Higher yields improve the strategic case for bonds, but duration remains a mixed tactical proposition

Overview

This report reassesses duration and bond allocations after a sharp rise in global yields and a historically poor five-year period for bond investors. Goldman Sachs concludes that elevated starting yields improve bonds' medium- and long-term role in balanced portfolios, but the near-term allocation case remains mixed because bond diversification is less reliable and equity earnings remain supportive.

Core views

Global bond yields have risen sharply year to date amid stronger nominal growth, lingering inflation, fiscal concerns, increased competition for capital from AI investment, higher energy prices and the Middle East conflict. The report places the sell-off in historical context: US 10-year yields at 5% are close to their 250-year average of 4.7%, or 4.5% excluding the 1970s, rather than unusually high. Yet the prior starting point of exceptionally low post-GFC yields left investors vulnerable: following the 2022 drawdown, five-year rolling nominal and real returns on US 10-year bonds were among the worst in more than a century. The report argues that the macro regime has shifted toward higher neutral rates, greater inflation volatility, larger fiscal risks and stronger competition for capital. The central question is whether those higher yields make bonds more attractive in balanced portfolios. Goldman Sachs frames the optimal bond weight through four variables: relative return, relative risk, equity-bond diversification, and the return available on cash. Higher yields improve carry and prospective buy-and-hold returns while creating a larger cushion against future yield increases, which should lower bond risk over longer horizons. Still, valuations have limited power to predict near-term bond returns, and relative returns versus equities are driven chiefly by equities. A model using 10-year rolling US real GDP growth and inflation explains broad yield trends since the 1960s and places current US 10-year yields close to recent long-term fair value, but the report stresses that forward-looking markets can overshoot or undershoot such estimates. Historical portfolio analysis illustrates why the conventional 60/40 allocation cannot simply be assumed to be optimal. A post-1950 efficient frontier combining the S&P 500 and US 10-year bonds produced a tangent portfolio with roughly a 36% bond weight, close to 60/40. Before the COVID-19 crisis, the optimal mix was more like 40/60, resembling risk parity; in the last five years, however, the efficient frontier became very steep and the estimated optimal allocation shifted toward 100% equities, with cash used to adjust overall risk tolerance. A US 60/40 portfolio's five-year rolling Sharpe ratio lagged an outright equity allocation after the COVID-19 crisis because both asset classes suffered in 2022 and bonds subsequently remained a drag on risk-adjusted results. The report's simulations show why the bond weight falls when equity-bond correlations become more positive, equity returns improve relative to bonds, bond volatility rises, or cash returns become more competitive. Its illustrative assumptions for these simulations are a 10% equity return with 15% volatility, a 4.5% bond return with 6% volatility, a 3.5% cash return and zero equity-bond correlation; Goldman Sachs cautions that unconstrained mean-variance optimization is highly sensitive to expected-return and covariance assumptions and should be read as a directional signal rather than a portfolio recommendation. In the last five years, cash materially outperformed long-dated bonds, making equity-plus-cash attractive for investors with a positive equity risk premium. As bond yields rise, however, the relative case for bonds improves. Equity performance remains the principal near-term determinant of equity-versus-bond returns. Although the equity/bond yield gap has narrowed as bond yields rose and Shiller P/Es reached Tech Bubble-level readings, Goldman Sachs argues that current elevated equity valuations are supported by high corporate profitability. The gap is near a macro-implied fair value based on 10-year inflation, GDP growth and S&P 500 return on equity. Earnings growth and expectations for ROE, particularly the effect of AI on profitability, are therefore more important than starting valuations for relative performance in the near term. Historically, equities could still outperform bonds from expensive relative valuations when earnings growth, especially real earnings growth, was strong. Higher yields should also reduce bond risk in the medium term. Since 2022, low initial yields and high duration made bonds unusually volatile relative to equities; over the last five years, equity volatility was only about twice that of 10-year bonds, whereas the average ratio since the 1990s has been closer to 2.5 times and has reached around three times during major equity drawdowns. Goldman Sachs expects bond risk relative to equities to moderate unless yields rise materially again, as they did in the early 1980s. Yet elevated near-term rates volatility, high equity valuations and uncertainty around AI profitability leave both sides of the balanced portfolio exposed to changing macro conditions. The diversification function of duration is likely to be weaker than in the post-GFC period. Negative equity-bond correlations were historically unusual and largely occurred when inflation was well anchored. With higher inflation volatility and fiscal risks, Goldman Sachs expects correlations to move closer to zero, making bonds more useful as income assets than as reliable risk mitigators. The report expects inflation to normalize into 2027, which should help correlations decline from their elevated recent levels, but it warns that a future recession may generate smaller long-yield declines than the Tech Bubble or GFC episodes. Higher nominal growth, potential fiscal spending and less central-bank quantitative easing could constrain the traditional bond buffer. For tactical asset allocation, Goldman Sachs is neutral across assets over three months. It identifies inflation momentum, energy prices, the Middle East conflict, central-bank policy and the evolution of the AI capex boom as the main variables for both bonds and equities. Yield increases greater than 25 basis points over three months have historically begun to weigh more on equities than on bonds, particularly when driven by energy prices or aggressive tightening. Under its macro baseline of falling energy prices, continuing core-inflation normalization and less hawkish central banks, the report expects bonds to stabilize into year-end and forecasts German, US, UK and Japanese 10-year yields of 3.25%, 4.75%, 5% and 3%, respectively. Over six to 12 months, a material growth slowdown or fading AI-related earnings tailwinds would strengthen the case for more bonds. However, Goldman Sachs expects double-digit, though slower, 2027 earnings growth broadly in line with consensus and therefore remains overweight equities, neutral bonds and underweight credit over 12 months. Over longer horizons, higher yields support a normalization of strategic bond allocations toward a 35–55% long-run range rather than an immediate return to a simple 60/40 template. Across six structural scenarios—including Goldilocks with an AI boom, reflation, an AI bust, stagnation and stagflation—the report finds bond weights higher than in the last five years and modestly above the post-1950 optimum, while emphasizing that equity outcomes still drive most cross-scenario return variation. Implementation should focus on carry, convexity and broader diversification rather than a blanket duration increase. Long-duration bonds offer greater recession convexity and helped 60/40 portfolios during the Tech Bubble collapse and the GFC, but their scope for large yield declines may be reduced in the next crisis. Goldman Sachs' rates team sees the greatest potential relief in the front end versus forwards, favoring the belly of the curve while recognizing its lower convexity. With relatively contained rates volatility, the report highlights shorter-dated receiver swaptions as a potentially more convex recession hedge and notes that three-month TLT puts can protect portfolios that already carry duration risk; long curve-volatility positions can hedge extreme growth or policy tails. Finally, the report proposes equity and hybrid-credit complements where bonds offer less protection. It favors diversification between global AI-exposed equities and high-dividend, low-volatility equities, whose correlation with the S&P 500 has declined and whose negative-beta representation has risen as technology and AI dominate benchmarks. These styles lagged the broad market since the COVID-19 crisis but outperformed during technology-led drawdowns and relative to the market excluding technology. Convertible bonds are presented as equity-like bonds that provide embedded-call upside with relatively low duration; increased issuance tied to AI funding needs and the option value created by higher volatility can make convertibles useful late in the cycle.

