TIPS breakevens track inflation only loosely — surveys and premium-adjusted models usually forecast it better at one to two years
Asked:
“Do TIPs breakevens accurately forecast inflation?”
Seven selected Federal Reserve and BLS studies or summaries, spanning samples from 1986 through 2021, evaluate how well market inflation compensation — nominal Treasury yield minus TIPS real yield — forecasts U.S. CPI inflation. Sample periods and horizons differ, so this page is a synthesis, not a pooled meta-analysis.
The verdict spectrum: where each study lands
raw breakeven evidence
premium- or liquidity-adjusted evidence
supporting (market-based, not direct TIPS)
Why a breakeven is not a forecast
Breakeven rate
Nominal Treasury yield minus TIPS real yield of matching maturity — better described as market inflation compensation.
= expected CPI inflation
The piece a forecaster wants: the market's average expected inflation over the maturity.
+ inflation-risk premium
Extra yield investors demand for bearing uncertain inflation; varies over time.
−/+ liquidity & technical effects
TIPS trade less liquidly than nominals; the sign and size depend on convention and model, and can swing sharply in crises.
Three different claims, three different tests
Unbiasedness
Errors average out to about zero. The St. Louis Fed found no significant average bias in five-year breakevens — but unbiased can still mean large errors either way.
Directional usefulness
Moves broadly with realized inflation. The BLS found breakevens reasonably approximated CPI-U over 2003–2018, more precisely at long maturities than short ones.
RMSE accuracy
Small point-forecast errors. Here raw breakevens often trail surveys and simple rules at one to two years; premium-adjusted models close much of the gap.
What the visual cannot say alone
Stripping estimated risk and liquidity premiums improved forecasts by 32% at 12 months to over 100% at three years in-sample, and cut out-of-sample RMSE by 4%–11% versus raw breakevens over 1999–2014 —
newyorkfed.orgAn earlier Fed comparison put a TIPS-based measure at RMSE 1.07 versus 0.92 for the SPF survey — suggestive rather than definitive, since it matched a five-year breakeven against a one-year horizon —
federalreserve.govFive-year TIPS breakevens showed no statistically significant average bias, while inflation-swap breakevens significantly overestimated inflation — lack of bias is not the same as small errors —
stlouisfed.orgAdding a TIPS liquidity factor cut one-year-ahead RMSE from 162.65 to 141.69 basis points over 2005–2013, making the market model competitive with Blue Chip at 150.93 —
frbsf.org
Practical takeaway
- Use TIPS breakevens as one real-time market signal, not a stand-alone point forecast.
- Compare like-for-like CPI horizons — a five-year breakeven is not a one-year forecast.
- Prefer zero-coupon or constant-maturity breakeven measures.
- Adjust or cross-check for liquidity and inflation-risk premiums; adjusted measures forecast better.
- Triangulate with SPF, Blue Chip, and model forecasts before acting.
- Distrust crisis-time breakeven moves at face value — premiums, not expectations, often drive them.
The evidence, study by study
| Study | Horizon | Sample | Finding | Comparative result | Caveat |
|---|
Method: evidence table of 7 selected Federal Reserve and BLS studies or summaries on U.S. TIPS breakevens and market-based inflation forecasts, assembled 2024. Each row records forecast horizon, sample period, headline finding, comparative accuracy result (RMSE where reported, in percent or basis points), and a methodological caveat. Spectrum positions are a qualitative reading of each study's verdict, not a pooled statistic; long quantitative passages are trimmed for space. Samples and horizons differ across studies, so results are synthesized, not pooled.