Fama, in an interview no longer available on Dimensional's website, said:
Interviewer: Some people... assume that you would recommend most investors have a big helping of small and value stocks in their portfolios. Is that a fair representation of your views?
Fama: Um, no. (Laughs) Basically this a risk story the way we tell it, so there is no optimal portfolio. The way I like to talk about it when I give presentations for DFA or other people is, in every asset pricing model, the market portfolio is always an efficient portfolio. It's always a relevant portfolio for an investor to hold. And investors can decide to tilt away from that based on their personal tastes. But that's what it amounts to. You can decide to tilt toward more value or smaller size based on your tastes for these dimensions of risk. But you needn't do it. You could also decide to go the other way. You could look at the premiums and say, no, I think I like the growth stocks better. Then, as long as you get a diversified portfolio of them, I can't argue with that either. So there's a whole multi-dimensional continuum here of efficient portfolios that anybody can decide to buy that I can't quarrel with. And I have no recommendations about because I think it's totally a matter of taste. If you eat oranges and I eat apples I can't really quarrel very much with that.
My interpretation of Fama, with several jumps and "read somewheres" added, is that Fama believes that, yes, there is a value premium but that it is explained entirely
by higher risk.
By "dimensions of risk" I assume that he means things like skew or kurtosis or something. For example, I've "read somewhere" that value stocks have negative skew--they have small but consistent outperformance most of the time, but when they crash they crash badly. And that growth stocks have positive skew--they have small but consistent underperformance most of the time, but once in a while they hit the jackpot.
I've also "read somewhere," probably one of Larry Swedroe's columns, that many investors prefer positive skew.
The traditional factor maven point of view is that this is irrational, that they are overpaying, and that the rational investor exploits this by investing in value. But this is only true if you assume that there's something objectively wrong with preferring positive skew.
And Fama says "You can decide to tilt toward more value or smaller size based on your tastes... You could also... say, no, I think I like the growth stocks better... I think it's totally a matter of taste. If you eat oranges and I eat apples I can't really quarrel very much with that."
So the rational
approach would seem to be to devise some kind of second-order risk tolerance measurement instrument, a risk tolerance questionnaire that assesses, not your tolerance for risk, but your preference for positive or negative skew. Or for whatever Fama means by "dimensions of risk." Let's say, a questionnaire that assesses your preference for growth or for value. OK, I don't think it's possible, but that would
be the rational approach.
And the conclusion is that you should figure out what you prefer
, and only tilt toward value, not because of anything you believe about risk or return, but because you prefer the "dimensions of risk" that value has.
As nearly as I can judge my personal preferences, I
don't want to touch negative skew with a ten-foot pole, and if that's what value is, it's not for me.
Throw into this a bias problem. We are too apt to use what I call the "athletic model of performance assessment," that is to say counts or percentages of wins and losses. An athlete gets the same gold medal whether she wins by a minute or a millisecond. This gives a competitive advantage to managers who can create negative skew by their strategy. If, like Bill Miller, you can beat the S&P 500 fifteen years in a row, you will become personally rich, even if negative skew results in losing all of the outperformance back again in just three years.
It should be noted, by the way, that the opposite of "lottery tickets" (positive skew) is "gambling systems." These are all forms of the martingale, usually obfuscated, which in essence consists of doubling your bet every time you lose until you eventually win. Such a strategy produces a steady, regular stream of small wins that make it appear that the system is working. It produces rare catastrophic losses (you go bankrupt before you can double the bet again), but because they are rare you can brush them off as "nobody could possibly have foreseen." In gambling, of course, the long-term expected return is negative. In investing, it is (hopefully) positive. But you can still a pattern of "high chance of small outperformance, misjudged as consistent outperformance, interrupted by small catastrophic losses, that can be explained away as rare anomalies that shouldn't count."
Negative skew. Ugh. I hatesssss it, I hatesssss it I doessss. If you like it, you are welcome to your risk premium for taking it.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.