Portfolio Theory and Asset-Pricing Models

Introduction and Context

Investment management requires more than selecting securities with attractive individual characteristics. The central question is how securities interact within a portfolio and how their risks should be priced. Portfolio theory explains how investors combine assets to obtain the most desirable trade-off between expected return and risk. Asset-pricing models then connect required return to systematic risk and identify the economic forces that determine expected returns.

This chapter develops the mean–variance framework, diversification, efficient frontiers, investor utility, optimal portfolios, the Capital Asset Pricing Model (CAPM), beta, multifactor models and Arbitrage Pricing Theory (APT). It also evaluates the assumptions and empirical limitations that affect professional use of these models.

Learning Objectives

By the end of this chapter, the learner should be able to:

1. Explain how diversification affects portfolio risk.
2. Calculate expected portfolio return, variance and covariance.
3. Interpret the efficient frontier and identify optimal portfolios.
4. Apply investor utility to portfolio selection.
5. Explain CAPM, beta, the security market line and the market risk premium.
6. Distinguish systematic from unsystematic risk.
7. Compare CAPM with multifactor models and APT.
8. Evaluate the assumptions, evidence and practical limitations of asset-pricing models.
9. Apply portfolio theory to professional investment decisions and governance.

1. Risk, Return and the Logic of Diversification

The expected return of an asset is the probability-weighted average of possible returns:

Here, E(R_i) is the expected return of asset i, p_s is the probability of state s, R_is is the return of asset i in state s, and S is the number of states. Probabilities must be non-negative and sum to one. The measure is appropriate when the investor has a specified probability distribution for possible outcomes.

For a portfolio of n assets, expected return is the weighted average of constituent expected returns:

where w_i is the proportion invested in asset i and the weights sum to one for a fully invested portfolio. Short selling permits negative weights; leverage may produce weights whose sum exceeds one if borrowing is included explicitly.

Risk is commonly measured by variance or standard deviation. Portfolio variance is:

where σ_p² is portfolio variance and Cov(R_i,R_j) is the covariance between returns i and j. In matrix notation:

where w is the vector of portfolio weights and Σ is the covariance matrix. Standard deviation is σ_p = √(σ_p²). If returns are expressed as decimals, variance is in squared decimal units and standard deviation is in decimal return units.

Covariance measures the direction in which two assets move together:

where ρ_ij is correlation, bounded between −1 and +1, and σ_i and σ_j are standard deviations. Correlation is standardized covariance and has no units.

Diversification

Diversification reduces portfolio risk when asset returns are not perfectly positively correlated. It does not require that every asset be low risk; it requires that the assets do not all respond identically to economic events. With two assets:

If ρ₁₂=1, diversification provides no reduction relative to a perfectly aligned combination. If ρ₁₂<1, risk may be reduced. At ρ₁₂=-1, a zero-risk combination may be possible if the weights satisfy the appropriate volatility offset condition.

Diversification separates total risk into:

• Unsystematic risk: firm-specific or diversifiable risk, such as litigation against one company or a product failure.
• Systematic risk: market-wide risk that cannot be eliminated by holding more securities, such as broad economic contractions or changes in interest rates.
• Total risk: the combined variability of portfolio returns, often measured by variance or standard deviation.

A well-diversified investor should generally not expect compensation for bearing avoidable unsystematic risk. This principle is central to CAPM, although practical investors may retain some concentrated exposures because of taxes, liquidity, information advantages or liabilities.

2. Mean–Variance Portfolio Theory

Harry Markowitz’s portfolio theory treats portfolio choice as a problem involving two dimensions: expected return and variance. The theory is a one-period model and is most directly applicable when investors evaluate portfolios using expected return and risk, and when the relevant return distribution is adequately summarized by its first two moments.

A portfolio A dominates portfolio B if A offers at least as much expected return for no greater risk, or no greater risk for at least as much expected return. A portfolio is mean–variance efficient if no other feasible portfolio dominates it.

The Feasible Set and Efficient Frontier

The feasible set consists of all attainable combinations of assets under stated constraints, such as budget, short-selling, leverage and liquidity rules. Its upper boundary is the efficient frontier. The lower boundary may contain portfolios that are inefficient because another portfolio offers higher return at the same risk.

The global minimum-variance portfolio is the feasible portfolio with the smallest variance. For two risky assets, the weight in asset 1 that minimizes variance is:

The weight in asset 2 is w₂^*=1−w₁^*. This formula assumes two risky assets, a fully invested portfolio, known variances and covariance, and no additional constraints such as no short selling. If the calculated weight lies outside 0 to 1, the unconstrained solution involves short selling.

