Indices insights: Factor diversification - Upside potential at lower risk

Investors have recognized and deployed the benefits of diversification since the early days of capital markets. However, the concept was largely popularized after the late Harry Markowitz's ground-breaking paper, "Portfolio Selection," in 1952.¹ Using mathematical terms, Markowitz formalized the idea that diversification can reduce risk without changing expected portfolio returns, leading to the birth of modern financial economics.

Often referred to as the only 'free lunch' in investing, diversification allows investors to reduce portfolio risk without sacrificing returns. Simply put, a well-diversified portfolio achieves a higher expected return per unit of risk or a lower risk level for a given expected return.

The power of diversification applies across multiple asset classes—such as combining stocks and bonds—as well as individual securities—like diversifying among different industries or geographical regions. In this Indices Insights animation, we demonstrate empirically that diversification also applies when allocating towards multiple factor premiums. In other words, by diversifying across multiple factors, such as value and momentum, factor investors can lower their portfolio risk without compromising expected returns.

To examine the benefits of multi-factor allocation, we compared the historical performance and tracking error of individual factors, namely value, momentum, quality, and low beta, against a multi-factor portfolio with equal investments in all factors. The findings were clear: over the past 50 years, the multi-factor portfolio delivered similar returns as other factors but with a consistently lower overall tracking error. Diversification across factors enables investors to enjoy the benefits of factor investing while simultaneously reducing the overall deviation from the benchmark.

Methodology

Our analysis uses the monthly US factor returns from the Professor Kenneth French Data Library.² These factors are determined by double-sorting stocks based on market capitalization (five groups) and a second variable (five groups). These latter variables are book-to-price ratios (value); return over the prior twelve months excluding the last month (momentum); univariate market beta over the preceding five years (low beta); operating profitability (high profitability) or change in assets between the last two fiscal years (low investments).

We average the top quintiles across all size groups to derive the final factors. For instance, the value factor averages the five size-sorted portfolios with the highest book-to-market ratio. The quality factor is calculated by equally weighting the high profitability and low investment portfolios. Lastly, we derive the multi-factor portfolio by equally weighting the returns of the four individual factors each month (value, momentum, quality, low-beta).

To calculate the outperformance of each factor portfolio, we subtract the market portfolio returns from the factor returns. The outperformance displayed in the visual (y-axis) is the annualized geometric average three-year outperformance. The tracking error (x-axis) is calculated as the annualized standard deviation of outperformance. Both statistics, denoted in USD and ranging from December 1966 to May 2023, are updated annually and animated in between.