One of my favorite investment thinkers, Morgan Housel, recently asked a great question on Twitter:
My favorite investing question is, “What new thing would we learn if we had 10,000 years of apples-to-apples market data?”
— Morgan Housel (@morganhousel) September 27, 2016
Matt Busigin weighed in with a thoughtful response:
@morganhousel My guess: investment returns are extremely lumpy and are much lower on average than commonly thought.
— Matt Busigin (@mbusigin) September 27, 2016
This got me thinking about what we would learn about all the factors and anomalies researchers have identified in the past 40 years or so? In other words, could we please, please, please end the smart beta debate one and for all?
Harvey et al. have argued that our current pool of returns data has been horribly polluted by data mining. We, meaning investors, should therefore be more skeptical of research purporting to have found another anomaly.
Another 1000 years of data would tell us whether the factors we have seen in recent history, and yes 50 years of data is recent in this context, are persistent or simply random blips in time. If they are permanent we could delve deeper into whether these factors are a function of “investor behavior and being compensated for bearing risk.”
In a certain respect we can think of the boom in smart beta ETFs as a real-time experiment in anomaly exploitation. Investors will be able to watch not only how these funds perform (or not) over time. In addition we will be able to watch how investors react to the relative performance of these funds.
There is already a substantial debate about whether investors should try to time these various factors in order to avoid periods of overvaluation and popularity. We probably shouldn’t be surprised that investors are trying to out-game the smart beta game. Investors just can’t help themselves.
You can see in the graphic above, investors are even less patient with smart beta strategies than they are with traditional active managers. This is at odds with the idea of smart beta. From the State Street Global Advisors report:
Measuring the performance of smart beta strategies over a short time horizon can be hazardous, as their benefits are typically designed to be realized over the longer term.
To answer Housel’s initial query I think that there would be relatively few anomalies that end up being significant over time. Like Busigin I think many factors will pop up and then recede. With 10,000 years of data you think that we would be able to settle the smart beta debate once and for all. Unfortunately I am not sure that would make much difference to impatient investors who are only looking in a very short term rear view mirror.