This blog is the first in a four-part series on factors of return.
Research has identified approximately 316 new factors in the academic literature from 1964 through 2013(1), so how can we be sure that the factors above are the ones we should be using to construct our portfolios?
To review, factors are independent variables in an equation, or model, which help to explain asset prices. In the seminal Fama-French three factor asset pricing model, the three variables, or factors, are the market premium, the small premium and the value premium. These premiums are represented in the model as follows:
* Market Premium — the return on the market over the risk-free rate
* Value Premium — the return of stocks with low relative price over stocks with high relative price, determined by the book-to-market ratio on stocks
* Small Premium — the return on stocks of smaller companies over stocks of larger companies, determined by the market capitalization of stocks
Improving our explanation for asset prices of stocks and bonds is the motivation for continued research in this area.
To be useful, factors must have explanatory power. But understanding whether a factor has explanatory power is more than just a t-stat in a difficult-to-read table full of regression coefficients and test statistics. Advisors can think about the explanatory power of a factor by using the framework employed by Dimensional Fund Advisors.
Factors must be:
Sensible — Sensible factors are those factors that have a sensible explanation for why they explain expected returns, whether it is risk (market, small, value, etc.) or a measure of a characteristic of the cross-section of companies (profitability).
Persistent — The factor must be able to explain returns throughout time. A factor will have limited explanatory power if it works during select time periods.
Pervasive — The factor must be able to explain returns across markets. A factor will have limited explanatory power if it works only in certain markets.
Robust — A factor must be robust to alternative specifications. For example, when evaluating the value factor, it must persist using a number of different metrics besides book to market (examples would include price to earnings, price to sales, price to cash flows, etc.).
Cost-Effective — A factor must also be cost-effective to implement. We’ve seen how costs negatively affect performance; costs also detract from the expected benefits of implementing a factor.
Evaluating whether factors meet the criteria above requires performing initial research, formulating a hypothesis, and designing and performing tests to see if the hypothesis was correct. This is the essence of economic science. The economic science underlying these factors has guided us towards these factors when constructing our model portfolios and provides confidence during time periods when these factors are negative.
Over three future blog posts, I will be taking a closer look at each of the small, value and market premiums to shed some light on how these factors have performed historically.