction window, see the Methodology section.

The Decade Without an Inversion

The standard description of US inflation expectations in the 2010s, repeated across central bank communications and market commentary, runs something like this: long-term expectations remained anchored near 2%, while short-term expectations responded to oil-price shocks, occasional growth surprises, and the gradual recovery from the financial crisis. In this telling, the breakeven curve is upward-sloping in the long run because investors demand compensation for inflation risk over longer horizons, and that risk premium translates into a higher 10-year breakeven than 5-year breakeven.

The data confirms the upward slope for the full decade following the GFC dysfunction. From January 2010 through December 2019, the 10-year breakeven exceeded the 5-year breakeven in every single one of 120 monthly observations. The mean spread (T5YIE minus T10YIE) was −0.26 percentage points; the median was −0.27pp. Not a single month flipped. Extending the window to include the immediate post-dysfunction recovery (from April 2009) and the pre-COVID Liftoff cycle through February 2020 yields the same result: 129 of 129 months with T10YIE above T5YIE.

Since April 2021, the configuration has reversed. The 5-year has exceeded the 10-year in 45 of 61 months — a frequency of 73.8% — with an additional one month at zero. The peak inversion of +0.56pp was reached in March 2022, the same month T5YIE itself peaked at 3.41% (the highest single monthly value in the FRED T5YIE series since its inception in January 2003). The mean spread across the Post-COVID phase is +0.10pp, compared to −0.26pp during 2010–2019: a differential of 0.35pp.

What Each Phase Looked Like

The Phase classification used throughout this page partitions the sample into six analytical windows defined by monetary policy, market structure, and macro regime. The frequency of term-structure reversal varies sharply across phases, providing a comparative reference for the post-2021 configuration.

Phase-by-Phase Term Structure Direction

Three observations stand out. First, the Pre-GFC phase shows that breakeven term structure inversion is not a new phenomenon: 24 of 68 months saw the 5-year exceed the 10-year, typically during brief oil-price episodes. What is new is the persistence — no prior 5-year window contains anything resembling a 74% inversion frequency. Second, the Recovery/QE and Liftoff phases are striking in their uniformity: across 131 combined months from April 2009 to February 2020, the 5-year never once exceeded the 10-year. Third, the GFC Dysfunction window is a separate analytical category — the −0.83pp mean spread reflects the collapse of T5YIE to −1.40% in November 2008, an episode that mechanically dominates volatility and distributional statistics if not isolated.

Volatility by Phase

Beyond the direction of the curve, the volatility of breakevens also varies across phases. Annualized 12-month rolling standard deviation of monthly changes provides a comparable measure across periods.

Across every phase except the GFC Dysfunction window, the 5-year breakeven is between 1.2 and 1.4 times as volatile as the 10-year. This is the structural reason why temporary shocks tend to widen the spread visibly: the shorter-dated series moves more on the same news. In the Post-COVID phase, the absolute level of 5-year volatility (0.49) is similar to the Recovery/QE phase (0.65) and roughly half the COVID Shock level (1.00), so the inversion is not driven by elevated turbulence — the curve has tilted, not destabilized. For broader context on inflation dynamics across these phases, see our complete CPI inflation history study and the consumer sentiment dataset.

A Legitimate Reading: Anchor Plus Path

A reading of the post-2021 configuration that does not require any departure from standard inflation-targeting theory runs as follows. Markets observed a large near-term inflation shock starting in 2021, reflecting energy prices, supply-chain disruption, and demand patterns from pandemic-era fiscal transfers. Five-year inflation pricing rose accordingly, because the next 5 years contain that shock. Ten-year inflation pricing rose by less, because averaging the shock over a longer window dilutes its effect. The result is a temporary inversion of the term structure that resolves slowly as the shock period rolls out of the 5-year window.

This interpretation has three things going for it. First, the median 10-year breakeven across the Post-COVID phase is 2.33% — approximately 33 basis points above the Fed’s 2% target, which is within the band typically described as anchored in central bank communications, and substantially below the peak of 2.85% reached in March 2022. Second, the spread has been compressing: from +0.56pp in March 2022 down to +0.22pp in April 2026. The configuration is moving toward, not further from, its prior state. Third, this exact mechanism — front-loaded near-term inflation pricing combined with a stable long-run anchor — is the outcome an inflation-targeting regime is supposed to produce in response to a transitory shock. Volcker, had he observed it in 1980, would have considered it evidence that monetary policy was working.

