What Is Roll’s Critique of CAPM and Why Does It Matter?

The Capital Asset Pricing Model (CAPM) is widely used in finance to estimate expected returns based on risk. However, financial economist Richard Roll argued that CAPM has a fundamental flaw that makes it impossible to test properly. His critique questions the validity of one of the most influential models in asset pricing.

Main Proposition

Richard Roll’s critique of CAPM centers on the idea that the model’s predictions rely on an unobservable benchmark. CAPM assumes investors hold the “market portfolio,” which includes every investable asset weighted by market value. In reality, no one can construct or observe this portfolio precisely. The closest approximation, such as a broad stock index like the S&P 500, captures only publicly traded stocks, leaving out private equity, real estate, human capital, and other assets.

Because the true market portfolio is unknown, any test of CAPM is inherently flawed. Researchers use stock indices as proxies, but these are incomplete representations of the full investment universe. This means any empirical validation or rejection of CAPM is based on an approximation rather than the actual theoretical construct. As a result, conclusions drawn from these tests may reflect the limitations of the chosen benchmark rather than the model’s true accuracy.

The Market Portfolio Challenge

A major obstacle in applying CAPM is defining and measuring the true market portfolio. While the model assumes investors hold a fully diversified collection of all investable assets, constructing such a portfolio is nearly impossible. Financial markets include not just publicly traded stocks but also bonds, commodities, real estate, intellectual property, and human capital. Many of these assets lack reliable pricing data or are not easily tradable, making their inclusion impractical.

Even if a broad approximation were attempted, determining the appropriate weightings for each asset presents further challenges. Market values fluctuate constantly, and certain assets, such as private businesses or rare collectibles, do not have transparent or frequently updated valuations. Without a precise way to aggregate these diverse holdings, any constructed portfolio will always be an imperfect representation of the theoretical benchmark.

Consequences for Empirical Tests

Testing CAPM relies on statistical methods that measure the relationship between risk and return. Researchers typically use regression analysis to examine whether an asset’s expected return aligns with its beta, which represents its sensitivity to market movements. However, if the true market portfolio cannot be observed, beta itself becomes an estimate based on an incomplete benchmark.

This affects studies on investment performance. When evaluating mutual funds or hedge funds, analysts compare returns to a benchmark index to determine whether a manager has generated excess returns, known as alpha. If the benchmark does not fully capture all sources of systematic risk, the calculated alpha may be misleading. A fund might appear to outperform due to an improper comparison rather than genuine skill.

The issue extends to corporate finance, where firms use CAPM to estimate the cost of equity. This figure influences investment decisions, valuations, and capital budgeting. If the model is tested using an incomplete proxy for the market portfolio, the derived cost of capital may be biased. Even small miscalculations can lead to inefficient resource allocation or flawed assessments of risk-adjusted returns.

Relevance to Asset Pricing

Roll’s critique raises broader questions about how asset prices are determined. If the market portfolio cannot be observed, then any risk-return relationship derived from an incomplete benchmark may misrepresent the true drivers of asset prices. This affects the construction of risk premia, which are used to estimate expected returns across different asset classes. Investors relying on models that assume a well-defined market portfolio may be basing decisions on incomplete information, leading to mispricing and inefficiencies.

This issue is particularly relevant in factor-based investing, where asset pricing models extend CAPM by incorporating additional risk factors such as size, value, and momentum. If the benchmark used to derive these factors is an imperfect representation of the investable universe, the identified risk premia may be biased or incomplete. This raises concerns about whether widely accepted factors genuinely capture systematic risk or if they are artifacts of flawed empirical testing. Portfolio managers constructing factor-based strategies must recognize that their expected return assumptions depend on the accuracy of the underlying asset pricing framework.