Editor’s Introduction for 2023 Special Issue on Factor Investing

INTRODUCTION

This eighth special issue on factor investing contains 13 articles, 11 are contributed by practitioners, one by a practitioner/academic team, and one academic. The issue begins with perspective pieces by three portfolio managers—David Blitz of Robeco (“Factor Investing: The Best Is Yet to Come”), Jennifer Bender of State Street Global Advisors (“A Tour of the Factor Funhouse”), and Harindra de Silva of Allspring Global Investments (“On Factor Purity in Investment Portfolios”).

Style factors—size, value, quality, momentum, and low volatility—exhibit time-varying trends. Andrew Ang in his article “Trends and Cycles of Style Factors in the 20th and 21st Centuries” employs various time-series models to document trends and cycles of factors, characterizing long-run and short-run components of style factor returns. He finds that in the 21st century, value, momentum, and low volatility factors exhibit lower trends compared to the 20th century while the trends of size and quality have increased. Focusing on value, Ang reports the underperformance of this factor from 2017 to 2022. According to the author, this finding is attributable to both a downturn in its long-term trend coupled with a negative cyclical component, making it much more severe than in the late 1990s where the long-term trend was positive and only the cyclical component was negative. The value drawdown can be divided into three parts: a late cycle from 2017 to 2019, the COVID-19 pandemic year of 2020 where value’s losses accelerated, and then a recovery starting in late 2020 to 2022.

There is a considerable body of literature that provides both theoretical and empirical support for factor investing. This literature suggests that factor-based strategies offer valuable diversification to traditional markets that is not dependent on market conditions or macroeconomic environments. Despite this body of literature, there remains confusion and debate regarding factor investing. Although some of the rhetoric and myths have been around since factor investing was first proposed, others have arisen as a result of the performance of factor investing from 2018 to 2020 and the subsequent turnaround. In “Fact, Fiction, and Factor Investing,” Michele Aghassi, Cliff Asness, Charles Fattouche, and Tobias J. Moskowitz analyze 10 claims about factor investing. Some of these claims are timeless. Others focus on specific concerns that have emerged recently. The main factors that the authors analyze are value, momentum, carry, and defensive/quality because they pervade the literature.

As a result of the mediocre performance during the past decade, the role of macro factor investing has come into question. Confirming this decline in profitability of macro factor investing, Chenfei Ma, Eddie Cheng, and Wai Lee investigate the importance and relevance of macro factors in their article “The Lost Decade: Have Macro Factor Risk Premia Become Irrelevant?” Using three different approaches, the authors analyze the power of macro factors for explaining asset risks and cross-sectional return variations. No evidence of the declining importance of macro factors over time is found. Ma, Cheng, and Lee discuss a few possible explanations for the apparently unreliable risk premia associated with these factors in the recent decade and suggest that portfolio managers may require a more dynamic macro factor investment approach in the future.

Rob Arnott, Vitali Kalesnik, and Lillian Wu explore several often underestimated risks associated with multifactor investing, and techniques to improve the risk-adjusted returns of individual factors and factor portfolios in their article “Mitigating the Hidden Risks of Factor Investing.” This research inspired them to develop and introduce a two-step volatility management method that adjusts the length of the estimation window to scale factor returns. They find that their method is effective in improving risk-adjusted returns as well as the trade-off between performance improvement and turnover characteristics. The authors argue that the two-step method they propose coupled with an optimization technique for capturing both volatility and correlation information results in (1) improved risk-adjusted performance, (2) lower volatility of volatility, and (3) improved kurtosis and drawdown characteristics.

In “Factor Information Decay: A Global Study,” Emlyn Flint and Rademeyer Vermaak investigate how the factor exposures of equity factor strategies decay over time. To analyze this empirical issue, they study five well-known factors—value, momentum, quality, investment, and low volatility—across 12 developed and emerging markets over the last 20 years. Flint and Vermaak calculate empirical factor exposure distributions for a selection of factor strategies and study how these distributions evolve over a 36-month holding period. To quantify how fast the target factor exposures decay, they introduce a factor half-life metric. The authors find that (1) although the strength of decay differs greatly per factor, over time the target factor exposure distributions of pure factor portfolios tend to exhibit smooth decay profiles, (2) some of the distributions display significant dispersion and so target factor mismatch risk can be sizable even at short holding periods, (3) although non-target factor exposures of pure portfolios are generally centered around zero, these exposures can display fairly large dispersion in the medium to long term, and (4) the pure value, low volatility, and quality factors are generally quite similar in terms of overall distribution shape and evolution over time, whereas the momentum and investment factors decay significantly faster on average and show markedly different distributional shapes compared with the other three factors.

