Key Concept 2: Multi-Factor Style Model

The investing community has long used “value” and “growth” to describe distinct investment “styles”.  Looking at the mutual fund universe, it is clear that managers, too, self-identify and align their products along style lines.  Value managers describe their process as one that involves trying to buy assets or cash flows at inexpensive prices by looking at “scaled price ratios” - (think P/E, B/M or other ratios of market value to variations on intrinsic accounting value).  Growth managers, on the other hand, often describe their process as one that involves determining which firms will grow (sales, earnings, cash flows, etc.) most quickly.

In the academic world, the past 20+ years have seen many journals filled with articles exploring the differences in returns between stocks (called the “cross-section of stock returns”). Seminal work by Fama and French, among others, noted stocks that are lower priced than their peers using a scaled price ratio tend to have higher future returns.  Using terms similar to practitioners, academics noted that these cheap “value” stocks tend to move together, as do stocks at the opposite end of the spectrum, the “growth” stocks (so called because valuation methods dictate that investors, rationally or not, believe they will experience higher future cash flows).  For academics, this indicated the presence of a common explanatory “value” factor.

Presumed opposites:

Shuffling the Gordon model so that there is a scaled price ratio (Price / Cash Flow) on the left shows a clear relationship with the discount rate and expected growth.

If not yet clear, the similarity between investor practice and academics seems to start and end with terms. For example, if we run an experiment and compare the returns that would have been generated by owning a portfolio of value stocks to one composed of growth stocks (using typical academic style definitions) we see that there is far less correlation between the two series than we observe when we compare the returns of portfolios of funds managed by investors that identify as value or growth investors.  In other words, value investors do not appear to simply be the opposite of growth investors. Accordingly, benchmarks that treat the two styles as opposites may not provide as rich a description of manager performance.

CRSP, in its index design, has sought to develop a series of more applicable benchmarks for industry, bridging the gap between industry practice and current academic thinking.

The CRSP multi-factor model:

CRSP believes that investment managers possess information beyond that contained in simple scaled price ratios and growth statistics.  The decisions these investors make are shaped by this unobservable information.  The importance of this philosophical consideration cannot be overstated: the CRSP U.S. Value and Growth Style Indexes are designed to be a cost-effective approximation of the process actual value and growth managers use to invest rather than an ad hoc style definition. 

A notable feature of the CRSP Indexes that allows for a better fit to industry behavior is the separation of value and growth into two distinct dimensions.

The value and growth dimensions are defined using multiple factors for each security.  The use of multiple factors follows current academic thought and manager behavior (managers look at multiple data points simultaneously when generating their investment ideas).  This ultimately allows for better estimation of the true, unobservable “value” or “growth” of a firm. 

Two distinct style dimensions:

A good proxy for the way investors view the world.

The factors used result from a combination of common industry practice and recent work in empirical finance.  CRSP is the first index provider to include investment rate and return on assets (“ROA”) as growth factors.  Empirical analysis shows that firms that invest more tend to grow faster, as do firms that are more profitable.  In addition, economic theory links both investment and ROA to expected stock returns.

CRSP and its faculty advisors place a premium on independent analysis.  As such, CRSP validated each factor individually and in concert with other factors using common econometric techniques. The five value factors, individually, have predictive power in the cross section of stock returns. A composite of the five value factors produces even more consistent results. Similarly, while the six growth factors individually predict growth, together they prove more effective.
 

The weights selected for each individual factor are the result of a process designed to select investments that behave like those value and growth managers would choose, while limiting portfolio turnover (where obvious transaction costs are incurred).  Using a combination of cluster analysis, regression and rank tests, CRSP assessed almost 2600 candidate models before determining a set of final factor weights. In contrast to many existing index providers, CRSP found that not all factors are created equal.  On the value side, earnings metrics have a plurality of the weight.  For growth, historical sales trends and analyst estimates proved most important.

To build the value and growth portfolios, stocks are ranked in cumulative market capitalization order first by a composite value score and then by a composite growth score.  This means that along the value dimension, stocks can be in either the top or bottom half of market capitalization ranked value (the same goes for growth).  Here, CRSP introduces two novel features:

  1. Value and growth scores are determined solely within the market cap segment evaluated, making scores statements of relative value and growth.
  2. CRSP simply averages the value and growth ranks scores to determine security placement, rather than coercing half of an index’s market cap to be value and the other growth. 

Investors should be able to immediately recognize how both decisions make the CRSP Indexes better measures of manager style performance.  For example, CRSP understands that a large cap value manager may only choose securities that look like value stocks within his or her universe; the use of a relative value score specific to the large cap universe better represents his or her opportunity set.  It also means that the same security may have a different style assignment in the CRSP mid cap or mega cap portfolios, which use their own relative scores (representing the opportunities restricted to those universes). The second decision, to resist coercing half the market into value and the other into growth, reflects an idea introduced earlier: value and growth styles, as practiced by investors, are not opposites; instead, they are best thought of as separate, though related, processes.