Prolepsia forecastNumber

What will the US 10-year Treasury yield be on 31 December 2027, as published by the Federal Reserve (FRED series DGS10)?

As of 17 November 2026, Prolepsia’s median forecast is 5.01, with an 80% chance it falls between 3.68 and 6.24.

5.01median

Prolepsia’s forecast distributionHow likely each value is for: What will the US 10-year Treasury yield be on 31 December 2027, as published by the Federal Reserve (FRED series DGS10)?. Median: 5.01. 80% range: 3.68 to 6.24.
Prolepsia forecast made

How Prolepsia reasons

The report Prolepsia wrote with this forecast.

How forecasts are made · Track record

TL;DR

For December 31, 2027, the forecast is a median (50th percentile) of 5.01%, a 10th percentile of 3.68%, and a 90th percentile of 6.235%. The central case is continued elevated long-term yields, not a return to the low-rate environment of the previous decade. These percentiles describe the single daily DGS10 observation, conditional on valid publication.

Key evidence

  • The starting point is already elevated. The latest official observation available for the November 17, 2026 forecast was 5.15% on November 13, updated November 16. That is the relevant published anchor, rather than a subsequent intraday market quote (FRED).
  • High nominal yields are not just an inflation story. The same-date real 10-year yield was 2.91%, leaving a nominal-minus-real spread of 2.24 percentage points. Between November 9 and November 13, real yields accounted for 25 of the nominal yield’s 27-basis-point increase. Disinflation alone therefore does not guarantee a large decline in nominal yields (H.15).
  • Policy and employment resist a rapid-easing narrative. The October 28 FOMC statement maintained the 3.75%–4.00% policy range, with two dissents favoring an increase (Federal Reserve). October payrolls rose by 165,000 and unemployment was 4.1% (BLS).
  • History supports substantial uncertainty, not automatic mean reversion. Historical DGS10 changes over comparable horizons commonly reach a percentage point. Tests of reversion toward past yield levels did not reliably beat a no-change benchmark. The historical record therefore supports staying near the current level while allowing meaningful moves in either direction (FRED history).
  • Fiscal pressure is real but not an immediate issuance shock. Treasury anticipated unchanged nominal coupon auction sizes for the next several quarters, tempering the argument that financing needs must immediately force yields higher (Treasury).

Uncertainties

The main downside risk is a downturn that lowers real-rate expectations and triggers monetary easing. Stronger productivity and easing energy pressure also support disinflation. The upside risks are persistent inflation, further tightening, weaker Treasury demand and greater compensation for holding long-duration debt.

The division between expected future rates and term premium remains uncertain. Historical samples contain relatively few independent episodes, and a single year-end observation adds market noise that quarterly or annual averages smooth away. If no authorized observation exists for the specified date by the resolution deadline, the question is annulled rather than resolved using another trading day.

Read the full report

TL;DR

The forecast for December 31, 2027 is a median (50th percentile) of 5.01%, a 10th percentile of 3.68%, and a 90th percentile of 6.235%. Use continued elevated long-term yields as the central planning case, with substantial exposure to moves in both directions. The decisive point is that high real yields support the current nominal yield, so lower inflation alone does not imply a return to low Treasury rates.

Context

This forecast is dated November 17, 2026 and targets the single daily observation dated December 31, 2027. DGS10 is the nominal, annualized yield on a hypothetical Treasury security with exactly 10 years remaining to maturity. It is not an annual average, an individual bond’s yield or the federal funds rate. The series is daily, not seasonally adjusted, and drawn from the Federal Reserve’s H.15 release (FRED definition).

The latest published observation was 5.15% for November 13, 2026, updated November 16. That is the official starting point, even though newer market quotes were available (FRED). The forecast horizon is 409 calendar days from the forecast date. The economic question is whether elevated real yields and persistent inflation endure, weaken with a downturn, or rise further as investors demand more compensation for holding Treasury debt.

