Strategic & Tactical Asset Allocation

Capital Market Assumptions and Expected Returns

The inputs that determine every strategic allocation decision.

Capital market assumptions (CMAs) are forward-looking estimates of expected return, volatility, and correlation for each asset class over a 10-year horizon. They are the primary inputs to strategic asset allocation — and getting them right, or at least less wrong than historical averages, is the most important analytical task in the policy portfolio design process.

By Swoopr Editorial Team

Published · Updated

AI-assisted content — disclosure

Direct Answer

Direct answer: Capital market assumptions are forward-looking estimates of expected return, volatility, and correlation for major asset classes, typically over a 10-year horizon. They are constructed using analytical frameworks rather than historical averages because markets are non-stationary: the US equity return of 10.5% per year from 1926 to 2025 reflects starting conditions — dividend yields, payout ratios, earnings growth, and valuations — that do not describe current conditions. Forward-looking CMAs for US equities as of mid-2026 range from 5.5% to 7.5% per year across major asset manager publications, reflecting lower starting dividend yields and higher starting valuations than the historical average.

The three most common CMA construction frameworks are: the building blocks method (risk-free rate + inflation premium + equity risk premium for each asset class), the dividend discount model or earnings yield framework for equities, and yield-to-maturity with credit spread decomposition for fixed income. Each framework makes explicit the economic inputs being forecasted, which makes disagreement reviewable and auditable rather than hidden inside historical statistics.

Key Takeaways

Core Concepts

1. The Building Blocks Method

The building blocks method decomposes expected return for any asset class into its component sources of return: real risk-free rate + expected inflation + asset class risk premium + any additional premia (illiquidity, credit, etc.). This decomposition makes the assumptions explicit and independently reviewable — if you disagree with the expected inflation assumption, you can change that component without changing the structure of the framework.

For US equities, the building blocks might be: real risk-free rate (proxied by TIPS yield) 2.0% + expected inflation 2.3% + equity risk premium 3.5% + small cap/factor premium 0.5% = nominal expected return 8.3%. Each component can be estimated independently — the TIPS yield is observable; inflation can be implied from the breakeven in TIPS vs. nominal Treasuries; the ERP is the most debated and uncertain element.

For investment-grade bonds, the building blocks are simpler: nominal risk-free rate + credit spread + duration adjustment for roll-down. A 10-year Treasury at 4.5% plus a IG credit spread of 100bps gives a 5.5% expected return before accounting for default losses. The critical input for bonds is simply the current yield curve, which makes bond CMAs considerably more precise than equity CMAs.

The building blocks method applied consistently across asset classes produces a set of expected returns that can be compared and stress-tested. If the equity ERP assumption is reduced from 3.5% to 2.0%, the framework immediately shows the impact on total equity expected return and the resulting change in optimal policy weights. This transparency is the main advantage over using historical averages.

2. Equity Expected Returns: CAPE and Earnings Yield Frameworks

Robert Shiller's cyclically adjusted price-to-earnings (CAPE) ratio divides the current price of the equity market by the average real earnings over the prior 10 years, smoothing out cyclical earnings fluctuations. The inverse of CAPE — the CAPE earnings yield — is a useful forward-looking expected return estimate. At a CAPE of 30 (as of mid-2026 for US equities), the CAPE earnings yield is approximately 3.3%, which represents a base estimate of real equity return before accounting for earnings growth beyond trend.

The full Shiller earnings yield framework for equity expected return is: CAPE earnings yield + expected real earnings growth + expected change in valuation (mean reversion in CAPE). If CAPE is 30 and the long-run average is 17, and mean reversion occurs over 10 years, the annual valuation drag is approximately ln(17/30)/10 ≈ −5.7%/10 ≈ −0.57% per year. Adding expected real earnings growth of 1.5% per year: 3.3% + 1.5% − 0.57% = 4.23% real equity return, or approximately 6.5% nominal with 2.3% inflation.

An alternative is the dividend discount model (DDM): expected return = dividend yield + expected dividend growth rate + expected change in valuation. With a dividend yield of 1.3%, expected real dividend growth of 2.0%, inflation of 2.3%, and zero valuation change, the expected nominal return is 1.3% + 2.0% + 2.3% = 5.6%. The DDM is more appropriate for high-quality, stable-dividend markets and less so for high-growth markets where dividends are low and buybacks dominate.

For international developed market equities (MSCI EAFE), starting valuations are typically lower (CAPE around 15–18 vs. 30+ for US), leading to higher expected returns: approximately 7–9% nominal in most institutional CMA sets. For emerging markets equities (MSCI EM), even lower starting valuations but higher political, currency, and liquidity risk suggest expected returns of 8–11% with commensurately higher volatility and risk of permanent capital loss.

