Your research shows that despite the CAPM’s many flaws, it is still the favored model among investors. How do you explain this paradox?
“The flaws of the CAPM, often called ‘anomalies’, have to do with the model’s ability to predict the average returns of characteristic sorted portfolios. The response of academics to these failures has often been to take the anomaly itself, and add it to the model as a new risk factor and claim victory.”
“My joint research with Jonathan Berk, from Stanford University, argues that this approach is very similar to the way early astronomers ‘fixed’ the earth-centered model of the universe: add a so-called ‘epicycle’ to the system every time you see a planetary movement you do not understand. In the end, so many epicycles were added that when Copernicus proposed the sun-centered model and erroneously assumed the planets moved in circles instead of ellipses, the earth-centered epicycle model predicted the movement of the planets more accurately than the sun-centered model. Obviously, that did not make the epicycle model right. This is a cautionary tale of what happens when you are too dismissive of new theories.”
“Furthermore, theories should be tested on data they were not designed to explain. In this case, data other than average returns. We use mutual fund flows instead. We find that even though the CAPM is certainly not perfect, it is still the best model we have. Whether or not a better model exists that both explains data on investor choices, that is mutual fund flows and average returns, is an exciting question for future research. I think that in the past few decades, we have seen many epicycles and little real progress. Hopefully somebody will discover the equivalent of a sun centered model. Or perhaps the CAPM is already that model, and all we have to do is swap circles for ellipses.”
Over the last few decades, evidence has been found for a large number of anomalies. What should investors do about these anomalies?
“First, investors should establish whether the anomaly represents a robust phenomenon or a statistical quirk. We can make that assessment only after observing out-of-sample data across different time periods, different regions and different asset classes. Second, if we decide that the anomaly is robust, we should try to establish whether or not it compensates for risk. My research on mutual fund flows suggests that several robustly documented anomalies do not compensate for risk in the eyes of investors. That is, they view these anomalies as alpha, rather than beta.”
You document a mismatch between the size of some anomalies and their impact on the real economy. What are the consequences?
“The question we focus on is whether the potential mispricing of stocks is harmful to the economy as a whole. After all, stock prices are an important signal for firm managers to base their real investment decisions on. If those prices are biased, they can distort the economy. My research with Christian Opp, at the Wharton School, suggests that we should be concerned with persistent anomalies such as the value premium or the profitability premium. Transient anomalies, such as momentum, while potentially interesting for investors, are not particularly important for the economy.”
“That’s because the price distortions related to momentum are too short lived for firm managers to respond to and thus these price distortions have little influence on real investment. We argue that if financial intermediaries, such as active mutual fund managers, can help eliminate persistent anomalies, they can potentially add a lot of value to the economy. This holds true even if they do not beat passive benchmarks. Think of it this way: if everybody invests passively, who will make sure that the price is right?”
In a competitive world, it is very hard for an ordinary investor to consistently achieve positive abnormal returns.
Active managers are often criticized for high fees and poor performance. Do you agree?
“I partly addressed this in my previous answer. More importantly, perhaps, the key to understanding active mutual funds is to frame the issue as an economic equilibrium, just as we would with stocks. Suppose an active mutual fund manager runs a 100-million-dollar fund, and has delivered a 2% extra return to investors. What do you think will happen? Well, many investors will jump at this amazing opportunity.”
“As the fund receives more inflows, the manager will find it harder and harder to identify good investment opportunities for this new inflow. As a result, the fund’s average performance per dollar invested will decline. When will the inflow stop? When the net alpha is driven down to zero. Consequently, in a competitive world, it is very hard for an ordinary investor to consistently achieve positive abnormal returns. This explains why the net alpha does not persist – the inflow ensures that it does not – and perhaps even more importantly, why the net alpha is not an accurate measure of the manager’s skill.”
“So, in summary, where people fiercely compete with each other for something they all want, in this case a skilled manager, they usually go back home with nothing. Only if you know something about a manager that other investors do not know, meaning you have a competitive advantage, can you achieve abnormal returns.”
So can at least some active managers can be considered skilled?
“Yes, absolutely. However, to establish the level of this skill, we need to measure it the right way. That is, we need to properly take into account the equilibrium I mentioned previously. When we measure skill based on the amount of money a manager makes or loses in markets for asset owners and for himself – which is the fee he charges plus the positive or negative alpha that he achieves, all multiplied by the amount of assets under management – instead of using a return measure, we do find persistent evidence of skill.”
This article was initially published in our Quant Quarterly magazine.