Evidence

The historical backbone favors a restrained directional view. Across 35 annual changes between year-end 1990 and year-end 2025, the standard deviation of DGS10 changes was 1.0811 percentage points. Absolute changes of at least one percentage point occurred in 12 of those 35 years; changes of at least two points occurred in three. These are calculations from historical daily endpoints, not annual-average yields (DGS10 history). They show why a forecast close to today’s yield still needs a wide range of outcomes.

Comparable-horizon evidence gives the same broad message. Rolling 409-day changes had a standard deviation of 0.932 percentage points for starts from 1990 onward and 0.860 for starts from 2000 onward. Restricting starting yields to 3.5%–6% reduced those figures to 0.833 and 0.648, respectively. The unrestricted history beginning in 1962 was wider, at 1.221 percentage points (historical series). The narrower recent samples are useful comparisons, but they contain fewer distinct episodes and lean heavily on a particular monetary regime.

The historical evidence also challenges automatic mean reversion. Level-based reversion lost to no change in all eight out-of-sample comparisons reported in the research. Reversion toward trailing historical averages also failed to improve reliably on no change. These are historical test results, not a law of interest rates. Their practical lesson is narrower: a yield that looks high against the recent past is not, by itself, evidence that it must fall.

The available dated DGS10 landmarks show the changing starting environment. These are daily observations, not a continuous history or monthly averages (FRED DGS10).

Observation date Yield
February 27, 2026 3.97%
September 10, 2026 4.95%
October 5, 2026 5.31%
November 5, 2026 4.84%
November 9, 2026 4.88%
November 12, 2026 5.10%
November 13, 2026 5.15%

The composition of the latest increase matters more than the headline alone. From November 9 to November 13, the nominal yield rose from 4.88% to 5.15%, while the inflation-indexed yield rose from 2.66% to 2.91%. The nominal-minus-real spread moved only from 2.22 to 2.24 percentage points (H.15). Thus, 25 of the 27 basis points of nominal increase came from the real-yield component.

The recent rise was predominantly a real-yield move, not a surge in inflation compensation. This does not identify how much came from expected real short-term rates versus risk premiums. Nor is the nominal-minus-real spread a pure inflation forecast: liquidity and risk compensation affect it. But the evidence rules out attributing the entire elevated nominal yield to inflation expectations.

Current inflation and employment provide little support for making rapid easing the central case. Headline CPI rose 0.6% in September and 0.4% in October, month over month. Core CPI rose 0.3% in each of August, September and October. These figures are seasonally adjusted. The usual October annual comparisons were unavailable because relevant October 2025 observations were missing (BLS CPI release). September PCE inflation was 3.6% year over year, with core PCE at 3.1%; the respective monthly increases were 0.5% and 0.3% (BEA).

October payroll growth of 165,000 and unemployment of 4.1% point to continued labor-market resilience. There was also a weaker detail: involuntary part-time employment increased by 339,000 to 4.8 million. A single household-survey movement does not establish a downturn, but it supports keeping a meaningful downside scenario (BLS employment report).

Productivity supplies another counterweight to inflation pressure. Preliminary third-quarter nonfarm-business productivity rose at a 3.6% annualized quarterly rate. Unit labor costs fell 0.7% on the same basis and rose 2.4% year over year (BLS productivity release). These figures support a less inflationary growth path. Yet stronger productivity can also sustain higher real interest rates. It is therefore not an unambiguous signal for lower nominal yields.

Policy remains restrictive rather than clearly poised for rapid easing. On October 28, the FOMC held the target range at 3.75%–4.00%, with two dissenters preferring an increase (FOMC statement). September projections put median PCE inflation at 3.7% for 2026 and 2.3% for 2027. They put the median federal funds rate at 4.1% at both year-ends, 3.9% in 2028 and 3.2% over the longer run (September projections). Those are conditional policy projections, not commitments or direct forecasts of DGS10.

Fiscal financing supports the upside risk, but the issuance details temper the strongest version of that argument. Fiscal-year 2026 receipts of $5.403 trillion and outlays of $7.396 trillion imply a $1.993 trillion deficit (Treasury refunding presentation). Treasury projected privately held net marketable borrowing of $618 billion in October–December 2026 and $800 billion in January–March 2027, conditional on its cash assumptions (borrowing estimates). These are substantial financing needs, but not direct measures of additional long-duration supply.

