Nobel Prize Winner Eugene Fama: Nothing is perfectly anything so there has to be some amount of inefficiency in markets. It seems to be very difficult to find it and it’s very difficult to identify it among portfolio managers.
Interviewer: And would that be due to human behavior or the nature of humans in terms of how they are trying to process the information? Whether they process it correctly or not.
Fama: Well I don’t know. The basic reason is that prices and returns are so noisy. There is so much randomness in them that it is difficult to identify actual skill.
Interviewer: So from that you had the surge in research of stock market anomalies. How and what an anomaly is like a falsification of the EFH where people…
Fama: It is usually a falsification of the other part of the hypothesis which is the model for risk and return.
Interviewer: It implies with the studies that markets can be beaten if you use a particular trading strategy. For example a momentum strategy.
Fama: Ya, Momentum is the biggest example. Very difficult after that to find anything that is actually robust.
Interviewer: Ya, especially if you look at something like the January effect. There are so many different types of effects that they all seem to smooth out over time. There is almost a randomness in that in terms of returns.
Fama: There has been studies that come out that look at the so called anomalies and say how many of them are observed in data outside of the data used to discover them and then you find that lots of them really disappear when you look at them in new data. But that is an area that still needs to be flushed out. See there is a problem in academics where everyone wants to publish new papers. That’s the way they advance and get tenure and get higher salaries. They also get notice on wall street for doing it. So there is an incentive to dredge the data and come up with things that will be attention grabbing. But won’t necessarily be there in new data and active places for investment strategies.
Interviewer: I find that based on readings of the studies myself even though I have not performed any of the research independently but looking at it we see trend lines where there is a disappearance of some of these anomalies and not the momentum effect. Did you find it entertaining or did you find it difficult at the time or did you just blank it when academics were putting this research out to try and make a name for themselves given the popularity of the topic at the time or did you find it a personal attack?
Fama: No, I don’t really have any vested interest in it. The first person that I know of that discovered momentum was Cliff Asness [I think that is who he said]. He was one of my PhD students. He is an investment manager at this point. But he came to me and showed it to me and thought I would be upset by it but I said “This is the data and this is what it says so publish it.” But he had already been scooped at that point because Jegadeesh and Titman had the same results like 6 months earlier. So Asness never got credit for it.
Interviewer: Ah that’s unfortunate.
Fama: He is very rich now so it doesn’t really matter.
Interviewer: Wow that’s what counts then actually
Fama: No, not really.
Interviewer: It doesn’t really matter
Fama: But he is quit successful actually so he doesn’t need that.
Interviewer: Ya, but even Jegadeesh and Titman themselves got criticized for perhaps data mining their work in order to have the data fit their model.
Fama: Well right. So robustness is the name of the game and their results were scrutinized by many people thereafter and applied to different time periods and different markets and they showed up pretty well so that is one anomaly that seems to be robust. And it is what it is. You have to live with it. It contradicts market efficiency I think but that’s the name of the game. All scientific theories have anomalies otherwise they are not theories. They are reality.
Interviewer: That’s true actually. That’s the beauty of economic science.
Fama: That’s the beauty of all science. All science is you purpose models. You test them and you come up with some stuff that say “this works pretty well” and you come up with other stuff that says ”this doesn’t work well” on this particular so called anomaly and so you either tweak the model to incorporate that or you just accept it as one of the shortcomings of the model. That’s why you call them models.