3. Fixed Income Expected Returns

For investment-grade fixed income, the yield-to-maturity at the time of construction is the single best predictor of subsequent 10-year realized return, with an R² exceeding 0.9 in empirical studies. This near-perfect predictability reflects the mechanical nature of bond return: the yield is a contractual cash flow commitment, not a forecast. If the bond is held to maturity and does not default, yield-to-maturity equals realized return.

For a diversified bond portfolio like the Bloomberg US Aggregate Bond Index (which has an average duration of approximately 6.5 years as of mid-2026 and a yield-to-maturity of approximately 4.8%), the 10-year expected return CMA is simply the starting yield, adjusted for any duration mismatch and realistic credit loss rates. With credit losses of approximately 0.2% per year on IG bonds, the net expected return is approximately 4.6%.

The main sources of uncertainty in bond CMAs are: (1) reinvestment rate risk — if yields decline significantly, the income from coupon reinvestment will be at lower rates than the current yield implies; (2) duration mismatch between the benchmark and the CMA horizon; and (3) credit migration and default losses, which are typically small for IG but can be significant in high-yield or credit-stressed environments. For TIPS (Treasury Inflation-Protected Securities), the real yield is directly observable and the only uncertainty is the path of inflation adjustment; TIPS CMAs are among the most reliable in any CMA set.

4. Alternative Asset Class CMAs

Real estate investment trusts (REITs) are often modeled as a blend of equity-like and bond-like return components: the property income yield (typically 3–5%) plus expected income growth (tracking inflation) plus a valuation component. REITs historically have had higher yields and lower valuation multiples than the broad equity market, leading to expected returns of 6–8% nominal, with higher volatility (20–25% annually) than investment-grade bonds but more income stability than pure equities.

Private equity CMAs are the most uncertain and controversial. The traditional approach adds an illiquidity premium (typically 2–4%) to public equity CMAs, based on empirical studies of private equity IRRs relative to public market equivalents. A simpler and arguably more honest approach for an LBO-focused private equity fund is to start with public equity CMA (7%), add leverage premium (1–2%) reflecting that PE funds use significant debt, add operational alpha (1–2% if the manager has demonstrated value creation), subtract fee drag (1.5–2%): net expected return 7–10% with uncertainty range ±3%. Private equity CMA estimates vary enormously across institutions.

Commodity CMAs are among the lowest in most institutional sets: the long-run expected return of a diversified commodity futures index is approximately 0–3% nominal, reflecting that commodities have low or negative risk premium (investors can earn the risk-free rate + positive roll yield in some conditions, but roll yield is negative in contango markets). The value of commodities in a portfolio is primarily diversification against inflation shocks, not expected return enhancement.

5. Constructing the Correlation Matrix

The correlation matrix used in strategic allocation optimization is as important as the expected return vector. Historical correlations are somewhat more stable than expected returns, making them slightly less problematic to estimate from history — but regime-dependence is a serious issue. The equity/bond correlation was negative from approximately 2000 to 2021 (both US Treasuries and equities rose in recessions as the Fed cut rates and investors fled to quality), but turned briefly positive in 2022 when both asset classes fell simultaneously due to inflation and rate hikes.

CMA correlation matrices should reflect the full range of economic regimes likely over a 10-year horizon, not just the recent period. A common approach is to blend: a normal-regime correlation matrix (reflecting the typical 1990–2020 experience) with a stress-regime correlation matrix (reflecting 1970s stagflation or 2022-style inflation shock). The stress-regime matrix typically shows higher equity/credit correlation, higher equity/real estate correlation, and reduced equity/bond diversification benefit.

For the optimization to be well-conditioned, the correlation matrix must be positive definite — all eigenvalues must be positive. Sample correlation matrices with many asset classes sometimes violate this condition due to estimation noise. Ledoit-Wolf shrinkage or Factor model shrinkage can be applied to the correlation matrix before use in optimization.

Worked Scenario

Building a simplified 5-asset-class CMA set as of August 2026 using the building blocks method.

