This In Practice piece gives a practitioner’s perspective on the article “Comparing Cost-Mitigation Techniques,” by Robert Novy-Marx and Mihail Velikov, published in the First Quarter 2019 issue of the Financial Analysts Journal.
What’s the Investment Issue?
Executing trades efficiently can be just as important as stock selection for portfolio managers trying to provide value to end investors. This is often overlooked in academic studies.
Reducing transaction costs requires either trading less frequently or selecting stocks that are cheaper to trade (or some combination of the two). There are three common techniques to reduce transaction costs. One is to target turnover by rebalancing a portfolio less frequently. The second is to use “banding”—also known as introducing a “buy/hold spread”—because the criteria to buy stocks is more stringent than the criteria to hold them. The third is to restrict security selection to the universe of cheap-to-trade stocks. This does not impact turnover but rather the cost of turnover. The authors investigate and compare the performance of these three techniques.
How Do the Authors Tackle the Issue?
The authors consider transaction costs of US stocks in a sample period between January 1975 and December 2016. They divide the stocks into three categories at the end of this period: large (companies in the top 90% of total market capitalisation), small (the next 9%), and micro (the next 0.9%).
To measure the cost of trading, they use a bid/ask spread estimator based on daily closing prices, gathered from the CRSP database. They use this measure to compute transaction costs in each size category for seven investment strategies that are popular in academic literature—designed to generate large gross returns but with no regard for transaction costs. The strategies include two accounting strategies, two defensive strategies, and three momentum strategies.
The authors use these seven strategies as the baseline against which to compare the effectiveness of the three transaction-cost-mitigation techniques. First, they look at transaction costs when only dealing in cheap-to-trade stocks, defined as stocks whose trading costs have been below the median in their size universe in the recent past. Second, they lower the frequency at which strategies are rebalanced, from monthly to quarterly. Third, they introduce a 10%/30% buy/hold spread, which means that stocks are only purchased when they enter the buy range of a given strategy (top 10%) but held until they fall out of the hold range (top 30%). They are also sold short using the same principles.
The authors also evaluate screens—a potential way to reduce costs by allowing investors to trade one strategy “on the margin” of another. Finally, they consider how equal-weighted returns, which are often reported in academic literature and disproportionally include micro stocks, compare with the performance of value-weighted strategies.
What Are the Findings?
Before transaction costs, each of the seven baseline strategies generates significant excess returns. Transaction costs substantially diminish their realised average returns. In the large-cap universe, none of the strategies produce statistically significantly net average returns. Small- and micro-cap stocks see their average realised spreads reduced by 55% and 77%, respectively.
Banding is the single most effective cost-mitigation technique. Introducing a buy/hold spread reduces trading costs in every case. It has the added benefit of not significantly reducing the average gross returns of the strategies, because it retains a higher exposure to the underlying signal. It is thus the only technique to consistently yield net gains—most prominently, for small- and micro-cap strategies.
Quarterly rather than monthly rebalancing cuts transaction costs for every strategy considered by a similar amount as banding. It also forces the strategies to use less timely signals. The ensuing decline in gross performance is of similar magnitude to gains from efficient execution, meaning that, on average, there was no net benefit.
The least effective technique is restricting trades to the universe of low-transaction-cost stocks. Although it, too, curtails transaction costs, it also reduces gross strategy performance. This results in a cut in net returns in 10 of the 21 cases examined.
The authors also find that screens can dramatically improve net performance. For example, adding a momentum screen to a size strategy achieves negative transaction costs and almost doubles the average net return.
What Are the Implications for Investors and Investment Professionals?
The authors’ key finding is that although reducing turnover and the cost of turnover are both effective techniques to curtail transaction costs but both diminish the success of the strategy. Banding is more effective than simply rebalancing less frequently because it uses the stock selection signal to inform its trading.
More broadly, this study suggests that transaction costs are vital to the performance of stock-picking strategies. For example, using banding realised net returns of more than 4% per year in almost all the strategies in the small- and micro-cap universes. The authors argue that portfolio managers deliver as much or more value through efficient execution as they do through stock selection. This is especially striking for investment strategies that have high turnover or are mainly geared to trading small- or micro-cap stocks, where each trade is typically more expensive. Yet, according to the authors, efficient execution is often largely overlooked by academics. This may have implications for the performance of popular factor strategies that are not explicitly designed to consider transaction costs.