There are a number of recent studies that propose a more rigorous criteria for evaluating the practical significance of factors published in academic research journals. First, Harvey, Liu, and Zhu (2015) argue that a t-stat of 3 should be replacing the old 2 as a rule for statistical significance. In 2017, Campbell Harvey was quoted claiming the following:
Half the financial products (promising outperformance) that companies are selling to clients are false.Also, McLean and Pontiff (2014), Chordia, Subrahmanyam and Tong (2014), and Hou, Xue, and Zhang (2017) document a post publication reduction in average strategy performance (across numerous anomalies), but surprisingly none of these papers really include an in-depth transaction cost analysis in their performance calculations. Finally, the Fama-French (and Carhart) factors (beta, value, size, momentum), which are the foundation for many smart beta strategies, were not designed with t-costs in mind and could potentially overstate what an investor can realize when investing in these strategies. (Here is a post with an introduction to factors.) Despite all the aforementioned attempts to question the validity of factor investing strategies, transaction costs are not really addressed in detail. The academic article that really sparked the debate on the importance of considering transaction costs for factor investment strategies was published by Frazzini, Israel, and Moskowitz (2014) (Here is a discussion of this study). The study sparked debate because it suggested that transaction costs were not that big a deal when one actually looks at live data (which was in contrast to prior academic research). But the academics were not satisfied with this answer and a more recent study conducted by Robert Novy-Marx and Mihahil Velikov, and published in the Review of Financial Studies at the beginning of 2016, takes the issue to the next level by evaluating a larger set of well-known anomalies. The article, “A Taxonomy of Anomalies and Their Trading Costs,” examines the after-transaction cost performance for 23 different factor investing strategies over longer horizons and across various market capitalization classes, an improvement over other studies. Interestingly, the authors calculate transaction costs using the effective bid-ask spread measure proposed by Hasbrouck (2009) (working paper version here). Considering that the bid/ask spread does not account for the price impact of large trades, it should be interpreted as the cost faced by a small liquidity demander. The authors also examine the relationship between low turnover and higher capacity across various factors. A summary of the main questions and insights include the following:
- What are the costs of trading the most important anomalies?
- What is the capacity that each of these strategies has to attract new capital before it becomes unprofitable to marginal trading?
- Are there effective transaction cost mitigation techniques?
ConclusionIt seems that everywhere you look there is a promotion related to factor investing and/or smart beta. The incentives to develop strategies with strong backtests are strong, both in academia and in industry. This natural conflict of interest should raise concern for investors who are trying to ascertain the validity of a particular study or investment approach. One must always consider the possibility of data-snooping, overfitting, and transaction costs — do they make the strong results null and void? This paper is also important because the results are a great contrast to the research presented in the Frazzini et al. paper. (detailed review here). Bottomline: investors need to be diligent and think critically when presented hypothetical (live results are arguably more dangerous) results.
A Taxonomy of Anomalies and Their Trading Costs
- Robert Novy-Marx
- Mihahil Velikov
We study the after-trading-cost performance of anomalies, and effectiveness of transaction cost mitigation techniques. Introducing a buy/hold spread, with more stringent requirements for establishing positions than for maintaining them, is the most effective cost mitigation technique. Most anomalies with turnover less than 50% per month generate significant net spreads when designed to mitigate transaction costs; few with higher turnover do. The extent to which new capital reduces strategy profitability is inversely related to turnover, and strategies based on size, value, and profitability have the greatest capacities to support new capital. Transaction costs always reduce strategy profitability.