Analysis framework

Goldman Sachs combines long-run historical comparisons, rolling return and volatility analysis, efficient-frontier portfolio simulations, macro-based yield analysis, conditional equity-versus-bond return comparisons, and structural scenario analysis. It then applies the findings to tactical, 6–12 month and longer-horizon asset-allocation decisions and discusses curve positioning, options hedges, equity style diversification and convertibles.

Methodology notes

• Quantitative, Factor, and Portfolio TheoryMean-Variance Analysis (Markowitz)

  Mean-variance optimization and the efficient frontier

  The report uses expected returns, volatility and equity-bond correlation to show how the risk-adjusted optimal bond weight changes across market regimes. It cautions that unconstrained optimization is highly sensitive to its assumptions.

• OtherYield curve analysis

  Yield-curve positioning and forward-rate analysis

  The report assesses higher neutral-rate expectations, term premia, curve shape and the relative appeal of the front end, belly and long end when managing duration and recession convexity.

• Quantitative, Factor, and Portfolio TheorySharpe and Information Ratios

  Rolling Sharpe-ratio comparisons

  Historical Sharpe ratios are used to compare the risk-adjusted performance of 60/40 portfolios, equities and different duration exposures.

• Other

  Macro-based long-term return scenario framework

  Goldman Sachs evaluates portfolio implications under six structural scenarios defined by growth, inflation and corporate profitability, including AI boom and AI bust cases.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• US 10-year bonds

  Core duration asset used to assess balanced-portfolio bond allocations

  Strengths

  Higher starting yields improve carry and prospective buy-and-hold returns, while reducing risk over longer horizons.

