Morningstar Magazine - December/January 2015 - 73

Now, having said that, we have another paper in
which we show that once you get beyond
two dimensions of expected return, the third adds
very little.7 No matter how you stack them,
it doesn't matter where you start. So, if you
start with size and value, say, and you add
profitability, you don't get much. If you start with
profitability and value, size doesn't get much.
The problem is that these things tend be correlated.
For example, momentum and value-well,
value stocks tend to be negative momentum
stocks. So, if you want to get momentum, you
have to give up a lot of value to do it. That's
the general thing that happens when you try
to combine these things. To get more of something,
you have to give up some of the other stuff that
you're focusing on.

Beta has always been just the market
betas, and all the other stuff is not really
about beta.
Eugene Fama

Now why? That's a different story. We're having
much less success telling stories about why
there are different dimensions of risk that are
priced differently. All we know is that they seem to
be there over long periods of time and across
many markets.
Johnson: As your own thinking has evolved, you've
gone from the three-factor model to now
drafting a paper on a five-factor model.6 Now, we're
seeing this proliferation of factors. Are you
concerned that some of these may not hold up-
that they haven't been vetted enough?

Fama: That's always a concern. One of my
students is writing a paper trying to enumerate

the possible number of factors, and I think
he quit at 150.
Johnson: That's a big model.

Fama: Right. But the five-factor model that we
developed is basically just an extension
of the three-factor model. The three-factor model
focuses on the book to market ratio. But we've
long known that that's not the pure measure
of expected return, that the market price reflects
the discount rate that goes into the pricing of cash
flows. But it also reflects the cash flow. So,
basically what the five-factor model does is that it
tries to control for the cash flows, so that you
get a more precise story about the expected return.

Johnson: What about some of the factors that we've
seen emerge outside of the context of your work?
We see things now like quality and low volatility. What
are your thoughts on those?

Fama: I can explain all of those with the five-factor
model, frankly. They're just manifestations
of this. With quality, you're talking about a tilt
toward profitable firms that don't invest a
lot. The five-factor model said, "Yeah, they will
do better, given their book to market ratio."
You have to control for all three of these things.
So, I can explain why the relation between
average return and beta is flat. The low-beta
stocks tend to be profitable stocks that don't
invest very much and have a slight tilt toward
value. The high-beta stocks tend to be
growth stocks-a lot of small-growth stocks
in there-that are unprofitable and invest
like crazy.

There's one group that appears everywhere in
every sort. It's the same group of stocks. They're
small, unprofitable growth stocks that invest
a lot. No matter which sort you do, if they pop up
in a portfolio, their asset-pricing model just
grinds to a halt. You cannot explain why their
return is so low. And that turns up in lots of these
so-called smart beta sorts.

6 Fama, Eugene F. and Kenneth R. French, "A Five-Factor Asset Pricing Model," September 2014, working paper.
7 Fama is referring to work he's done with DFA, whereby they've recently introduced "direct profitability" into the mix with size and value. It adds a bit of incremental expected return, but not
much, and tamps down portfolio volatility a bit-resulting in improved Sharpe ratios.

global.morningstar.com/Morningstarmagazine

http://global.morningstar.com/Morningstarmagazine

Morningstar Magazine - December/January 2015

Table of Contents for the Digital Edition of Morningstar Magazine - December/January 2015