Treasury also anticipated unchanged nominal coupon auction sizes for the next several quarters and up to $70 billion of quarterly liquidity-support buybacks (refunding statement). Buybacks are not quantitative easing or an equivalent reduction in net borrowing. Recent auction evidence showed softer demand, but changing participation between indirect and direct bidders did not establish a wholesale buyer withdrawal. The inference is persistent financing risk, not a confirmed immediate supply shock.

What's non-obvious

The obvious narrative is that inflation will eventually cool, so long yields should fall sharply. It misses two points. First, the recent yield rise was concentrated in real yields. Second, disinflation can coexist with strong productivity and relatively high real rates. Lower inflation therefore supports the downside without determining it. A large sustained rally needs more than a benign inflation release: weaker growth, lower expected real rates or reduced risk compensation would strengthen that case.

The other trap is treating calm recent markets or a large count of rolling historical observations as proof of precision. Many rolling windows cover the same episodes. The research identified only about 33 independent 409-day windows since 1990, with still fewer at comparable starting yields. Recent consensus commentary generally favored some normalization, but some forecasts concerned quarterly or annual averages rather than the target day. Neither a smoother target nor repeated observations of the same historical shock justify confidence that the daily endpoint will stay close to the median.

Uncertainties

The central unresolved issue is the persistence of high real yields. A downturn, weaker investment or a reversal in the AI investment cycle would support lower yields. Persistent inflation, stronger real growth or deteriorating Treasury demand would support higher yields. Energy prices and the Iran conflict add uncertainty in both directions: easing disruption would help disinflation, while renewed pressure would complicate monetary policy.

The current term premium is not pinned down by a consistently verified estimate. Estimates discussed in the research differ by method and date. A matched series separating expected future short rates from duration risk compensation would clarify whether elevated yields chiefly reflect the economic outlook or the price investors demand for bearing risk. More auction observations would also help distinguish a temporary weak result from sustained demand deterioration.

Inflation measurement has a specific gap in the missing October annual CPI comparisons. Productivity figures are preliminary, and the labor-market warning comes from a single month. Subsequent inflation, employment and productivity releases would help distinguish orderly disinflation from weakening demand. Historical tail evidence remains limited because large moves cluster in a small number of episodes.

Finally, this is a forecast conditional on valid publication. The target is December 31, 2027, not a nearby trading day. Under the resolution rules, FRED takes precedence; an explicitly dated, comparable H.15 observation can serve as the fallback if necessary by January 30, 2028. If neither authorized source supplies a valid observation by that deadline, the question is annulled. The percentile precision does not imply equivalent precision in the underlying economic outlook.

Forecast history

How Prolepsia’s forecast has movedOne forecast so far: 5.01 on 17 Nov 2026.
DateForecast
Median 5.01, 80% range 3.68 to 6.24

Sources Prolepsia read

  1. Market Yield on U.S. Treasury Securities at 10-Year Constant Maturity, Quoted on an Investment Basis (DGS10) | FRED | St. Louis Fedfred.stlouisfed.org
  2. Federal Reserve Board - H.15 - Selected Interest Rates (Daily) - November 16, 2026www.federalreserve.gov
  3. Consumer Price Index Summary - 2026 M10 Resultswww.bls.gov
  4. Personal Income and Outlays, September 2026 | U.S. Bureau of Economic Analysis (BEA)www.bea.gov
  5. # Highlights of Treasury’s November 2026 Quarterly Refunding Presentation to the Treasury Borrowing Advisory Committee (TBAC)home.treasury.gov
  6. Treasury Term Premia - FEDERAL RESERVE BANK of NEW YORKwww.newyorkfed.org
  7. PDFwww.treasurydirect.gov
  8. November 16, 2026: 10-Year U.S. Treasury 15 min · 8:45–9:00 AM EST Prediction Marketrobinhood.com
  9. 10-year Treasury yield moves to 5.17% as traders weigh producer price reportwww.cnbc.com
  10. 10-Year Treasury Yield Hits 5.15%, Squeezing Stocks and Mortgages247wallst.com

