Trading costs have predictive power

BOSTON — Forget expense ratios. Portfolio trading costs can be an even bigger drag on mutual fund performance.
On average, U.S. stock funds spent $1.44 per $100 of assets on trading costs, compared with $1.21 on costs that comprised their expense ratios, according to a study that examined trading costs of 1,706 funds between 1995 and 2005.
The study, “Scale Effects in Mutual Fund Performance: The Role of Trading Costs,” also found that every dollar the average fund spent on trading reduced the fund’s assets by about 42 cents. Big trades, those made to accommodate inflows or outflows and trades related to soft-dollar arrangements were even more detrimental to fund performance, the study found.
Since trading costs cut into returns, they are a powerful predictor of negative performance, according to the study. Furthermore, since trading costs vary more widely among similarly managed funds than do expense ratios, understanding those costs goes a long way toward explaining variations in returns.
“To the extent that trading costs explain fund return performance, it’s going to have a much greater potential than the expense ratio, because there is much more variation,” said Gregory B. Kadlec, a professor at Virginia Tech in Blacksburg, Va., and a co-author of the study with Roger Edelen of Boston College in Chestnut Hill, Mass., and Richard Evans of the University of Virginia in Charlottesville.
Morningstar interested
The study has caught the attention of Chicago’s Morningstar Inc., which is looking at ways to apply its findings to the fund data it makes available to financial advisers.
“This could be one of the big things in the future,” said Russel Kinnel, Morningstar’s director of mutual fund research.
Although the fund world is full of interesting data points, few have real “predictive power” when it comes to fund returns, he said.
Trading costs could be among those few, Mr. Kinnel added.
Over the past several years, the Securities and Exchange Commission has been scrutinizing various costs associated with mutual funds. In early 2005, the staff at the SEC said that it was considering a proposal that would have required more disclosure around trading costs.
So far, however, little has been done to mandate more disclosure on how much funds spend to buy and sell stocks.
Unlike other costs, which are reported in a fund’s expense ratio, trading costs are more opaque.
Although funds disclose how much they pay in commissions, brokerage costs account for only about 20% of a fund’s total trading costs, Mr. Kadlec said. Other costs, such as the difference between the price to buy and sell securities, and market impact costs, are buried within a fund’s trading costs, he said.
How do you get a handle on fund trading costs, given that most of what constitutes the total isn’t reported?
Using portfolio turnover isn’t the best barometer — particularly when comparing different asset classes and trade sizes — because it assumes that per-share trading costs are constant across all funds, Mr. Kadlec said.
“As anybody who’s involved with trading can tell you, trading costs vary dramatically across both asset class and trade size,” the professor said, adding that when it comes to estimating how much a fund spends on trading costs annually, turnover is just half the story. “The per-unit trading cost is the other half.”
For example, small-cap stocks can cost as much as three times to trade as large-caps, Mr. Kadlec said.
Turnover is “an unreliable proxy for a fund’s actual trading costs,” he said. “We estimate trading costs using an approach that captures not only the frequency of trade — which is what turnover is capturing — but also recognizing the differential per-unit costs of trade.”
The research also revealed that the effect of trading on fund performance hinges on who is trading and why, Mr. Kadlec said.
Large trades, as well as those driven by asset inflows and outflows, are particularly costly to performance.
Soft-dollar-related trades increase both the cost per trade and trade volume, with no clear benefit, Mr. Kadlec said.
Small trade advantage
“By contrast, when funds trade in small quantities for discretionary purposes, the relationship between performance and trading costs is actually positive,” he said.
The researchers separated the funds by those whose trade size was above the median for their asset class and those whose trade size was below it.
When they calculated the average trading costs for each group, they found that funds with large trades incurred annual trading costs of nearly 1 percentage point more a year than funds with small trades.
“Large funds, for every dollar they [spent] on trading costs, fund value falls by 88 cents,” Mr. Kadlec said, adding that, in contrast, every dollar small funds spent on trading costs was associated with a 38-cent increase in fund value. “So small funds appear to be more than recovering their costs, and actually adding value to their trades.”
Flow-related trades can be costly because they are time constrained — that is, a manager may have to trade quickly — and because they generally are being executed for reasons unrelated to the manager’s investment strategy.
“They typically don’t have an alpha motive behind it,” Mr. Kadlec said.
In soft-dollar arrangements, funds agree to pay more to trade a stock in exchange for an array of brokerage-related services, such as stock research.
“Funds in soft-dollar arrangements may be induced to trade excessively in order to generate the rebates they desire,” Mr. Kadlec said.
Terry Otton, chief executive of San Francisco-based RS Investments, said that although the study was “very good, and we take transparency very seriously,” he urged some caution in applying its findings.
“If you were an individual investor, and you took these raw statistics and then went out and applied it as [Mr. Kadlec] had done formulaically to your mutual fund, you could come to the wrong conclusion,” said Mr. Otton, whose mutual fund firm manages $18 billion in total assets. “He’s just simply saying, ‘Here’s an estimate of the trading costs.’ I’m saying you’ve got to take into consideration the value attached to the cost. What did the trade generate in investor returns?”