# Premia
Source: https://www.yieldcurve.pro/premia
Every Treasury yield is the sum of two things: what markets expect short-term rates to
average over the life of the bond and the extra yield investors demand as compensation
for committing capital for longer. That extra yield is the **term premium**. The
decomposition takes the form
$$y_n = \hat{y}_n + \xi_n$$
where
* $y_n$ — observed Treasury yield at tenor $n$
* $\hat{y}_n$ — risk-neutral yield (expected average short rate over the bond's life)
* $\xi_n$ — term premium
The term premium is time-varying and can turn negative. During the Fed's quantitative
easing programs (2008–2022), large-scale asset purchases suppressed term premia —
long-term yields fell not because rate cut expectations changed, but because the risk
compensation demanded by investors collapsed. Understanding which component is driving a
yield move is essential for reading monetary policy signals accurately.
This tool estimates term premia using the **ACM model** (Adrian, Crump, and Moench 2013),
the Federal Reserve Bank of New York's official decomposition methodology. It uses a
three-step linear regression on principal components of the yield curve to separate the
expectations and risk-compensation components at each maturity.
## Instructions
* use the **View** menu to switch between **Premia** and **Components**
* in **Premia** view: select one or more tenors (1Y–30Y) using the checkboxes and a date
range; the chart shows each selected tenor labeled as e.g. **10 Yr (premium)**
* in **Components** view: select a single tenor from the dropdown to see the full
decomposition — three lines labeled **10 Yr**, **10 Yr (expected)**, and **10 Yr (premium)**
* the chart updates automatically when you change dates, tenors, or view mode
* click the **sparkle** button to open **ChatYCP** and ask questions about the data (login required)
## Notes
* term premia are estimated using all available daily yield curve history; the model
re-fits after new data arrives and estimates for earlier dates may be revised slightly
(normal for time-series models of this type)
* negative term premia indicate that investors accept a lower yield than the expected
short-rate path — common when safe-haven demand or central bank buying compresses risk
compensation
* try asking ChatYCP: *"Current 10Y term premium?"* or *"How many days has 10Y TP been negative total?"*
* **our estimates will run approximately 1–2 percentage points higher than the NY Fed's
published ACM series.** the ACM model's VAR learns the unconditional mean short rate from
history. the NY Fed trains on data back to 1961 (including the Volcker-era rates of
5–15%); this site uses data from 2001-present, a window dominated by the zero-rate era
(2008–2021). as a result, our VAR estimates that rates mean-revert to roughly 2%,
attributing more of the current yield to "premium" and less to "expected short rates."
this is a known limitation of re-estimating the ACM model on a shorter history, not an
error in the implementation. the NY Fed's published series is linked below.
Use Cases
* **parsing yield moves**: when the 10Y yield rises, check whether the term premium or
the expectations component is driving it; a term-premium rise signals shifting risk
appetite, an expectations rise signals a more hawkish Fed repricing
* **monitoring QE/QT impact**: term premia compress during asset purchase programs and
widen during balance-sheet runoff; this page makes that dynamic visible in real time
* **cross-maturity risk structure**: compare 2Y vs. 10Y term premia to understand where
in the curve investors are demanding the most compensation
* **historical context**: place today's term premium in context — is it elevated relative
to the past decade, or near the post-QE floor?
Further Reading
* Adrian, Crump, Moench (2013) — Pricing the Term Structure with Linear Regressions
* NY Fed Term Premium Data and Methodology
* Forward Rates as Market Forecasts