Commentary: Using a forward-looking approach to factor investing boosts its benefits

New methods for defining and weighting factors can significantly enhance the benefits of factor investing. Factor-based investment strategies launched since the 1980s have generated unexpected results when the drivers of return change from one market cycle to the next. A notable example of this is the underperformance of book-to-price, or BP, since the beginning of the global financial crisis, a period during which asset-light companies have dominated returns. One of the great idioms of investing is the importance of "avoiding the rearview mirror." Applying this lesson to factor investing can provide meaningful insights to generate more consistent returns and withstand the changing market forces that shape equity returns.

As investors consider the choice of factors, they generally look to history as a guide. Countless papers have discussed the merits of blending quality, value and momentum. Theory supported by empirical evidence provides ample justification for their inclusion as factors for stock selection. Companies with rising fortunes led by high-quality management teams trading at reasonable multiples are attractive investments for both fundamental and factor-based managers.

The advantage of factor investing is its systematic approach for identifying stocks with these desirable characteristics across a broad opportunity set such as the MSCI ACWI Investable Market index. Where the process can break down, however, is in the definition of each factor. Quality, value and momentum are all multidimensional, so selecting a single dimension to measure any of these factors based on statistical significance may be appealing but can carry unwanted risks as market dynamics change.

Exhibit 1 displays the returns to book-to-price since 2001. What is remarkable is the stability of its return, first in a positive direction, followed by generally negative returns. Positive returns since the financial crisis have been limited to brief periods of economic recovery (e.g., 2009, 2016 and 2021). Are the negative returns an anomaly or are there fundamental reasons why BP has underperformed? This is an important question for both index providers and managers who employ this factor for stock selection.

An analysis of corporate earnings over this period clearly shows that they were largely driven by technologies that rely on human capital rather than hard assets to grow profits. As the cash flows of asset-heavy, high BP companies grew at a slower pace relative to the market, BP underperformed. Understanding the past is helpful, but the critical question is "what is a better measure of value that can adapt to changes in market forces?"

Fortunately, early proponents of value investing, Benjamin Graham and David Dodd, provide a solution. Forward-looking dividend discount rates consider multiple dimensions by measuring value relative to investor expectations for earnings and dividend growth.

Dividend discount models have been employed by managers for decades, but today's investment technology greatly enhances their value by providing solutions to overcome their primary weaknesses. However, as practitioners have learned, dividend discount models present multiple challenges, such as making difficult assumptions for companies without earnings and dividends, estimating longer-term growth and default rates, and covering a broad universe of stocks. Any errors in the return estimation process may in turn exacerbate errors in the portfolio construction process.

Fortunately, factor-based technologies solve many of these problems by enabling managers to adjust raw dividend discount rates for each company's risk characteristics. This can be accomplished through a regression-based "forward attribution" process whereby discount rates are restated based on a company's current risk exposures and the forward price of risk from the attribution analysis. This factor adjustment greatly reduces return estimation errors caused by uncertain stock level assumptions.

Exhibit 2 displays returns for factor adjusted dividend discount rates over the same period. By dynamically adjusting dividend discount rates for each company's level of risk, the rates can identify value more consistently as market conditions change. Cyclical underperformance will still occur during periods with strong megacap momentum or economic shocks that harm the economic prospects for value stocks. The cumulative return, however, is far more consistent with theory. A low BP stock will not be penalized if its growth expectations justify its price.

Similar examples may be developed for quality and momentum factors. By bringing in multiple dimensions, they also are better able to adapt to changing market conditions. Momentum may be broken down into its "good and bad" components, whereby price changes driven by fundamentals and earnings news are considered positive attributes while unexplained price changes are not. The recent drawdown from the Nov. 9, 2020, momentum crash following the COVID-19 vaccine announcement would have been largely avoided by separating momentum into its two components. Quality factors may be adjusted for forward-looking metrics such as long-term growth rates or changing ESG scores. New techniques such as machine learning and the incorporation of unstructured data may be employed to assist in the development of better forward-looking factor definitions. These techniques enable quantitative approaches to use information previously considered only by fundamental managers.

For investment strategies that employ multiple factors, weighting is another important decision. Many quantitative strategies do this by embracing long-term weights derived from historical returns, risks and correlations. Other managers have employed a more dynamic approach based on a best-fit analysis over rolling time periods. In either case, however, these approaches assume that the past is prologue to the future. In the case of BP, a dynamic approach would eventually downweight the factor, but only after the factor has underperformed for a period of time. By optimizing on the recent past, one must assume that factor efficacy changes slowly and in predictable ways. Both the quant crisis of 2007 and the recent dominance of megacap stocks suggest this assumption does not always hold true.

An alternative approach for weighting factors is to estimate their current risk premiums based on today's prices rather than historic returns. While auction markets provide useful information for real-time pricing of fixed-income factors, the same cannot be said for equity factors. Advanced statistical techniques such as Ridge regression, however, may be employed to disaggregate today's stock prices into their factor components. By doing this, managers can weight factors based on expected rather than historical returns. This approach avoids the problems associated with fitting models to the past or relying on uncertain beliefs about the future.

Will BP perform well in the future? Today, the answer may be yes, but under what conditions will its success continue and for how long? The valuations of asset-light companies may be contracting in today's market but aren't their futures still bright if their fundamentals stay strong? Market pricing provides valuable information to answer these important questions.

Innovation has always been critical for success in competitive fields such as investing. For investors who recognize the value of factor investing as a means to create portfolios with desirable return and risk characteristics, multidimensional forward-looking factors are an important tool to achieve consistent results as market conditions change. Today's analytical capabilities combined with the abundant availability of both financial and unstructured data sources present new opportunities for both factor selection and weighting. While investors will always make investment decisions against the backdrop of an uncertain future, innovation backed by investment experience provides solutions to avoid the rearview mirror.

This content represents the views of the author. It was submitted and edited under Pensions & Investments guidelines but is not a product of P&I's editorial team.