  1. Observable inputs: 10-year TIPS yield = 2.1% (real risk-free rate); 10-year breakeven inflation = 2.3%; 10-year nominal Treasury yield = 4.4%; US IG Corporate index yield-to-maturity = 5.0%; US equity CAPE = 29.5; US equity dividend yield = 1.4%; MSCI EAFE CAPE = 16.8 (implied earnings yield = 5.95%).
  2. US equities: CAPE earnings yield = 1/29.5 = 3.4%. Add expected real earnings growth = 1.5%. Add inflation = 2.3%. Subtract valuation drag from mean reversion (assume partial mean reversion, drag = −0.3%). Total: 3.4 + 1.5 + 2.3 − 0.3 = 6.9% nominal expected return. Volatility assumption: 15.5% per year.
  3. International developed equities: MSCI EAFE earnings yield = 5.95%. Add expected real earnings growth = 1.2%. Add inflation = 2.3%. Subtract valuation drag = −0.1%. Total: 5.95 + 1.2 + 2.3 − 0.1 = 9.35%, rounded to 9.0% to reflect currency risk discount. Volatility: 17.0% per year.
  4. US investment-grade bonds: Yield-to-maturity = 5.0%. Subtract expected credit losses = 0.2%. Subtract estimation uncertainty haircut = 0.2%. Total: 4.6% nominal expected return. Volatility: 6.5% per year.
  5. US TIPS: Real yield = 2.1%. Total real expected return = 2.1%; nominal = 4.4%. Volatility: 8% per year (higher than nominal bonds due to inflation adjustment uncertainty in short run).
  6. REITs: Property income yield = 3.8%. Add real income growth = 1.5%. Add inflation = 2.3%. Subtract valuation drag = −0.2%. Total: 7.4%. Volatility: 20% per year.
  7. Cross-asset correlations: Equity/IG bonds = −0.1 (normal regime) to +0.3 (inflation regime), use 0.0 as base; Equity/TIPS = −0.2; Equity/REIT = 0.65; International equity/US equity = 0.75. Final CMA set documented with methodology and all input assumptions for governance review.

Measurement Framework

MeasurementWhat it tells you
CAPE earnings yield vs. historical rangeIs the current starting valuation above or below the historical average CAPE? High CAPE reduces forward equity expected return relative to history.
Current yield-to-maturity of bond benchmarkBest single estimate of annualized bond return over the next 10 years for a buy-and-hold strategy. Compare to last year's CMA to see how much bond expected returns have shifted.
TIPS real yieldThe real return available from inflation-protected government bonds — the risk-free real rate that all other CMAs should be assessed against as a risk premium above this floor.
Equity risk premium (equity CMA minus bond CMA)Is the compensation for bearing equity risk above or below the historical average? If ERP is compressed, the case for high equity allocations weakens.
Dispersion across CMA sets (major manager range)How much do BlackRock, Vanguard, JP Morgan, and other major CMA publishers disagree? Large dispersion signals high uncertainty and argues for wider scenario analysis.
Policy portfolio expected return vs. required returnDoes the policy mix, at these CMAs, meet the investor's required return? If not, the required return or the allocation must change.

Common Failure Modes

Using Historical Returns as CMAs

The 10.5% historical nominal return on US equities from 1926 to 2025 includes periods of very low starting valuations (CAPE of 5–10 in the 1940s and early 1980s) that are structurally different from today's starting conditions. Using this average as a forward CMA leads to return expectations that are approximately 2–4% per year too high at current valuations, with compounding effects that produce wildly optimistic long-run wealth projections.

The correct test is to ask: given current dividend yields, earnings yields, and reasonable assumptions about earnings growth, what return can equities mechanically be expected to generate? The answer at a CAPE of 29–30 is approximately 6–7% nominal, not 10.5%. The discipline is to derive CMAs from current data, not copy the long-run historical average.

Ignoring the Geometric-Arithmetic Return Gap

A 7% arithmetic expected return with 15% volatility compounds to a 10-year geometric return of approximately 5.9% (7% − 0.5 × 0.15² = 5.875%). If strategic allocation projections use arithmetic returns for compounding calculations, they overstate projected terminal wealth. For a $1 million portfolio, 7% compounded over 10 years = $1.967 million; 5.9% compounded over 10 years = $1.775 million — a difference of nearly $200,000. This is not a rounding error; it is a structural bias that systematically overstates the achievability of return targets.

Extrapolating Correlation Structure from a Single Regime

The negative equity/bond correlation of 1998–2021 was a specific feature of a low-inflation, Fed-dominant monetary regime where the Fed could cut rates aggressively in recessions, driving up bond prices and partially offsetting equity losses. This correlation cannot be assumed to persist in an inflation-volatile environment. Building a portfolio with a 30% bond allocation based on an assumed −0.3 equity/bond correlation, then experiencing a 2022-style event where both fall 15–20%, is a correlation assumption failure that is entirely foreseeable given the historical record.

Failing to Distinguish Nominal from Real CMAs

CMAs expressed in nominal terms are the right input for nominal return targets; CMAs expressed in real terms are the right input for inflation-linked liability matching. Mixing the two — using real return CMAs to assess progress toward a nominal target, or using nominal CMAs to compare against an inflation-linked liability — produces systematic errors. The discipline is to specify at the outset whether the required return is nominal or real and to use consistently-denominated CMAs throughout the analysis.

Frequently Asked Questions

How often should capital market assumptions be updated?