For a target expected return, the minimum-variance portfolio can be obtained by solving:

Here μ is the vector of expected returns, μ_T is the target return, 1 is a vector of ones, and w is the portfolio-weight vector. This quadratic optimization problem forms the basis of strategic asset allocation, although real-world applications add constraints and estimation procedures.

The Two-Fund Separation Result

Under the standard mean–variance framework, every efficient risky portfolio can be represented as a combination of two distinct efficient portfolios, provided the relevant assumptions and unconstrained conditions hold. This is called two-fund separation. It does not mean that all investors hold identical portfolios; they may choose different combinations according to risk tolerance.

3. Risk-Free Asset, Capital Allocation and the Sharpe Ratio

A risk-free asset has a known return R_f over the investment horizon, zero variance and zero covariance with risky assets in the idealized model. Combining it with a risky portfolio produces a capital allocation line (CAL):

where y is the proportion invested in risky portfolio P. If y exceeds one, the investor borrows at the risk-free rate; if y is negative, the investor lends and holds less than zero in the risky portfolio.

The slope of the CAL is the Sharpe ratio:

The Sharpe ratio measures expected excess return per unit of total risk. It is useful for comparing diversified portfolios, but it can be misleading for non-diversified portfolios, asymmetric returns or portfolios whose returns are smoothed or illiquid.

When a single risky portfolio is optimal for all investors and investors can borrow or lend at the risk-free rate, that portfolio is the market portfolio in the CAPM equilibrium. The resulting line is the capital market line (CML):

The CML applies to efficient portfolios formed from the risk-free asset and the market portfolio. It should not be confused with the security market line, which applies to individual assets and relates return to beta.

4. Investor Utility and Optimal Portfolios

Mean–variance investors can be represented by a quadratic utility function:

where U is investor utility and A is the coefficient of risk aversion. A higher A indicates greater dislike of variance. The expression assumes that expected return and variance are sufficient decision statistics, which is exact for normally distributed returns or quadratic preferences, but only an approximation in many markets.

For two portfolios, the investor chooses the one with the highest utility. Graphically, indifference curves in expected return–standard deviation space slope upward: greater risk must be compensated by greater expected return. More risk-averse investors have steeper indifference curves and choose portfolios closer to the risk-free asset or the minimum-variance region.

Worked Example: Two-Asset Diversification

Suppose an investor allocates 60% to Asset A and 40% to Asset B. Expected returns are 10% and 6%, standard deviations are 20% and 10%, and correlation is 0.25.

Expected portfolio return is:

E(R_p)=0.60(0.10)+0.40(0.06)=0.084=8.4%

Covariance is:

Cov_AB=0.25(0.20)(0.10)=0.005

Portfolio variance is:

σ_p²=(0.60²)(0.20²)+(0.40²)(0.10²)+2(0.60)(0.40)(0.005)

σ_p²=0.0144+0.0016+0.0024=0.0184

Thus:

σ_p=√0.0184≈0.1356=13.56%

The portfolio’s risk is below the weighted average of the individual standard deviations because the correlation is below one. The result demonstrates that portfolio risk cannot be calculated by simply averaging individual standard deviations.

5. The Capital Asset Pricing Model

The CAPM states that the expected return on an asset equals the risk-free rate plus compensation for its systematic risk:

where E(R_i) is the required or expected return on asset i, R_f is the risk-free rate, β_i is the asset’s market beta, and E(R_M)−R_f is the expected market risk premium.

The security market line (SML) is the graphical representation of CAPM. Its intercept is R_f, and its slope is the market risk premium. An asset with beta zero has a CAPM required return equal to the risk-free rate. An asset with beta one has the expected market return. A beta greater than one implies greater sensitivity to market movements; a negative beta implies that the asset tends to move opposite to the market.

Beta

Beta is the covariance of an asset’s return with the market return divided by market variance:

It can also be estimated as the slope coefficient in a time-series regression:

Here α_i is the intercept, ε_it is the residual, and t indexes time. The regression assumes a sufficiently appropriate market proxy, stable or interpretable parameters, and residual conditions suitable for estimation. Beta is an estimate, not a permanent physical characteristic; it may change with leverage, business strategy, operating conditions and measurement interval.

Worked CAPM Application

Assume a risk-free rate of 4%, expected market return of 10% and an asset beta of 1.3. The required return is:

E(R_i)=0.04+1.3(0.10−0.04)=0.04+0.078=0.118=11.8%

If analysts forecast the asset’s return at 14%, the implied positive alpha relative to the CAPM benchmark is approximately 2.2 percentage points, before considering estimation error, transaction costs, taxes and model risk. A forecast above required return does not guarantee outperformance.

