Some of the largest, most liquid stocks have some of the highest trading costs, a new study shows.
The wide-ranging study measured the market impact of every trade of every stock last year in Standard & Poor's three broad market indexes and the Nasdaq 100.
These high, unexpected costs for "no-brainer" trades contributed to magnifying the overall cost of trading in terms of market impact, said Laszlo Birinyi Jr., principal of Birinyi Associates Inc., Greenwich, Conn., which conducted the study.
The findings potentially could hurt soft-dollar brokerage firms - and waves a warning flag for such users, according to one portfolio manager - although the study doesn't raise this issue.
Microsoft Corp., General Motors Corp. and AT&T Corp. - three of the largest stocks in the S&P 500 in terms of market capitalization - were among the most expensive stocks to trade in 1994, the study noted.
The costs of trading the stocks was almost six times that of other stocks among the biggest in market value in the S&P 500.
"Many liquid stocks actually have high associated costs of trading," according to the study.
"A stock's associated cost of trading is not necessarily a function of its market capitalization nor its volume."
General Electric Co., the largest stock in terms of market capitalization, has a trading impact almost 45% lower than the second largest stock, AT&T.
The study's findings suggest a need to redefine liquidity in its commonly understood meaning as low-cost trading, Mr. Birinyi said.
Typically, liquidity "is being able to trade large amounts of stock at relatively stable prices," Mr. Birinyi said in an interview. But the study found many liquid stocks have large market impacts.
He said the results of the study suggest portfolio managers would better serve their performance, and their client, by giving more attention to trading large stocks, rather than just on small, difficult-to-trade stocks. Large stocks typically account for a large part of a portfolio's value. With smaller stocks, the individual market impact cost may be great, but the relative impact on a portfolio is not.
"Maybe they should spend more attention to trading GM and maybe they can save more money by trading GM," he said.
"We should have a different focus on trading."
Commenting on the Birinyi findings on liquidity, Theodore R. Aronson, chief investment officer, Aronson + Fogler, Philadelphia, who hasn't yet seen the study, said: "Even though it's counterintuitive, it makes sense."
"He's absolutely right," Mr. Aronson said. "There are no no-brainer trades on Wall Street. Thoughtful people do have to pay attention to every trade."
"You can get zapped with something in this game you don't think is a problem," such as trading supposedly highly liquid stocks, he added.
Portfolio managers and pension funds trying to earn soft-dollar credits typically send trades on larger capitalization, more liquid stocks to soft-dollar brokerage firms.
But the study's findings may mean such trades require more talent or capital than assumed to obtain best execution.
"I've often though that line of reasoning" - that you can send the no-brianer trades to soft-dollar brokers without hurting best execution - was flawed," Mr. Aronson said.
"Birinyi's evidence waves a warning flag," he added. "Everybody is in agreement that you want lots of talent and capital for trades in difficult stocks. But now you have a warning for the easier trades."
Among the stocks in the S&P 500, Intel Corp. was the most expensive in 1994 to trade, costing investors $74.89 million in terms of market impact.
Alberto-Culver Co. was the cheapest of the S&P 500 stocks to trade, costing a mere $246,263 in market impact in 1994.
"We were surprised to discover .*.*. the cost of trading is markedly different from liquidity," the study noted.
The study also shows:
The total market impact of trading last year increased from 1993;
Market execution trading costs appear to exhibit large degrees of randomness; and
Institutional portfolio manager underperformance may be a function of underestimation of the impact of trading;
The cost of trading the 1,500 stocks in the three S&P indexes -the S&P 500, the S&P MidCap 400 and the S&P 600 - totaled $6.102 billion last year, the study calculated from the 37.8 million transactions.
Broken down by indexes, the study calculated the S&P 500's market impact costs last year at $2.93 billion, or 17.76 basis points of its total trading value of $1.65 trillion.
The S&P MidCap 400's market impact costs totaled $1.499 billion, or 28.63 basis points of its total trading value of volume of $523 billion.
The S&P 600's market impact costs totaled $1.67 billion for the year. The study didn't provide comparable trading impact and trade volume value, because the S&P started the small-cap 600 index in October.
The cost of trading the Nasdaq 100 stocks, which include some issues in the S&P indexes, totaled $1.546 billion, or 23.75 basis points of its total trading volume of $651 billion.
All of the market impact costs were paid by anyone doing stock transactions, including portfolio managers, securities dealers and individuals.
For institutions and individuals, trading costs are even higher than the average in the study, Mr. Birinyi said. The study measures all trades - even those of floor brokers and specialists, who may trade at a lower market impact, he added.
The study defines market impact as a purchase or sale of a stock that results in a change in price. Thus, if an institution buys 10,000 shares of a stock whose last sale price is $22 a share and pays $22.25, its execution cost, or market impact, is 25 cents a share, or $2,500 for the entire trade. The cost of trading in the study excludes brokerage commissions.
Mr. Birinyi suggested the market impact of trading domestic equities rose substantially in the past year, although he didn't do such an extensive study of trading in 1993.
The study found no compelling relationship between a stock's market capitalization and its trading costs. Tribune Co., the 239th largest stock, has a similar impact to General Electric.
Likewise, the study found no clear relationship between trading activity and market impact. The study found the cost of trading AT&T, the most active among the 20 largest, was almost twice as high as the second most active, IBM Corp.
In 1994, the average cost per trade, adjusted for the size of the trade, for AT&T was 8.7 basis points. For IBM, it was 5.1 basis points. Both had similar trading volume, number of trades per day and average trade size.
"It would appear the total cost of trading is not explained by either market value or volume," according to the study.
The study doesn't have advice for investors on how to trade to reduce market impact, nor does it explain why it cost more to trade stocks with otherwise similar characteristics.
"I can't explain the difference in the cost impact between, say, GE and AT&T," Mr. Birinyi said.
"We always assume stocks like GM and Wal-Mart are very liquid," he added. "If you don't focus on market impact, you could overlook the fact that there are lots of one-eighth-point moves. You should handle stocks like GM with more care."
For instance, he noted about a third of the trades in Wal-Mart Stores Inc. stock have a one-eighth-point market impact, while only a little more than 10% of IBM's trade have a one-eighth-point impact. On the other hand, about a quarter of the trades in Wal-Mart have zero market impact, while close to 50% of IBM's trades have no market impact.
Mr. Birinyi drew attention to the stocks with the highest absolute dollar costs to trade. Other stocks have higher costs per trade.
Among stocks in the S&P 500, Scientific-Atlanta Inc.'s cost per trade was 10 times as much as Intel (66.3 basis points vs. 6.9 basis points).
But Intel had the higher absolute trading costs, because its trading volume was much higher than Scientific-Atlanta's.
"This shows what people really did," where they spent more of their money, he said. "This is where the traffic is."
"You can take an abstract, where the average cost per trade is very high," he added. "But if there are only three trades a day in the stock," it's not that significant for the overall trading value.
The cost of trading has hurt manager performance, Mr. Birinyi said the study suggests.
"It's an area we don't understand as much as we should," he said. "Managers focus on the economy, on corporate activity, on corporate management - all important issues. Managers just aren't aware of how important it (trading cost) is.
"Trading cost has been treated as a second-class citizen."