Inside ETFs: That makes sense; you'd want an easier benchmark. You wouldn't want anything that might stand in your way of showing that your strategy might be working.
Arnott: Exactly. And fundamental index, in some markets, shows a residual alpha even net of multivariate Fama-French factor attribution. And in some markets, neutral or slightly negative.
Does that mean it's not working? No. It means it's actually capturing the factor alphas pretty darn well without incurring a lot of trading cost to do so. So it's actually capturing those factor returns. Factor strategies don't do that. Ouch!
Now, one of the other things you made mention of in one of your questions was factor or smart-beta strategies. I liked your choice of words—factor or smart beta.
Inside ETFs: With so many products, and strategies being labeled “smart beta,” can we lump both factors and so-called smart beta together?
Arnott: I don't think of factor tilts as smart beta. Most factor-tilt strategies start with cap weight and then put on a factor tilt. Cap weight is not smart beta. It doesn't break the link with price. And if you put on a factor tilt, the factor tilt might be smart, it might be stupid. It might be smart, but poorly timed because it's currently expensive. But it's not smart beta; it doesn't break the link with price.
So I think the redefinition of smart beta to encompass practically everything robs it of any meaning. I mean, if smart beta encompasses everything, it means nothing.
Inside ETFs: Research Affiliates constantly says, “Keep the smart in smart beta.” What exactly do you mean by "the smart"?
Arnott: To our way of thinking, the original way the term was used was that strategies that break the link between the price of a stock and its weight in the portfolio have a structural alpha.
Structural alpha is that you're not going to automatically overweight overvalue and underweight undervalue companies. Now, you may make other mistakes. You may have negative alpha from time to time, but overvalued companies will not automatically be overweight just by dint of their price.
That's the Achilles' heel of cap weight. If a share price doubles, the weight of the portfolio doubles. Now, is that a good reason to double the weight in the portfolio? Of course not.
Now, the other thing that's interesting is that you can reverse-engineer the mathematics, and you can find that, tacitly, within capitalization weighting: If the price of a stock doubles and its weight in the portfolio doubles, then by definition, all else equal, you must be expecting a higher return after the price doubled than before it doubled. You must be expecting a higher return; otherwise, the weight wouldn't be higher.