Goldman Sachs: 10-Year U.S. Treasury Yield to Fall to 4.6% by 2027
nashnova research
Goldman Sachs forecasts the 10-year U.S. Treasury yield will drop from 5.13% to 4.6% by end-2027, a roughly 50-basis-point decline; this means → the bank sees the hiking cycle peaking soon, making bonds worth holding again.
What road map did Goldman draw for global bonds?
Goldman's rates team, led by William Marshall, titled the outlook "A Better Cushion." This means → yields are high enough to act as a safety net — even if prices stay flat, coupon income alone delivers a return.
Specific forecasts: U.S. 10-year 5.13% → 4.6%, U.K. gilts 5.32% → 4.8%, German Bunds 3.37% → 3.1%, Japan JGBs rangebound near 2.75%.
Goldman expects Treasuries and gilts to deliver roughly 10% total return over the full cycle, with most of that coming from coupon income rather than price gains. In plain terms = you earn "interest money," not "trading money."
When do rate cuts actually arrive?
Goldman expects the Fed to hike two more times. Core PCE — the Fed's preferred inflation gauge, stripping out food and energy — needs to fall from 3.2% at end-2026 to 2.4% at end-2027 before the Fed starts cutting.
This means → yields will likely stay rangebound in the near term; the real turning signal comes in the second half of 2027.
The ECB is expected to hike once more in December; the Bank of England is forecast to stop at a 4.25% terminal rate in February 2027; the Bank of Japan is projected to hike three consecutive times at 25 bp per quarter, reaching a 2% terminal rate.
What does the AI investment boom have to do with bonds?
Goldman notes that AI-related capital spending may provide a floor for yields in the first half of 2027. In plain terms = companies borrowing heavily to fund AI keep demand for capital high, which keeps rates from falling easily.
Only once AI's impact on the labor market comes into view will bonds see a clearer bullish signal. This means → AI pushes rates up first, then down — the inflection depends on how fast AI displaces workers.
Is bonds' "insurance function" coming back?
Goldman's data show that when core CPI is above 2.5%, the one-year rolling correlation between the S&P 500 and 10-year Treasuries is roughly +30 — stocks and bonds move together. When core inflation drops below 1.75%, the correlation flips to about -35, and bonds regain their hedging power.
In plain terms = when inflation is high, stocks fall and bonds fall too — owning bonds doesn't protect your portfolio. Once inflation comes down, bonds rise when stocks drop, acting as insurance.
Recent volatility in AI-related tech stocks has already shown early signs of duration assets' diversification value, and falling energy prices may reinforce this. But with core inflation still well above 2.5%, bonds can hedge "AI risk" only — they cannot yet hedge "inflation risk." Last week's simultaneous stock-bond selloff confirmed this limitation.
How solid is Goldman's forecast?
Goldman's valuation model shows the U.S. 5-year/5-year forward rate — the market-implied rate for years 5 through 10 — is slightly above fair value; Germany is slightly below; Japan and the U.K. are notably underpriced.
The key test: whether the Fed pivots to cuts by end-2027 as Goldman expects. This reflects the fact that the biggest variable in the entire thesis is still the pace of inflation's decline.
Notably, Goldman's economists added an extra hike to their forecast after last week's above-consensus CPI print. This means → the bank's own base case still has room to be revised upward, and the timing of cuts could be pushed back.
市场有风险,内容仅供研究参考,不构成投资建议。