The empirical record does not falsify this reading. What it does, however, is establish the persistence of the configuration as unusual. The post-COVID inversion has now lasted longer than any documented inflation shock period in the FRED breakeven sample. A shock that takes 5+ years to roll out of a 5-year window is, by construction, no longer a short-run phenomenon. Whether this matters for monetary policy or merely reflects path uncertainty is a question the breakeven data alone cannot answer; it would require triangulation against survey-based long-term expectations dispersion, professional forecaster forecasts, and direct measures of inflation uncertainty (for example, the NY Fed’s SCE measures of forecast dispersion).

What Happened Next? Realized Inflation by Breakeven Regime

Standard forward-return tables on Eco3min compare a market signal at observation date to subsequent equity returns. For breakeven inflation, the more direct comparison is to the realized inflation that subsequently occurred. The table below classifies each observation by the term-structure regime at that date and reports the median realized annualized CPI growth over the following 5 years. The sample excludes the seven months of GFC dysfunction (September 2008 – March 2009) and includes only observations through January 2021, the last date for which a complete 5-year forward window of CPI data is available.

Methodological note: the regime classification at observation date uses a ±10 basis point threshold around zero spread, identifying “Flat” observations where the two breakevens are functionally tied. The 5-year forward CPI growth is computed as the geometric annualized rate: (CPI[t+60mo] / CPI[t])^(1/5) − 1. Sample size for the reversed regime (n=8 with available forward window) is modest; these observations are concentrated in the brief Pre-GFC episodes (e.g. 2005, when oil prices were rising).

Past distributions are not predictive of future outcomes. Regime-conditional statistics describe historical patterns, not expected returns or expected inflation.

Regime Classification: Twenty-Three Years of Term Structure Direction

Five-year minus ten-year breakeven spread, by month (2003–2026)

Two decades of normal upward curvature visible as cool blue cells; the 2021–2026 reversal visible as warm orange cells. The clean break is the page’s central empirical finding.

Note: November 2008 cell (−2.04pp) reflects TIPS market liquidity dysfunction; color scale capped at ±0.7pp for readability. Sources: FRED (T5YIEM, T10YIEM). Chart: Eco3min Research.

PNG
SVG
PDF
Free to use with attribution · Cite

Regime Interpretation Grid

Did Breakevens Predict Realized Inflation?

5-year breakeven (T5YIE) at observation date vs realized annualized CPI growth over next 5 years

Points above the 45° line indicate the breakeven underestimated subsequent realized inflation. The COVID-era cluster sits high above the line: breakeven priced near-zero, inflation came at 4.4-4.6% annualized.

Source: FRED (T5YIEM, CPIAUCSL). Chart: Eco3min Research.

The chart provides a direct test of whether breakevens have been informative about future realized inflation. For each observation date from January 2003 through January 2021 (the last date where a complete 5-year forward window is available), the 5-year breakeven is plotted against the geometric annualized CPI growth that subsequently occurred. The 45° line represents perfect prediction. The dispersion around the line is large but not random: COVID-era breakevens (priced in early 2020 at 0.7-1.0%) sit roughly 3.8 percentage points below what subsequently realized; pre-GFC observations from mid-2008 (priced around 2.5%) sit somewhat above realization because the GFC suppressed inflation in 2009-2010.

Two implications follow. First, the largest single forecasting failure in the FRED breakeven series is the COVID-era cluster — not the post-2022 inversion. At the moment when long-term anchoring was most visibly preserved (breakevens crashed to 0.7%), the subsequent realized inflation diverged from breakeven pricing by the largest margin observed in 18 years. Second, the post-COVID period — which contains the recent inversion regime — does not yet have a completed 5-year forward window, so this analysis cannot judge whether the current inverted breakeven structure is “correctly” priced relative to inflation that has yet to occur.

Historical Turning Points

January 2003 — T5YIE Series Begins

The FRED T5YIE series, derived from 5-year nominal and TIPS yields published by the US Treasury, begins on January 2, 2003. The first monthly observation reads 1.40% for T5YIE and 1.76% for T10YIE, a spread of −0.36 percentage points. The realized annualized CPI growth from January 2003 to January 2008 was subsequently 3.05% — a forecast miss of approximately −1.6pp, indicating that breakevens substantially underestimated the 2003–2008 inflation episode driven by oil prices and housing.

