In new research Ken French and Eugene Fama are expanding their famed “size, value and momentum” work on the US market to an international data sample.
It ues Bloomberg data, which has been cleaned by Dimensional Fund Advisors to which French is a consultant, and includes small cap stocks. It used data from 23 developed countries, in the time period November 1989 to September 2010 and looked at 38,851 unique firms.
The research looked at the excess returns for 25 portfolios formed on size and book to market, across four regions: North America, Europe, Japan and Asia ex-Japan.
“We know the value, momentum effect is much greater in small stocks in the US, now we can test that in the international market,” French said at a Dimensional investment symposium.
“We wanted to see if we can we used the model to explain the patterns. We found there was no size effect for this 20 year period globally,” French said.
The research concluded there were value premiums in all four regions except Japan. And momentum in all regions except Japan.
French said such a model can provide an important performance evaluation tool for investors, and that the Fama French three-factor model was already used by most academics and practitioners in the US.
“This can measure whether fees are good money spent, or can you do it cheaper somewhere else,” he said.
However the model’s application to the international sample was somewhat flawed as the “global factors weren’t able to explain the country portfolios.”
Addressing the fact the model “doesn’t necessarily work” French said investors could use it as an additional evaluation tool.
“By using the three factor model and a comparison of your manager to the index together it is more effective than using them separately,” he said.
While the model in the US is widely used, French is quick to point out it is “an empirically motivated asset pricing model, which is a nice way of saying there’s not a lot of theory behind it”.
Instead of building the model from a theoretical base, the two academics saw that there were patterns – such as a value effect – and built a model to capture that.
French said he and Fama tried to change one of the factors a few years ago, to better reflect what they were trying to measure, but there was revolt by their academic colleagues.
“We tried to change HML a couple of years ago, our colleagues rebelled and said you can’t it’s in all our computer codes.”
They wanted to change the high book to market and low book to market measures, to growth and value.
An interesting test of the model, is on the data of the well-respected Peter Lynch, portfolio manager of the Magellan fund at Fidelity fund from June 1977 to May 1990.
“When applying this model you learn very quickly about his tilting,” French said.