CAPM Assumptions

The traditional one-period CAPM relies on strong assumptions:

• Investors are rational, risk-averse and choose portfolios using expected return and variance.
• Investors have homogeneous expectations about returns, variances and covariances.
• Markets are frictionless: no taxes, transaction costs, restrictions or liquidity differences.
• Investors can borrow and lend at a common risk-free rate.
• Assets are infinitely divisible and publicly traded.
• Investors can trade during the same horizon.
• The market portfolio includes all risky assets and is mean–variance efficient.

These assumptions make the model analytically clear but are not literal descriptions of actual markets.

6. Multifactor Models

Evidence that variables other than market beta explain return differences motivated multifactor models. A general linear factor model is:

F_j represents a factor risk premium or factor return, β_ij measures asset i’s sensitivity to factor j, and ε_i is an asset-specific component. Depending on model design, factors may represent macroeconomic variables, statistical factors or traded portfolios.

Common factor dimensions include:

• Market excess return.
• Size, often measured through small-company minus large-company returns.
• Value, often measured through high book-to-market minus low book-to-market returns.
• Momentum, based on recent winner minus loser performance.
• Profitability and investment characteristics.
• Interest-rate, inflation, credit and commodity risks.

A factor model may improve risk attribution and portfolio construction, but a factor’s historical premium does not establish a causal or permanent reward. Factors can be crowded, expensive to trade, unstable across countries and vulnerable to data-mining.

7. Arbitrage Pricing Theory

APT, associated with Ross, derives expected return from the absence of arbitrage in a multifactor environment. Its factor form is:

where b_ij is asset i’s sensitivity to factor j and λ_j is the risk premium associated with factor j. APT does not require the market portfolio to be mean–variance efficient. Instead, it relies on a factor structure, sufficiently diversified portfolios and the principle that economically equivalent exposures should not have persistently different prices.

APT is attractive because it is less restrictive than CAPM, but it does not uniquely identify the number or identity of factors. Empirical implementation requires the researcher or manager to select factors, estimate exposures and determine whether pricing errors are economically and statistically meaningful.

CAPM and APT Compared

8. Empirical Limitations and Professional Interpretation

Estimation Error

Expected returns, variances and covariances are not known; they are estimated from historical or forward-looking data. Small changes in inputs can produce large changes in optimized weights, especially when assets are highly correlated. Sample covariance matrices may be unstable, and historical relationships may fail during crises.

Professional responses include shrinkage estimators, Bayesian methods, robust optimization, resampling, factor-based covariance models, constraints and scenario analysis. These techniques reduce sensitivity but do not eliminate uncertainty.

The Market Portfolio Problem

The true market portfolio should include equities, bonds, private businesses, real estate, human capital and other investable claims. In practice, analysts use proxies such as broad equity indices. A proxy can cause measurement error in beta and the market risk premium, particularly for assets with exposure to illiquid or non-traded risks.

Non-Normality and Tail Risk

Returns may exhibit skewness, excess kurtosis, volatility clustering and regime changes. Variance treats positive and negative deviations symmetrically and may understate the danger of rare losses. Portfolio managers should supplement mean–variance analysis with stress tests, drawdown measures, expected shortfall, liquidity analysis and scenario analysis.

Time-Varying Risk and Premiums

Betas, correlations and factor premiums can change over time. Leverage increases equity beta; changing monetary policy affects duration and credit exposures; and correlations often rise during market stress. A static estimate may therefore be unsuitable for dynamic risk management.

Anomalies and Joint Hypothesis Problems

Observed return patterns such as value, momentum, size and profitability challenge a single-factor CAPM. However, rejecting CAPM is not straightforward because tests jointly examine the model and the market portfolio used as its proxy. A poor proxy may make a correct conceptual model appear defective. Conversely, a statistically significant anomaly may reflect transaction costs, data mining, risk omitted from the model or compensation for illiquidity.

Performance Measurement

Alpha is model-dependent. In a single-factor model, alpha is the intercept after controlling for market beta. In a multifactor model, alpha is the residual after controlling for all included factors. A positive alpha may reflect omitted exposures, luck, leverage, stale pricing or incorrect benchmark selection rather than managerial skill.

9. Professional Application: Portfolio Construction and Governance

A disciplined portfolio process commonly follows these steps:

1. Define objectives, liabilities, horizon, liquidity needs and legal constraints.
2. Establish strategic asset-class exposures using capital-market assumptions.
3. Estimate expected returns, covariance and downside risks using multiple information sources.
4. Generate efficient or constrained portfolios.
5. Select a portfolio consistent with risk capacity and risk tolerance.
6. Implement with attention to transaction costs, taxes, market impact and liquidity.
7. Monitor factor exposures, concentration, tracking error and stress losses.
8. Rebalance according to a documented policy.
9. Evaluate performance against appropriate total-return and risk-adjusted benchmarks.