November 2008 — Peak GFC Dysfunction

The 5-year breakeven crashed to −1.40% in November 2008, the lowest single monthly value in the entire FRED T5YIE series. The 10-year held at 0.64%, producing a spread of −2.04pp — the most extreme reading in either direction across 280 observations. The mechanical cause was not deflation expectations but TIPS market illiquidity: dealer balance sheets contracted, the natural buyer base for inflation-linked paper retreated, and TIPS yields rose sharply against nominal yields. This episode is treated as a separate analytical category throughout the page; its values are not comparable to other regime data.

March 2020 — COVID Breakeven Collapse

In March 2020, T5YIE fell to 0.67% and T10YIE to 0.99% as markets repriced for a deflationary recession. Both values were the lowest since the GFC dysfunction window. The subsequent 5-year realized annualized CPI growth from March 2020 to March 2025 was 4.38% — meaning the breakeven at that date underestimated future inflation by 3.71 percentage points annualized, the third-largest miss in the series. The May 2020 observation (T5YIE 0.77%, realized 4.62%) holds the record for largest miss at −3.85pp.

March 2022 — Peak T5YIE Post-COVID

T5YIE reached 3.41% in March 2022, the highest single monthly value in the FRED T5YIE series since its inception in January 2003. T10YIE was at 2.85%, producing the maximum spread reversal in the entire series: +0.56pp. CPI year-over-year growth at that date stood at 8.57%, near the cycle peak. The current Post-COVID phase began roughly a year earlier (April 2021) and has continued through the publication date with persistent positive spread.

April 2026 — Current Observation

The latest monthly observation places T5YIE at 2.60% and T10YIE at 2.38%, a spread of +0.22pp. The 5-year breakeven sits 60 basis points above the 2% Fed target, slightly more than the median 10-year breakeven during the Post-COVID phase (2.33%, 33bp above target). The post-COVID inversion regime has persisted for 61 consecutive months, with 45 of those months showing T5YIE above T10YIE. The trailing 6-month average spread has stayed continuously above zero since February 2025 (15 consecutive months).

Methodology

This dataset combines the FRED-published monthly breakeven inflation series (T5YIEM and T10YIEM) with the seasonally-adjusted CPI for All Urban Consumers (CPIAUCSL), derived columns computed in Python from the merged data, and a manually-defined phase taxonomy reflecting documented macro regimes. All numbers in the HTML page are derived from the published CSV.

Core Formula

Filter Definitions

Regime Selection Criteria

Sensitivity test on the reversal threshold: setting the band to ±0.05pp produces 55 reversed, 36 flat, 189 normal-upward observations; setting it to ±0.15pp produces 21 reversed, 135 flat, 124 normal. The core empirical finding — the absence of any T5YIE > T10YIE observation in 2010–2019 — is robust to all threshold choices because the maximum spread during that decade was −0.03pp (June 2016), well below any reasonable noise threshold. Setting the threshold to zero produces the same headline: 0 of 120 months in 2010–2019 show 5Y > 10Y.

Sensitivity test on the “2010–2019” window choice: substituting the window 2009-04 to 2019-12 (the immediate post-dysfunction recovery extended through the Liftoff period) yields 129 of 129 months with T10YIE > T5YIE. Substituting 2011-01 to 2019-12 yields 108 of 108. Substituting 2010-01 to 2020-02 (pre-COVID) yields 122 of 122. The absence-of-inversion claim is invariant to reasonable boundary shifts.

Dataset Design

Python Reproduction Code

# Reproduce the core dataset from FRED
import pandas as pd
import numpy as np

# Fetch from FRED (assumes pandas_datareader or direct download)
t5  = pd.read_csv("T5YIEM.csv",  parse_dates=["date"])
t10 = pd.read_csv("T10YIEM.csv", parse_dates=["date"])
cpi = pd.read_csv("CPIAUCSL.csv", parse_dates=["date"])

df = t5.merge(t10, on="date").merge(cpi, on="date", how="left")
df = df.sort_values("date").reset_index(drop=True)

# Derived columns
df["spread_5_10"] = df["t5yie"] - df["t10yie"]
df["cpi_yoy"] = df["cpi"].pct_change(periods=12) * 100
df["cpi_fwd_5y_realized"] = ((df["cpi"].shift(-60) / df["cpi"]) ** (1/5) - 1) * 100
df["error_5y"] = df["t5yie"] - df["cpi_fwd_5y_realized"]

