Factor-based investing has grown significantly in the years since Eugene Fama and Kenneth French first published (1992) their groundbreaking research on the “three-factor model” to explain the return of stocks. Now, a growing number of investors view their portfolios as “collections of various risk-factor exposures,” including risks to particular asset classes and specific “styles,” such as value, size and momentum.
Investors reasonably expect to be rewarded for taking on these various types of risk. Understanding the source of returns has also made it difficult for investment managers to pass off factor-based returns as “alpha” – i.e., something that they (the manager) should be paid for having produced. But in order for investors to be sure they’re not overpaying for factor-based returns falsely portrayed as alpha, they must first be able to measure their exposures to the various risk factors – and this is trickier than one might expect.
In a recent white paper from AQR, Ronen Israel and Adrienne Ross consider the challenges associated with measuring factor exposures. The authors draw a distinction between academic and practitioner models, favoring the latter for being more practical to implement.
When conducting factor analysis, investors should ask themselves two questions:
- Exactly what factors am I using?
- Are they the same as those I’m getting in my portfolio?
The answers to those questions can significantly affect alpha and beta estimates. Factor design is also important and can lead to major discrepancies, too. When comparing alphas and betas across managers, investors should make sure they’re using factors being captured by both portfolios – otherwise, they risk overpaying for inappropriately attributed alpha.
For portfolios with more than one risk factor, multivariate statistical models are most appropriate. Mr. Israel and Ms. Ross caution investors to consider t-stats – measurements of statistical significance – and not just betas, especially when comparing portfolios with different volatilities.
Mr. Israel and Ms. Ross examine a hypothetical long-only stock portfolio designed to capture returns from value, momentum, and size style premia. The portfolio was designed with a 50/50 weight on value (book-to-price) and momentum (12-month trailing returns), entirely within the small-cap universe.
From January 1980 through December 2014, the hypothetical portfolio would have returned an annualized 13.8% above the return on cash.
Mr. Israel and Ms. Ross start with one factor – equity market risk – and build from there. First, a value factor is added (“HML”), and then momentum (“UMD”) and finally size (“SMB”). The HML, UMD, and SMB abbreviations refer to “common academic” definitions:
- HML (high-minus low) – Long/short value methodology; long high-value stocks/short low-value stocks;
- UMD (up minus down) – Long/short momentum methodology; long the stocks up the most/short the stocks down the most; and
- SMB (small minus big) – Long/short “size” strategy; long small stocks/short big stocks.
As you can see, when only considering a single factor (“the market”) in Model 1, it appeared that the portfolio generated nearly half of its returns from manager alpha. But as more factors are accounted for, it became clear that alpha-generation was actually much smaller. As an investor, you shouldn’t have to pay active-manager fees for factor exposures presented as alpha.