Governance is essential because optimization can create portfolios that are mathematically efficient but operationally unsuitable. Investment committees should document assumptions, approve permissible instruments, review model changes and distinguish forecast uncertainty from measured risk. Fiduciary decisions should consider beneficiaries’ objectives rather than merely maximizing a statistical ratio.

Technology supports portfolio analytics through automated data pipelines, factor dashboards, optimization engines and scenario systems. Artificial intelligence can assist in forecasting, document analysis and anomaly detection, but it introduces risks of overfitting, opaque decisions, unstable relationships, biased data and cyber compromise. Human oversight, validation, explainability and model-risk controls remain necessary.

10. Illustrative Case: Northbridge Pension Fund

Background

Northbridge Pension Fund is a fictional defined-benefit pension plan with a long-term liability horizon. Its existing portfolio is concentrated in domestic equities and government bonds. The investment committee is considering adding global equities, infrastructure and high-quality corporate bonds.

Problem

A mean–variance optimizer recommends a large allocation to infrastructure because its historical correlation with equities is low and its estimated return is attractive. However, the committee is concerned that the historical data contain smoothed private-market valuations and that infrastructure may become highly correlated with equities during a recession.

Analysis

The team first develops a strategic factor view rather than relying solely on asset labels. It identifies equity beta, duration, credit spread, inflation sensitivity, liquidity and currency exposure. It then estimates covariance using both historical returns and a factor model. Stress scenarios include an inflation shock, a sharp rise in interest rates, an equity-market decline and a liquidity freeze.

The optimizer is run under constraints: maximum allocation to infrastructure, minimum liquidity, limits on equity beta and a permitted tracking-error range. The resulting portfolio has a lower concentration in domestic equities and a more balanced exposure to equity, duration, inflation and liquidity factors.

Approach and Outcome

The committee does not select the mathematically highest-return portfolio. It selects a constrained portfolio whose expected return is modestly lower but whose liquidity, stress resilience and governance characteristics are more suitable. It adopts quarterly factor monitoring and an annual review of expected returns and covariance assumptions.

Learning Points

The case illustrates that diversification must be evaluated by economic risk factors, not merely by security count or asset-class labels. It also shows why optimization outputs require professional judgment, constraints, scenario analysis and governance. A low historical correlation is not a guarantee of future diversification.

Challenges, Ethical Issues and Best Practices

Asset-pricing models can influence capital allocation, executive incentives and client outcomes. Ethical practice requires transparent assumptions, fair treatment of clients, accurate disclosure of uncertainty and avoidance of benchmark manipulation. Managers should not present estimated alpha as guaranteed skill or use a convenient factor model solely to make performance appear superior.

Best practices include:

• Use economically justified factors and disclose their definitions.
• Report confidence intervals or ranges rather than false precision.
• Test results out of sample and across markets.
• Include transaction costs, taxes and liquidity constraints.
• Monitor model drift and structural breaks.
• Use multiple models rather than relying on a single forecast.
• Separate research discovery from final investment approval.
• Maintain independent model validation and audit trails.
• Align risk measures with the investor’s actual liabilities and loss capacity.

Chapter Summary

Portfolio theory shows that the risk of a portfolio depends on weights, individual volatility and covariance, not simply on the average risk of its holdings. Diversification reduces unsystematic risk when assets are imperfectly correlated. The efficient frontier contains portfolios offering the highest expected return for each level of risk, while investor utility determines which efficient portfolio is preferred.

With a risk-free asset, the capital allocation line identifies the trade-off between risk and return; the Sharpe ratio measures excess return per unit of total risk. CAPM prices systematic risk through beta and produces the security market line. Multifactor models and APT extend the analysis by allowing several sources of systematic risk, although factor selection and estimation remain difficult.

In practice, model outputs are affected by uncertain inputs, changing correlations, non-normal returns, liquidity, transaction costs, proxy-market problems and omitted risks. Professional investment management therefore combines quantitative models with constraints, scenario analysis, governance, ethical disclosure and informed judgment.

Key Terms

Arbitrage: A strategy seeking to exploit inconsistent prices for economically equivalent exposures.

Asset allocation: The distribution of capital among asset classes or risk exposures.

Beta: Sensitivity of an asset’s return to the market return.

Capital allocation line: Risk–return line formed by combining a risk-free asset with a risky portfolio.

