Content area
Full Text
The early tests of the capital asset pricing model (CAPM) in the 1970s showed that the empirical risk–return relation is too flat. This low-risk effect is now regarded as one of the very first stock market anomalies. Many other anomalies have been documented since, such as the size, value, quality, and momentum effects. However, the “publish or perish” culture in academia may give a bias toward many false positive results. Harvey, Liu, and Zhu [2016] articulated this concern, documented the recent explosion in the number of factors, and suggested that thresholds for statistical significance be raised. Hou, Xue, and Zhang [2017] took a fresh look at the anomalies literature and found that many results cannot be replicated and are therefore likely the result of factor fishing. Moreover, the significance of many factors turns out to be critically dependent on methodological assumptions such as frictionless rebalancing, no leverage costs, and unlimited short-selling of micro caps.
Investors who want to profit from academic insights therefore have to be very careful. First, they should be highly critical as to which factors work and which do not because many results are spurious. The number of factors needs to be narrowed down to a subset that is persistent and robust. Second, translating multiple factors into one easily implementable investment strategy is not straightforward. Today, a good starting point for a limited set of persistent factors might be the ones identified by Fama and French [2015]. These factors are rigorously tested, complement each other, and require a limited amount of trading. To benefit, investors should tilt their portfolio toward small, attractively priced, profitable firms with low levels of investments. However, Fama and French did not create one integrated investment portfolio, but rather separate long–short portfolios for each factor, with many overlapping positions. This is of little use for investors looking for an easy and effective factor-based investment strategy. One could consider identifying stocks that score well on each of the Fama–French factors, but with such an approach the resulting number of stocks will be very low. For example, the expected number of stocks with a top quintile score on each of five independent factors is less than 1 in 3,000. Moreover, most investors face constraints on leverage and short selling. Therefore,...