GF Securities: 10-Year U.S. Treasury Yield Center May Settle at 4.0%-4.5% by 2027
nashnova research
GF Securities (广发证券) projects that the 10-year U.S. Treasury yield will center at 4.0%–4.5% in 2027, based on a 1.3× historical ratio between nominal GDP growth and the benchmark yield — with the floor hinging on whether AI capital-spending narratives shift.
Where does the 1.3× ratio come from?
From 2000 to 2019, the average ratio of U.S. nominal GDP growth to the 10-year Treasury yield was 1.3×.
Over the 13 quarters since Q2 2023, the ratio reverted to 1.3×; the two-year quarterly average sits at the same level.
This means → dividing nominal GDP growth by 1.3 gives a rough anchor for the 10-year yield's "theoretical center." GF's entire framework rests on this empirical rule.
Why did Q3 yields overshoot the theoretical center?
In Q1 and Q2, nominal GDP growth ran 5.8% and 6.3%; the 10-year yield averaged 4.2% and 4.4% — both at or below fair value.
Q3 nominal GDP growth was 5.9%, barely changed from Q2. At 1.3×, the theoretical center was roughly 4.6% — yet actual monthly averages climbed from 4.6% in June to 5.1% in October.
GF attributes the overshoot to two forces: oil prices rose from July to October, stoking inflation fears and brief speculation about a Fed October hike; and AI companies issued large volumes of long-dated bonds, reshaping long-end supply-demand — a phenomenon with no prior precedent.
Since early November, with the October-hike scenario ruled out and oil prices stabilizing, the yield has pulled back to 4.86% (Nov 6), beginning to revert toward the theoretical center.
Why 4.0%–4.5% for 2027?
GF assumes a higher inflation base will slow U.S. nominal GDP growth to roughly 5.2%. At 1.3×, the theoretical center is about 4.0%.
In plain terms = one notch lower on growth, one notch lower on yields — that is the through-line of the entire estimate.
The Fed's own 2026–2027 forecasts: real growth of 2.3%–2.4%, PCE inflation falling from 3.7% to 2.3%. But the GDP deflator — a broader price gauge — is stickier, so nominal GDP will not drop as fast.
Which three factors could keep yields above the center?
Factor one: nominal GDP slows gradually with rate hikes; yield declines will be equally gradual, not a one-step move.
Factor two: at least through H1 2027, AI capex remains in a trend expansion phase. AI firms will keep issuing long-dated debt, pushing up the term premium — the extra return investors demand for holding longer bonds.
Factor three: consensus expects the U.S. may still be hiking in H1 2027. Rising policy rates plus policy uncertainty constrain how far market rates can fall.
This means → the theoretical 4.0% is a floor, not a midpoint. Real-world yields will most likely run in the 4.0%–4.5% band, similar to H1 2026 levels.
What would it take to break below 4.0%?
GF states explicitly: breaching the lower bound depends on whether a demand-side shift emerges in the AI capex narrative.
In plain terms = only if AI investment cools, long-bond issuance shrinks, and the term premium recedes would yields have room to fall below 4.0%.
This reflects a larger signal: AI-sector financing behavior has become a structural variable in Treasury pricing — no longer just an intra-tech story.
市场有风险，内容仅供研究参考，不构成投资建议。