Capital market line: CAL based on the market portfolio in the CAPM framework.

CAPM: Model linking expected return to risk-free return and systematic market beta.

Correlation: Standardized measure of co-movement between two returns.

Covariance: Unstandardized measure of joint movement between two returns.

Diversification: Combining imperfectly correlated assets to reduce avoidable risk.

Efficient frontier: Set of portfolios offering maximum expected return for each risk level.

Factor model: Model explaining returns through exposures to systematic factors.

Global minimum-variance portfolio: Feasible portfolio with the lowest variance.

Market risk premium: Expected market return minus the risk-free rate.

Multifactor model: Asset-pricing model using more than one systematic factor.

Portfolio variance: Weighted aggregate measure of portfolio return dispersion.

Security market line: CAPM relationship between expected return and beta.

Sharpe ratio: Expected excess return divided by standard deviation.

Systematic risk: Market-wide risk that cannot be eliminated through diversification.

Unsystematic risk: Asset-specific risk that can generally be diversified away.

Utility: Investor preference measure balancing expected return against risk.

Review Questions

Short Answer Questions

1. What is the difference between systematic and unsystematic risk?
2. Why does correlation matter for portfolio diversification?
3. Define the efficient frontier.
4. What does a beta of 1.3 indicate?
5. Distinguish the capital market line from the security market line.

Long Answer Questions

1. Explain the Markowitz mean–variance framework, including its assumptions and practical limitations.
2. Derive and interpret the two-asset portfolio variance equation.
3. Explain CAPM and critically evaluate its assumptions.
4. Compare CAPM, multifactor models and APT.
5. Discuss how estimation error and non-normal returns affect portfolio optimization.

Analytical / Application Questions

1. An investor allocates 50% to an asset with expected return 12% and 50% to an asset with expected return 8%. Calculate the portfolio’s expected return and explain why its risk cannot be determined without covariance information.
2. A portfolio has a beta of 0.8, the risk-free rate is 3% and the expected market return is 9%. Calculate its CAPM required return and interpret the result.
3. A portfolio optimizer recommends a highly concentrated allocation based on historical covariance estimates. Design a governance and stress-testing process to determine whether the recommendation should be implemented.

Multiple Choice Questions with Answers

Question 1

Which portfolio characteristic directly captures the interaction between two asset returns?

A. Expected return

B. Covariance

C. Risk-free rate

D. Alpha

Correct Answer: B. Covariance

Question 2

If two assets have correlation less than +1, combining them may reduce portfolio risk because their returns are

A. Identical

B. Perfectly aligned

C. Imperfectly correlated

D. Always negative

Correct Answer: C. Imperfectly correlated

Question 3

An asset has beta 1.2, a risk-free rate of 4% and an expected market return of 9%. What is its CAPM required return?

A. 6.0%

B. 9.0%

C. 10.0%

D. 14.8%

Correct Answer: C. 10.0%

Question 4

Which line relates expected return to beta?

A. Capital market line

B. Security market line

C. Budget line

D. Minimum-variance line

Correct Answer: B. Security market line

Question 5

The Sharpe ratio measures

A. Total return divided by beta

B. Expected excess return per unit of standard deviation

C. Alpha divided by covariance

D. Market return divided by inflation

Correct Answer: B. Expected excess return per unit of standard deviation

Question 6

In CAPM, the risk that investors are compensated for bearing is primarily

A. Unsystematic risk

B. Accounting risk

C. Systematic risk

D. Operational risk only

Correct Answer: C. Systematic risk

Question 7

Which statement best describes APT?

A. It requires the market portfolio to be mean–variance efficient

B. It prices assets using multiple systematic factors under a no-arbitrage logic

C. It assumes all assets have beta equal to one

D. It excludes factor risk premiums

Correct Answer: B. It prices assets using multiple systematic factors under a no-arbitrage logic

Question 8

A portfolio invests 60% in an asset returning 10% and 40% in an asset returning 5%. What is its expected return?

A. 6.0%

B. 7.0%

C. 8.0%

D. 9.0%

Correct Answer: C. 8.0%

Question 9

Which is a major practical limitation of unconstrained mean–variance optimization?

A. It never uses expected returns

B. It is insensitive to estimation error

C. Small input errors can produce unstable weights

D. It cannot calculate portfolio variance

Correct Answer: C. Small input errors can produce unstable weights

Question 10

A positive alpha in a multifactor model means that

A. The asset has zero systematic risk

B. The model intercept is positive after controlling for included factors

C. The asset must outperform in every period

D. The asset has a negative beta

Correct Answer: B. The model intercept is positive after controlling for included factors