The Costs Of Turnover
Turnover is a measure of trading activity and represents how often a portfolio manager buys and sells the aggregate value of a portfolio. Unlike expense ratios, which are an explicit cost of money management disclosed to mutual fund investors, turnover costs are not disclosed.
The four primary costs typically associated with turnover are:
1. The bid/ask spread: The bid price is the price at which you can sell a security, while the ask price is the price at which you can buy the security. The difference between these prices is known as the spread. Spreads have been decreasing over time as market volumes have increased. One notable event was the New York Stock Exchange's (NYSE's) conversion to decimal pricing in 2001. Previously, NYSE stocks were sold in fractional units of a dollar (e.g., 1/4, 1/8, 1/16, or 1/32), where the minimum spread was the minimum fraction in which the share was sold. For example, a security with a 1/32 spread had a minimum spread of $.03125. Now, spreads can be as low as a penny.
2. Commissions: Commission costs range based upon a variety of factors, such as the size of the trade, the type of the trade (e.g., limit versus market) and the exchange the security trades on (e.g., NASDAQ versus NYSE). Before 1975, there were fixed (and high) minimum commission rates; with the introduction of competitive pricing in 1975, commission costs have decreased dramatically. Individual investors can now trade securities for as little as $6.99 per trade on popular Internet brokers like E*Trade.com; institutional investors can trade for even less.
3. Tax implications: Investors pay taxes on gains as they are realized within the portfolio. Therefore, the more frequently a portfolio manager trades, the more often an investor will have taxable income (assuming the portfolio is making profitable trades). The length of the holding period for securities in the fund is important. Gains (for taxable investors) on investments held for more than one year (i.e., long-term gains) are currently taxed at 15 percent, whereas short-term gains can be taxed as high as 35 percent. Exchange-traded funds limit this tax liability, but open-end funds do not.
4. Market Impact (also referred to as slippage): A portfolio manager who either seeks to buy or sell a large position in a security is likely to impact the market (usually, to his disadvantage). Despite the fact that total market volume has increased (for example, since 1997, the monthly dollar volume on the NASDAQ has more than doubled from $17.830 billion to $41.408 billion), securities that are thinly traded, such as domestic small-cap securities and certain international stocks, can experience dramatic price movements when large investors enter (or exit) a position.
The net impact for each of these costs will differ based upon the market exposure and the experience of the portfolio manager. See, for example, Keim and Madhaven (1997), wherein the authors note how trading costs differ with trader-specific factors such as investment style and order submission strategy, as well as stock-specific factors such as exchange listing (NYSE/AMEX versus NASDAQ). Through an empirical analysis, however, it is possible to determine the overall net impact of turnover on historical mutual fund performance.
An empirical analysis was conducted analyzing the historical returns of mutual funds in order to determine the pre-tax impact of turnover on portfolio performance. The annualized three-year performance for mutual funds, gross of fees, was reviewed: i.e., expense ratios were added back to the three-year annualized return for the analysis. Three-year performance was used since it represents the average investor holding period for mutual funds.1 The time period selected to assess the impact of turnover should be based upon the average holding time in the actual investment. Using shorter periods (e.g., connecting annual periods) ignores the nature in which investors purchase mutual funds (i.e., they tend to hold the same fund), while using longer periods (e.g., 10 years) increases the impact of survivorship bias and the potential for style drift during the test period.
Performance is measured in gross terms, since expense ratios represent an explicit cost that is a drag on performance. While investment management fees should be considered when quantifying the (ex-post) benefit of active management, the purpose of this paper is to quantify the costs of turnover, not to discuss the merits of active management. Mutual funds that have lower turnover rates also tend to have lower expense ratios (e.g., index funds); therefore, if expense ratios were not netted back in this analysis, part of the costs associated with turnover would actually be associated with the underlying expense ratio.
Three-year gross performance for six rolling calendar periods from 2001-2006 was considered for the analysis. This represented the longest period of complete data available to the author. While the overlap between periods is a potential concern (e.g., the 2003 and 2004 three-year test periods would share two years), each calendar period is determined independently. Using multiple periods increased the available data set.
In order for an investment to be included in the analysis, it had to be a publicly traded, open-end mutual fund with performance, turnover, standard deviation (three-year) and expense ratio information available at the calendar year-end test date. All information was obtained from Morningstar. The "style" used for comparison purposes was the investment's asset category at the end of each calendar year (as defined by Morningstar). In order for a mutual fund to be included in the test population, it must have had the same asset category for the previous three quarters as well. This screen was applied to minimize the impact of style drift during the test period. The mutual fund's turnover and expense ratio at the end of the calendar period was assumed to be constant for the entire three-year test period.
The mutual funds tested were limited to one share class per fund to ensure that those funds with multiple share classes were not overweighted compared with funds with fewer share classes. The share class with the lowest expense ratio and complete information was used. (Not limiting the test population to distinct investments is a common error in tests that address the historical benefits of active management in mutual funds.) The lowest expense ratio share class was selected since it is assumed to be the most efficient share class available (typically, this was the institutional share class, which tended also to have larger minimums than A, B, C, Investor or Retirement share classes).