In “When Do and Which Fama–French Factors Explain Industry Returns?” Nikiforos T. Laopodis investigates for each decade since the 1960s and by industry the statistical significance of the five Fama–French factors and several macroeconomic variables. He finds that not all factors were significant in each decade and for each industry. Moreover, the author reports that when the Fama–French factors were present in the regressions, the macroeconomic variables often lost their statistical significance for the industries studied in each decade. When constructing factors out of the macroeconomics variables, Laopodis reports that they were significant for many industries from the 1970s through the 1990s and part of the 2010s. These findings have implications for portfolio managers when selecting industries for investments based on factor models. Before designing an investment strategy via multi-factor modeling, portfolio managers need to understand the fundamental changes of each industry since the 1960s and 1970s.

An ongoing debate that is the cornerstone of numerous equity allocation decisions is the relative performance of value with respect to growth. A framework for quantifying the extent of over or undervaluation of value relative to growth and for identifying the key factors driving performance is provided by Olga Lepigina, Kevin J. DiCiurcio, and Ian Kresnak in their article “A Fair Value Approach to Forecasting Value versus Growth Returns.” In their analysis, the authors introduce a method for estimating a “fair value” of value versus growth relative valuation by putting valuation drivers—inflation, Treasury yield, equity volatility, and growth of corporate profits—into a time-series model (i.e., a vector error-correction model) and then forecast future value and growth returns. The authors find that their methodology offers a significant improvement over the use of historical average as a future return estimation based on their out-of-sample value versus growth historical return forecast. They conclude that the methodology that they propose gives portfolio managers an alternative robust solution to forecasting value versus growth returns that can be further applied to asset allocation decisions and risk management.

Empirical evidence suggests that stock returns are not linear in size, yet the Fama-French model includes only a single size factor, measured by the difference between small-cap returns and large-cap returns. In “Improving Equity Fund Alpha Estimates with a Second Size Factor,” Nanqing Dong, Luka Jankovic, Anne Stewart, and Scott Stewart confirm the nonlinearity of the size-return relationship and go on to show how to better explain equity mutual fund returns. They do so by breaking up the size factor in the Fama-French factor model into two size factors: (1) small-cap minus mid-cap returns and (2) mid-cap minus large-cap returns. By using two size factors, the authors demonstrate improved explanatory power and produce different alpha estimates, equity fund rankings, and evidence of manager skill.

There are many challenges faced by portfolio managers in seeking to implement factor investing in credit markets. In “Putting Credit Factor Investing into Practice,” Hendrik Kaufmann, Philip Messow, and Frederik Wisser identify two of the major challenges and how portfolio managers can deal with them. The first challenge is that the investment universe is reduced by non-tradable assets. The authors explain how to avoid investing in these bonds by constructing a tradability measure based on past transaction data and recommend a realistic strategy to implement in one that targets currently tradable bonds. The second challenge faced by portfolio managers is that typically their performance when implementing a factor strategy is measured against a benchmark. By limiting the deviation between the portfolio and the benchmark in key dimensions, the authors show how a fair performance comparison can be achieved. By doing so, not only the bonds with the best signals can be included in the portfolio. By reducing the universe of acceptable bonds and imposing other restrictions and higher transaction costs than on the equity side, the authors show a substantial performance decrease in the four strategies. However, despite this reduction in performance, Kaufmann, Messow, and Wisser show that factor investing in credit is still a successful strategy if it is approached with realistic expectations and common pitfalls are avoided.

Vishv Jeet and Amit Partani in their article “Brinson-Style Attribution over Continuous Factors” offer new insights into the relationship between two existing performance attributions methods—factor-based attribution and Brinson-style attribution. More specifically, they introduce a simple or intuitive methodology to perform a Brinson-style attribution over a set of factors that may have continuous exposures to the assets. They show how the Brinson-style method is a special case of factor-based attribution when all factors are binary, such as industry, sectors, or countries. The two methods are identical for a specific benchmark (or set of factor returns). The implication is that the allocation effect is the same as factor contribution and selection effect is the same as specific contributions.