The question

This question asks for the yield on U.S. Treasury securities at a constant maturity of 10 years on December 31, 2027, as reported in the Federal Reserve Bank of St. Louis's Federal Reserve Economic Data (FRED) series DGS10, officially titled 'Market Yield on U.S. Treasury Securities at 10-Year Constant Maturity, Quoted on an Investment Basis.' The forecast is made as of November 17, 2026, and concerns the daily observation dated December 31, 2027. DGS10 is expressed as an annualized percentage yield, is not seasonally adjusted, and is sourced from the Board of Governors of the Federal Reserve System's H.15 Selected Interest Rates release. It represents the market yield associated with a hypothetical Treasury security having exactly 10 years remaining to maturity, rather than the yield on any particular outstanding Treasury note. As background, the DGS10 yield was 4.95% on September 10, 2026, and approximately 5.0% in early November 2026, reflecting an environment of relatively elevated long-term U.S. interest rates. The yield on December 31, 2027, will depend on developments in inflation, monetary policy, economic growth, fiscal conditions, and investor demand for Treasury securities. The forecast target is the single daily DGS10 observation for December 31, 2027, not an annual average, monthly average, intraday market quote, or year-end yield estimate from another provider. Source: FRED series DGS10 (https://fred.stlouisfed.org/series/DGS10).

How it resolves

The question will resolve to the numerical value of the observation dated December 31, 2027, in FRED series DGS10, expressed as a percentage. The primary resolution source is the Federal Reserve Bank of St. Louis's FRED database at https://fred.stlouisfed.org/series/DGS10. The underlying official source is the Board of Governors of the Federal Reserve System's H.15 Selected Interest Rates release, specifically the 10-year Treasury constant-maturity yield. December 31, 2027, falls on a Friday. If FRED publishes a valid numerical DGS10 observation for that date, that value will determine the outcome, regardless of when the observation is actually released. Resolution should occur after the observation becomes available, allowing up to 30 calendar days after December 31, 2027, for ordinary publication delays or initial corrections. If FRED has not published the observation by January 30, 2028, the corresponding official H.15 observation from the Federal Reserve Board may be used, provided it is explicitly dated December 31, 2027, and uses the same 10-year Treasury constant-maturity yield definition. If an observation is available from either authorized source by that deadline, the question will resolve using that observation. If both sources provide values and they differ, the FRED DGS10 value takes precedence. If neither source provides a valid numerical observation for December 31, 2027, by January 30, 2028, the question will be annulled rather than using a substitute date or interpolated value. The reported yield will be interpreted directly in percentage points: for example, a published DGS10 value of 4.75 resolves to 4.75%. The outcome will use the published numerical precision without additional rounding. Subsequent historical revisions published after resolution will not change the outcome.

Fine print

The specified date refers to the observation date in the U.S. Federal Reserve's daily interest-rate dataset, not the date on which the data becomes publicly available. The forecast concerns the nominal 10-year Treasury constant-maturity yield, not the inflation-adjusted real yield, a Treasury Inflation-Protected Securities (TIPS) yield, the federal funds rate, or the yield on an individual Treasury security. The DGS10 series is measured in percent per annum, so 5.00 represents a 5.00% annualized yield, not 0.05%. The outcome may have decimal precision, so this is a continuous numeric forecast rather than a discrete integer forecast. The forecast range of 2% to 9% is intended to cover a broad set of plausible outcomes, not to impose hard limits on the yield. Values outside that range remain valid. If December 31, 2027, has no published observation because of a market holiday, an extraordinary market closure, a reporting interruption, or another reason, the question will be annulled under the resolution criteria rather than substituting the preceding or following trading day's yield. If the DGS10 series is renamed without a substantive methodological change, the corresponding successor observation may be used. If the series is discontinued or materially redefined before the target date and no directly comparable official observation is available, the question will be annulled.