Value stocks have long been shown to outperform growth or glamour stocks over long investment periods. Small stocks also seem to outperform large stocks. More recently, momentum stocks have attracted attention.
But are these effects global, or limited to the U.S. marketplace? Recently Chicago Booth School’s Eugene Fama and Dartmouth Tuck School’s Kenneth French, in a paper called “Size, Value, and Momentum in International Stock Returns” applied their three-factor analysis to four global regions, adding a momentum component.
In the three-factor model, beta is the market return (the CAPM return of the market over the risk-free rate) while value and size represent additional sources of return. Fama and French use a truncated period, 1990 to 2010, in which to examine global stocks, in order to dampen a bias towards big stocks that would ensue from using a longer time frame where price data for small stocks is lacking.
They find that neither a three-factor, nor a four-factor model is adequate to the pattern of returns and instead propose a six-factor model, one that splits value and momentum according to size. That seems a better fit to the data – except for micro-caps.
To provide a sufficient sample size, they selected 23 countries and divided markets into four groupings: North America, Europe, Japan and Asia-Pacific (Australia, New Zealand, Singapore and Hong Kong).
They find no evidence of a small-cap premium. But the value phenomenon persists, even in Japan where the market beta is negative. Contradicting past research, the premium for small-cap value stocks is higher than the large-cap value premium – except in Japan, where they are the same.
Similarly, momentum persists, and is stronger for small-caps – except in Japan.
But the absence of a size premium creates anomalies for the three-factor model, Fama and French note. Small-cap growth stocks underperform large-cap growth stocks – except in Japan. That leads them to reject the three-factor model as a global model.
Because momentum is stronger for small-caps, that also leads them to reject adding momentum to create a global four-factor model.
A second part of their paper is to determine whether global factors are at work, or local ones. They find little support for a global asset pricing model. Instead, on a global basis, large-cap growth is what provides the excess returns in their model, contrary to observed patterns.
That leads to the introduction of a six-factor model, where the value component is split between small and large-cap stocks, and the same with the momentum component. That’s “overkill for explaining size-B/M [book value/market value]returns; splitting WML [winners minus losers] into its small and big components doesn’t add much,” they admit. “Likewise, splitting HML [high minus low] into its small and big components doesn’t add much in the later tests on size-momentum returns. We focus on [a six-factor model] to have a single model that potentially provides a unified story for value and momentum returns.”
There are models, and there are models. Thankfully they’re on the finance track rather than the fashion runway.
A four-factor model would suffice, they note, were it not for the performance of micro-cap stocks. But not in all regions. Still, a six-factor model fails to explain return patterns in the Asia-Pacific region. And then, as usual, there’s Japan. “We can’t reject global CAPM, three-factor, four-factor, or six-factor pricing for Japan, but in economic terms, the global models fail badly for Japan.”
So much for global models. Do regional models have more explanatory power?
A three-factor model does for Japan, where there is no momentum or size effect. In Europe, a six-factor model provides little advantage over a three-factor model. There are no distinguishing models in Asia Pacific. And in North America, microcaps befuddle explanation.
“In sum, local three-factor models are passable stories for average returns on local size-B/M portfolios in Japan and Europe,” Fama and French write. “Subject to a caveat about low precision, the local six-factor model is the choice for Asia- acific. Microcaps do not have a special role in the local models for Japan, Europe, and Asia Pacific, but microcaps are important in the rejections of all local models for North America. Microcaps aside, the local four-factor model does a reasonable job explaining average returns on the North American size-B/M portfolios, and it is tempting to conclude that North America is an important contributor to the similar conclusion from the earlier tests of global models on global size-B/M portfolios.”
Takeaways: local factors, not global ones, explain regional stock outperformance. And size doesn’t matter. Value does. So do microcaps – though those are probably too small for institutional investors to get a decent weight in.