AA.1: Capital Market Expectations Framework

Predict

A CIO asks you to prepare 10-year capital market expectations. What are the biggest challenges you face, and how would your approach differ from simply extrapolating historical returns?

Challenges in Forecasting Capital Market Expectations

Capital market expectations (CMEs) are the long-term risk and return assumptions that drive strategic asset allocation. Developing them is both art and science, fraught with challenges:

Nine Major Challenges

Challenge	Description
Limitations of economic data	Data revisions, rebasing, regime changes reduce reliability
Data measurement errors	Transcription errors, survivorship bias, appraisal smoothing
Limitations of historical estimates	Non-stationarity — past regimes may not repeat
Ex-post vs. ex-ante	Realized returns include surprises; expected returns do not
Non-repeating data patterns	Data mining and overfitting historical anomalies
Failing to condition on current values	Current yields, valuations, and spreads matter
Misinterpretation of correlations	Correlations are unstable, especially in crises
Psychological biases	Anchoring, recency, representativeness, confirmation bias
Model uncertainty	All models are simplifications; none capture full reality

The key principle: condition expectations on current economic and market conditions rather than relying solely on unconditional historical averages.

Check Your Understanding

Why is it problematic to use the average historical equity risk premium from 1926-2025 as your forward-looking estimate?

Economic Analysis Framework for CMEs

Business Cycle Analysis

The business cycle provides a structured framework for setting shorter-term (1-3 year) CMEs:

Phase	GDP	Inflation	Policy	Asset Implications
Initial Recovery	Turning up	Low	Accommodative	Equities rally, credit tightens (positive), short-duration bonds
Early Expansion	Accelerating	Low-rising	Becoming less accommodative	Equities continue strong, spreads compress
Late Expansion	Slowing from peak	Rising	Restrictive	Equities vulnerable, commodities strong, flatten curve
Slowdown	Decelerating	Peaking	Peak restrictive → easing	Bonds rally, equities decline, steepen curve
Contraction	Falling	Falling	Accommodative	Government bonds outperform, credit spreads widen

Checklist Approach

Many practitioners use a checklist of leading indicators:

Yield curve slope (inversion signals recession)
Credit conditions (lending standards, spreads)
Consumer/business confidence surveys
Leading economic indicators (LEI)
Housing starts, durable goods orders
Labor market indicators (initial claims, quit rate)

Approaches to Setting CMEs

1. Risk Premium Approach (Building Blocks)

Expected return = Risk-free rate + Risk premiums

For each asset class, estimate:

Short-term default-free rate (anchor)
Term premium (for duration risk)
Credit premium (for default risk)
Equity risk premium (for systematic equity risk)
Illiquidity premium (for private markets)

2. Equilibrium Models

Singer-Terhaar Model

Combines the international CAPM with a segmentation adjustment:

$ERP_i = \rho_{i,M} \cdot \sigma_i \cdot \frac{ERP_M}{\sigma_M}$

Where:

$\rho_{i,M}$ = correlation of asset $i$ with global market portfolio
$\sigma_i$ = standard deviation of asset $i$
$ERP_M / \sigma_M$ = global market Sharpe ratio

Segmentation adjustment: Markets are not fully integrated. For a partially segmented market:

$ERP_i = \delta \cdot ERP_{integrated} + (1 - \delta) \cdot ERP_{segmented}$

Where $\delta$ = degree of integration (0 to 1), and the fully segmented premium uses $\rho = 1$ (only own risk matters).

Key insight: Segmented markets have higher risk premiums because diversification benefits are unavailable.

3. Statistical Methods

Shrinkage estimators: Blend sample estimates with structured estimates (reduce estimation error)
Time-series models: VAR, VECM for shorter-horizon forecasts
Regime-switching models: Different return distributions in different economic states

Exogenous Shocks and Tail Risks

CMEs must account for events outside normal models:

Geopolitical events: Wars, sanctions, trade conflicts
Pandemics and natural disasters: Supply chain disruption, demand shocks
Policy regime changes: Unexpected monetary/fiscal shifts
Technology disruptions: AI, energy transitions
Financial crises: Contagion, liquidity freezes

Stress Testing CMEs

Rather than single-point estimates, best practice includes:

Base case (most likely scenario, ~60% weight)
Optimistic scenario (~20% weight)
Pessimistic scenario (~20% weight)
Tail risk scenarios (unweighted stress tests)

The weighted scenarios can be combined to produce scenario-weighted expected returns that incorporate tail risk asymmetries.

Try It Yourself

Constructed Response: Capital Market Expectations

Capital Market Expectations Practice

Problem 1 of 3(advanced)

An analyst estimates the equity risk premium for a market with the following characteristics: standard deviation of 22%, correlation with the global portfolio of 0.55, global market Sharpe ratio of 0.28, and degree of market integration of 0.80. The risk-free rate is 3%. Using the Singer-Terhaar approach, the expected return for this market is closest to:

Explain Back

Explain the Singer-Terhaar model to a junior analyst. Cover what it does, why the segmentation adjustment matters, and how the final expected return is constructed. Use a specific numerical example.

Reflect

1. Which approach to setting CMEs (risk premium, equilibrium, statistical) do you find most intuitive, and why?
2. How would you handle a situation where your model-based CMEs differ significantly from market consensus?
3. What exogenous shocks should be incorporated into current CMEs that historical models would miss?

Key Takeaway

Key Concepts:

CMEs should be conditioned on current market conditions, not just historical averages
Nine major challenges include data limitations, psychological biases, and model uncertainty
Business cycle analysis provides a framework for shorter-term CME adjustments
Singer-Terhaar model: Blends integrated and segmented risk premiums based on degree of market integration
Segmented markets demand higher risk premiums (less diversification available)
Risk premium approach: Expected return = Risk-free rate + Sum of risk premiums
Stress testing across scenarios captures tail risk better than point estimates

Connect

Next, we'll explore **Forecasting Returns** in detail, building on the framework here to develop specific return expectations for equities (Grinold-Kroner), fixed income, real estate, and other asset classes — the quantitative inputs that feed directly into your asset allocation optimization.