# Headline finding
decade = df[(df["date"] >= "2010-01-01") & (df["date"] <= "2019-12-01")]
n_inversions = (decade["spread_5_10"] > 0).sum()
print(f"2010-2019 inversions: {n_inversions} of {len(decade)}")
# Output: 2010-2019 inversions: 0 of 120
  

Limitations on Breakeven Interpretation

This dataset documents the market price of inflation insurance, computed mechanically from Treasury and TIPS yields. It is not equivalent to consumer or survey-based inflation expectations. For triangulation, the Michigan inflation expectations series (FRED: MICH, 1-year; expectations panel 5-10 year horizon), the New York Fed Survey of Consumer Expectations (available from 2013), and the Philadelphia Fed Survey of Professional Forecasters provide complementary measures. Decomposing breakevens into expected inflation, inflation risk premium, and liquidity premium requires modelling assumptions; the Adrian-Crump-Moench model published by the NY Fed is one widely-cited reference framework.

Dataset Download & Reproducibility

280 monthly observations · January 2003 – April 2026 · Licensed under CC BY 4.0.

Data Sources & References

• Primary Federal Reserve Bank of St. Louis, 5-Year Breakeven Inflation Rate (T5YIEM), monthly, retrieved from FRED. Data derived from 5-Year Treasury Constant Maturity and 5-Year TIPS Constant Maturity yields published by the US Treasury Department.
• Primary Federal Reserve Bank of St. Louis, 10-Year Breakeven Inflation Rate (T10YIEM), monthly, retrieved from FRED. Data derived from 10-Year Treasury Constant Maturity and 10-Year TIPS Constant Maturity yields.
• Primary US Bureau of Labor Statistics, Consumer Price Index for All Urban Consumers: All Items (CPIAUCSL), monthly seasonally adjusted, retrieved from FRED.
• Research Adrian, T., Crump, R. K., & Moench, E. (2013). “Pricing the Term Structure with Linear Regressions.” Journal of Financial Economics, 110(1), 110–138.
• Research D’Amico, S., Kim, D. H., & Wei, M. (2018). “Tips from TIPS: The Informational Content of Treasury Inflation-Protected Security Prices.” Journal of Financial and Quantitative Analysis, 53(1), 395–436.
• Research Gürkaynak, R. S., Sack, B., & Wright, J. H. (2010). “The TIPS Yield Curve and Inflation Compensation.” American Economic Journal: Macroeconomics, 2(1), 70–92.
• Reference Federal Reserve Bank of New York, Survey of Consumer Expectations (SCE), available from 2013, quarterly reports on 1-year and 3-year inflation expectations.
• Reference University of Michigan, Surveys of Consumers, Inflation Expectations (FRED series MICH).

Methodological Limitations

• Breakevens are a composite of expected inflation, inflation risk premium, and TIPS liquidity premium. The page does not decompose these components; readers ought not treat breakevens as a clean measure of expectations alone.
• The TIPS market matured during the early portion of the sample. Direct comparison across 2003-2007 and 2015-2025 must acknowledge institutional evolution; pre-2003 breakeven data is not available in FRED for the 5-year series.
• The September 2008 – March 2009 dysfunction window is treated as a separate analytical category. Including it in distributional statistics distorts every measure of central tendency; excluding it removes seven months from a 280-month sample.
• Forward CPI realization columns are mechanically limited by data availability: cpi_fwd_5y_realized has data through January 2021 (last date with complete 60-month forward window); cpi_fwd_10y_realized through January 2016.
• Forward-window observations overlap by construction (consecutive monthly windows share 59 months of CPI data). Standard hypothesis testing on the forecast-error distribution would require adjustment for this overlap; the page does not perform formal inference.
• The phase taxonomy reflects analyst judgment about macro-regime boundaries, not algorithmic classification. Different boundary choices would shift specific summary statistics but do not affect the core empirical finding of zero inversions in 2010–2019.

Frequently Asked Questions

Source

Related Eco3min Research

• US Inflation Is Not Linear (1914–2026) — Complete CPI history with regime decomposition.
• Yield Curve Inversion History (2s10s) — The other term-structure inversion that markets watch.
• Real Interest Rates History (1962–2026) — DGS10 minus realized CPI, the companion deflated yield.
• Federal Funds Rate Track Record — 70 years of Fed rate decisions audited against inflation realizations.
• 5-Year Breakeven Inflation Dataset — The raw T5YIE dataset page.
• 10-Year Breakeven Inflation Dataset — The raw T10YIE dataset page.