Most institutional investors update CMAs annually, typically in the fourth quarter for use in next-year policy reviews. The key inputs — equity valuations, yield levels, credit spreads — change continuously, but annual updates capture meaningful changes without creating excessive policy instability from short-run fluctuations. A major structural shift (e.g., interest rates rising from 0% to 5% in one year, as happened in 2022) warrants an out-of-cycle update to avoid using stale bond CMAs.

Why do major asset managers' CMAs differ significantly?

CMAs from BlackRock, Vanguard, JP Morgan, and State Street can differ by 1–3% per year for the same asset class because the managers use different ERP models, different earnings growth assumptions, different treatments of valuation mean reversion, and different inflation forecasts. For equities, the biggest source of disagreement is typically the assumed speed and degree of CAPE mean reversion. These differences are meaningful for policy portfolio design and argue for building the policy around a range of CMAs rather than a single set.

What is the equity risk premium and why is it uncertain?

The equity risk premium (ERP) is the expected excess return of equities over the risk-free rate, compensating investors for bearing equity market risk. Historical ERP estimates range from 3% to 6% depending on the sample period, country, and whether the arithmetic or geometric mean is used. Forward-looking ERP estimates are derived from CAPE earnings yields minus current risk-free rates: at a CAPE of 30 and a 10-year TIPS yield of 2.1%, the implied real ERP is approximately 3.4% − 2.1% = 1.3%. This compressed ERP at high valuations is a structural reason to expect lower equity returns than the historical average.

Are emerging market CMAs reliable?

Emerging market CMAs are the least reliable in most institutional sets. While lower starting valuations suggest higher expected returns (EM CAPE earnings yields of 6–8% vs. 3–4% for US), EM portfolios face political risk, currency risk, capital controls, accounting quality risk, and liquidity risk that are difficult to quantify. Empirically, EM equities have provided lower Sharpe ratios than developed market equities over most 20-year windows despite higher expected return assumptions, suggesting that the premia for EM risks are imperfectly captured in CMA models. EM allocations should be accompanied by explicit scenario analysis of political and currency risk rather than reliance on a single expected return number.

How do CMAs handle alternative asset classes like infrastructure or private credit?

Infrastructure and private credit CMAs are built by starting with the closest public-market equivalent and adding an illiquidity premium. Infrastructure (airports, toll roads, utilities) is modeled as bond-like with equity-like income growth: a typical CMA might be 5.5% nominal return with 8–10% volatility. Private credit (direct lending to middle-market companies) is modeled as high-yield credit plus an illiquidity and complexity premium: perhaps LIBOR/SOFR + 400–600 bps, or approximately 8–10% floating-rate return in the current environment. Both CMAs carry high uncertainty because the performance track record for these asset classes is short, self-reported, and potentially smoothed.

Can I use someone else's CMAs rather than building my own?

Yes — most institutions use published CMAs from major asset managers (BlackRock's Capital Market Assumptions, Vanguard's Economic and Market Outlook, JP Morgan's Long-Term Capital Market Assumptions) rather than building their own from scratch. Using a published CMA set has advantages: it is independently documented, methodologically transparent, and updated annually with full audit trail. The main risk is that you do not understand the assumptions embedded in the numbers and cannot stress-test them sensibly. At minimum, read the methodology section of whatever CMA set you use before importing the numbers into your policy portfolio model.

What is the difference between a 5-year and a 10-year CMA horizon?

The choice of CMA horizon should match the policy portfolio's review horizon. A 10-year horizon is standard for long-term institutional investors (pension funds, endowments) because it smooths out cyclical variations in return and focuses on structural drivers. A 5-year horizon is sometimes used for tactical overlays or for investors with shorter spending horizons. Critically, equity CMAs on shorter horizons have much higher uncertainty — the starting CAPE has predictive power over 10 years but much less over 1–3 years, where momentum and short-run macro factors dominate valuation mean reversion.

Why is the correlation between equities and bonds not always negative?

The equity/bond correlation depends on the economic regime. In disinflationary recessions (US experience from 2000–2021), central banks cut rates aggressively when equities fell, driving up bond prices — this produced negative equity/bond correlation and made bonds an effective equity hedge. In inflationary environments (1970s, 2022), rising inflation and interest rate risk drove both equities and bonds down simultaneously, producing positive correlation. The switch between these regimes is regime-dependent and hard to predict in advance, which is why correlation assumptions should be scenario-weighted rather than assuming the recent negative correlation persists indefinitely.

Sources and Further Verification

Educational Disclaimer

This guide is for educational purposes only. Capital market assumption models are analytical tools with significant uncertainty; actual returns can and regularly do differ substantially from any forward-looking estimate. No specific investment or allocation is recommended. Consult a qualified financial professional before making investment decisions.