Prepared by Rafael Resendes, Daniel J. Obrycki, Derek Bergen, CFA and John Holt, CFA The Applied Finance Group, Ltd. – Internal Working Paper
Asset pricing model research has been dominated by the book to price (HML) factor following its introduction in the Fama French 3 Factor model in 1992. Over the initial study horizon of 1963 to 1991, book to price delivered performance characteristics that essentially absorbed the cross-sectional return information of numerous other factors, including leverage and earnings to price, on an ex-ante basis. Ongoing refinements to asset pricing models tend to build upon the foundation of this initial three factor research, highlighting the broad acceptance of these specific factors when seeking to explain cross-sectional returns. It has also become commonplace to refer to book to price simultaneously as a valuation factor (cheap vs. expensive, Asness, Frazzini, Israel and Moskowitz, 2015), a style factor (value vs. growth, Zhang, 2015, as well as Russell, S&P, and Morningstar style methodologies), and a leverage factor (high vs. low leverage, Penman, Richardson, and Tuna, 2005). In this paper, we provide compelling evidence that asset pricing models based on direct measures of intrinsic value, stewardship (which loosely aligns with style based on the reliance on external financing for early-stage growth stocks and the return of capital to shareholders for mature value stocks) and leverage offer substantial improvements in minimizing residual alpha compared to models that conflate book to price as a proxy for these competing themes.
As a research firm with expertise in measuring economic corporate performance and estimating intrinsic value, we have long been frustrated with the conflation of book to price as a valuation factor. Intrinsic value captures a current estimate of forecasted wealth creation over the remaining life of the business. Book value simply captures historical levels of paid-in capital and retained earnings, while further netting share repurchases. There is a substantial source of significant alpha available from the information gap between intrinsic value and book equity. Fama French later introduced independent sorting of growth and profitability rooted in the dividend discount model approach to valuation to motivate the expansion to a five-factor model. A dividend-based approach is too simplistic in estimating firm value compared to a robust valuation framework that captures economic measures of profitability, growth, competition, and risk; the unconditional preference of negative growth misrepresents the role of growth in creating shareholder value. In addition to this, the low growth and high profitability preferences derived by the dividend discount model’s emphasis of shareholder yield can be parsimoniously captured through a single factor based on financing cash flow yield that captures a firm’s stewardship of resources in the best interest of shareholders.
The initial three factor asset pricing model study, in addition to later refinements, has provided crucial insight towards a more comprehensive understanding of cross-sectional stock returns. The collective work of the academics and practitioners involved have contributed significantly to the knowledge base of the finance industry. With that in mind, we offer compelling evidence that calls for a paradigm shift in the construction of asset pricing models. As of June 2020, we are now equipped with 28.5 years of ex-post research on book to price, matching the initial study horizon from July 1963 to December 1991 applied in the original three factor paper. We can use this extended horizon to compare recent performance against the original in sample observations regarding the book to price factor. It is noteworthy that book to price no longer absorbs the role of other factors over our study horizon when factors are constructed in a manner that ensures comprehensive universe coverage. Using lagged data directly from Fama French data sources, we observe that the statistical nature of the unique contribution related to book to price has changed drastically since 1991. We further observe that a direct measure of leverage supersedes a book to price proxy over our study horizon.
We will present residual alpha studies that compare the cross-sectional explanatory power of commonly referenced asset pricing models against an alternative candidate model that incorporates a robust estimate of intrinsic value alongside direct measures of stewardship and leverage. This study ultimately confirms the application of size and leverage in asset pricing models, while providing a compelling framework to motivate the further inclusion of valuation and stewardship factors rooted in the complex, multidimensional relationships of profitability and growth. Evidence of this is clear in tests of factor redundancy when comparing direct measures of valuation, stewardship, and leverage against their correlated factor peers. This is further affirmed in studying the residual alphas and significance of a vast array of portfolio returns, factor returns, passive indices, and sector returns against each candidate asset pricing model.
- Book to price is an inadequate proxy for a robust measure of intrinsic value. Portfolios formed on the classification error between intrinsic value and book to price provide a significant unique source of alpha.
- The dividend discount model tautology that motivates Fama French’s transition from the three-factor model to the five-factor model lacks completeness as a comprehensive valuation framework. The preference towards negative growth as an independently sorted factor misrepresents the role of growth in creating shareholder value.
- A factor based on excess intrinsic value over book equity directly focuses on the dynamic relationship of profitability and growth in wealth creation, subsuming single year levels of profitability as an asset pricing factor.
- The dividend discount model framework that motivates independent preferences towards high profitability and low growth supports a financing yield factor that emphasizes flows to shareholders. This factor provides parsimonious benefits, rendering both profitability and growth redundant, while also capturing stewardship attributes by addressing the agency issue in corporate governance.
- Ex-post studies of book to price provide a stark contrast to characteristics delivered on ex-ante basis. On an out of sample basis, the book to price factor is subsumed by a direct measure of leverage.
- A valuation-based asset pricing model that incorporates excess intrinsic value, financing yield and leverage drastically improves the explanatory power of cross-sectional returns compared to other commonly studied asset pricing models.
- Active managers should incorporate intrinsic value in stock selection. Factor-based strategies are incomplete when valuation beta and stewardship beta are ignored. The negative alpha associated with passive investment further diverges from zero with a higher degree of significance against a valuation- based asset pricing model.