  Weaknesses

  Near-term returns remain sensitive to macro conditions and bond diversification benefits are less reliable with higher equity-bond correlations.

  Comparison

  Relative performance versus equities is driven largely by equity earnings growth; cash materially outperformed long-dated bonds over the last five years.

  Risks

  Further material yield increases, inflation volatility and fiscal risks could sustain elevated duration volatility.

• Long-duration government bonds

  Recession-convexity component of balanced portfolios

  Strengths

  Historically supported 60/40 portfolios during the Tech Bubble collapse and the GFC.

  Weaknesses

  Long duration detracted from risk-adjusted returns over the last five years and may have less scope for large yield declines in a future crisis.

  Comparison

  The report sees relatively greater potential for front-end relief than for forwards, favoring the belly of the curve but with less convexity.

  Risks

  Higher energy prices, aggressive central-bank tightening and persistent fiscal pressure could push long yields higher.

• Low-volatility and high-dividend equities

  Bond-like equity diversification complement

  Strengths

  Lower correlation with the S&P 500 and a growing set of negative-beta stocks can improve diversification during technology-led drawdowns.

  Weaknesses

  These stocks have lagged the broad market since the COVID-19 crisis.

  Comparison

  They performed better relative to the market excluding technology and materially outperformed in recent technology-driven drawdowns.

  Risks

  Europe, which benefits from lower technology exposure, faces separate headwinds from higher gas prices and politics.

• Convertible bonds

  Equity-like bond exposure with embedded optionality

  Strengths

  Offers equity upside through an embedded call option, relatively low duration and potential downside protection late in the cycle.

  Weaknesses

  Its effectiveness depends on the value of the embedded option and market volatility.

  Comparison

  The report states that convertibles outperformed a 60/40 portfolio during the late 1990s and around the GFC.

  Risks

  A decline in option value or adverse equity-market outcomes can weaken the asymmetric payoff.

Key data

• US 10-year Treasury yield5%Close to the 250-year average of 4.7%; 4.5% excluding the 1970s.
• Post-1950 tangent portfolio bond weightc.36%Efficient-frontier estimate for a portfolio combining the S&P 500 and US 10-year bonds.
• Long-run strategic bond allocation range35–55%The report's normalization range after the unusually low bond allocations of the past five years.
• Illustrative portfolio-simulation assumptions10% equity return, 15% equity volatility; 4.5% bond return, 6% bond volatility; 3.5% cash return; 0 correlationUsed in Exhibits 7–10 to illustrate allocation sensitivities.
• Tactical year-end 10-year yield forecastsGermany 3.25%; US 4.75%; UK 5%; Japan 3%Goldman Sachs rates-team forecasts cited in the report.
• Yield-change thresholdMore than 25bp over three monthsHistorically, such sharp yield increases begin to weigh more on equities than bonds.
• Median US 10-year yield at S&P 500 bear-market peaks since 19504.7% nominal; 2.3% real; 2.5% breakeven inflationUsed to assess the likely size of the bond buffer in future equity sell-offs.

Impact & implications

The report sees a gradual strategic restoration of bonds' role as higher yields improve carry and reduce medium-term duration risk. It does not expect a return to a mechanically constructed 60/40 portfolio: less-negative equity-bond correlations, attractive cash returns and resilient equity earnings require more selective use of duration, curve exposure, options and equity-style diversification.

Risks

• Persistent inflation volatility, fiscal risks and higher neutral rates could keep equity-bond correlations near zero and reduce bonds' diversification value.
• Further sharp increases in yields, especially from higher energy prices or more aggressive central-bank tightening, could pressure equities and sustain bond volatility.
• A deterioration in profitability among mega-cap technology companies before AI adoption benefits emerge could materially weaken equity returns.
• In a future recession, higher nominal growth, fiscal spending and less central-bank quantitative easing could limit the decline in long-dated yields and reduce duration's protective value.

What to watch

• Inflation momentum, energy prices, the Middle East conflict and central-bank policy over the tactical horizon.
• The evolution of AI capex, earnings revisions and return-on-equity expectations as drivers of equity performance relative to bonds.
• Whether inflation normalizes into 2027 and equity-bond correlations decline from elevated levels.
• Evidence of a material growth slowdown or fading AI-related earnings tailwinds, which would strengthen the case for higher bond allocations.
• Term premia, neutral-rate pricing, yield-curve movements and rates volatility when selecting duration and options hedges.