Research Affiliates: Measuring factor exposure, beta does not equal alpha

Research Affiliates: Measuring factor exposure, beta does not equal alpha

Factor Investing Risk Management
Vitali Kalesnik (Photo Archive Research Affiliates).jpg

By Vitali Kalesnik, Director of Research for Europe at Research Affiliates Global Advisors

Strategy factor exposure can be measured by factor characteristics (for example P/B ratio) or risk-based measures (for example beta). The latter aids in understanding risk, but is a poor predictor of alpha, making characteristics a superior measure.

Investors have two ways to measure factor exposures: characteristics and betas. The measure selected has very practical implications for the risk-and-return profile of the investor’s portfolio.

Characteristics, such as price-to-book ratio, measure a portfolio’s fundamental tilts by aggregating financial metrics. Betas measure the co-movement of a strategy’s return sources and predefined factor portfolios.

Better predictors, but factor risks

A characteristics-based approach follows fundamental investment principles supported by empirical evidence that factors provide persistent sources of return premia. A beta-based approach follows the theoretical argument that factor premia exist to compensate for exposure to factor risks. The two approaches are not mutually exclusive and can be complementary to the other.

Our research shows that characteristics directly measure fundamental factor exposures and thus are better predictors of future returns. Portfolios constructed using characteristics, however, introduce exposure to factor risks.

Betas, by measuring the co-movements of returns, offer investors valuable information to manage factor risks. Importantly, characteristics-based portfolios display significantly better returns compared to risk-based portfolios.

Benefits of beta-based factors

We construct three optimal unconstrained risky portfolios (characteristics only, betas only, and a combination of the two) to measure how investors can use beta-based factors to manage factor risks.

The portfolios incorporate the five factors in the Fama–French (2015) model: market, size, value, profitability, and investment. We analyze the portfolios over the period July 1968–September 2020.

The characteristics-based portfolio tilts significantly toward profitability (29%) and investment (54%), earning a Sharpe ratio of 1.02. This is an in-sample maximum Sharpe ratio and should not be viewed as suggesting investors could have outperformed the market in real-time.

The beta-based portfolio earns a Sharpe ratio of 0.36 (lower than the market’s Sharpe ratio of 0.41 over this period) because all factors but size have a low Sharpe ratio. The investor must take extreme positions (159% weight to size and −115% weight to the market) to reach the 0.36 Sharpe ratio.

The Sharpe ratio of the combined portfolio (1.48) contrasts vividly to the Sharpe ratios of the other portfolios. In this case, the optimal portfolio’s factor allocations (except for size) are in the opposite direction for the characteristics- versus beta-based factor. For example, the weights of the characteristics-based market and value factors are 32% and 19%, respectively, while the weights of the beta-based market and value factors are − 51% and − 30%, respectively.

High tracking errors

The high Sharpe ratio portfolios (characteristics-only and combined portfolios) have high tracking errors (TE). At 14.8% and 15.3%, respectively, these portfolios’ TEs are very similar to the market’s volatility over the analysis period. An investor who constructs a portfolio solely by maximizing its Sharpe ratio can create a portfolio entirely disconnected from the market!

How should an investor with a TE constraint solve this investment problem? We consider two TE-constrained portfolios: one with maximum 5% TE relative to the market and the other with maximum 5% TE relative to the value index (the characteristics-based long–short value factor). To satisfy the TE constraint, each portfolio could simply hold the respective factor, however, allocations to other factors can maximize their Sharpe ratios.

Again, we construct three portfolios — characteristics only, betas only, and a combination — for each constraint. Neither beta-based factor portfolio meets the 5% TE constraint. The two optimal constrained portfolios using both characteristics- and beta-based factors do, however, produce a meaningful benefit.

By shorting four of the beta-based factors, the portfolios gain exposure to the characteristics-based factors, earning Sharpe ratios of 0.78 versus the market and 0.91 versus the value index. In contrast, the optimal constrained characteristics-based factor portfolios earned Sharpe ratios of 0.52 and 0.59, respectively.

The bottom line is characteristics are good predictors of future return, and risk-based measures are lousy predictors of future return, but good predictors of future risk. Investors can use both to create portfolios that earn higher Sharpe ratios than portfolios constructed by either one alone. Beta-based measures are of particular help as a hedge to lower portfolio risk.

Disclaimer: Please refer to our disclosures