Buy low, sell high seems sage enough investment advice, but it does not explain why stock prices continue to climb if earnings are good—or fall if earnings are bad—months after company earnings announcements.
“If you knew the stock price was going to go up, then you’d buy more now so you would profit from it,” says Tjomme Rusticus, (Assistant Professor of Accounting Information and Management at the Kellogg School of Management).
So how to explain the well-documented drift in stock prices that dog earnings reports?
Researchers suspected the drift might be due to transaction costs—fees paid to market makers that stand ready to buy and sell a particular stock at a publicly quoted price on a regular and continuous basis. An analysis of earnings reports eighteen years of quarterly trading data shows this is exactly the case, reports Rusticus and his co-authors Jeff Ng and Rodrigo Verdi (Assistant Professors of Accounting at MIT Sloan School of Management).
“In the long distance past—that’s about forty years ago—the first large-scale data on company’s earnings and stock prices became available. That’s when accounting researchers could start investigating how the stock market uses earnings information,” Rusticus says.
“The first thing they looked at was a rather basic question: If we have good earnings—higher than expected—is that something that investors value? If you think about it, you would say yes: If a company makes a lot of money and has higher earnings, then the stock price should go up.”
An Accounting Puzzle
Unsurprisingly, better earnings were followed by a boost in a company’s stock price. Conversely, lower earnings triggered a decline. However, the analysis showed something else as well: Stock prices not only went up or down after earnings announcements, but they continued to do so for months afterward. “That was the start of a puzzle in accounting research,” Rusticus says.
Rusticus and his colleagues measured the transaction costs by looking at the difference between the price traders were charging to sell a stock and the price they were charging to buy.
More research followed using different data, time periods, and measures in an attempt to explain the phenomena. Researchers investigated whether the result could be due to risk, or whether somehow investors did not pay enough attention to or did not understand earnings reports. Some wondered whether the returns in the data were really attainable given that investors incur costs when trading stocks.
“One of the explanations was that there are costs involved with trading,” Rusticus says. “The returns that you see in the database as the change in price are not exactly the money that you would make. When you buy a stock, you tend not to get the best price available. You tend to trade with a market maker, for example, who will sell you a stock for a little more than he or she bought it for. There’s an intermediary who takes some profit every time they sell a share, so the profit that the trader makes is less than the profit observed in the database.”
Rusticus and his colleagues measured the transaction costs by looking at the difference between the price traders were charging to sell a stock and the price they were charging to buy. That difference between the ask price and the bid price was their profit—a cost that was normally hidden in day-to-day trading, but one that could explain the drift in stock prices months after earnings reports.
For example, suppose there is an investor who knows that the true value of a stock is $100 a share. A favorable earnings report comes out and each share of the company’s stock is now worth $105. If buying will cost $1 and there will be another $1 fee to sell at some point in the future, then the savvy investor knows that he or she can keep buying until the share price reaches $103. If you pay more than that, you would lose money in the trade. Pay less and you would gain.
The Bid-Ask Spread Factor
To show that transaction costs were systematically affecting stock prices over time, Rusticus grouped stocks based on the spread in their bid and ask prices, the profit a trader stood to make in the transaction. The idea was that stocks with a smaller spread, which are cheaper to trade, would have a greater initial bump in price following earnings reports and less change in the price months later. In contrast, stocks with a bigger bid-ask spread would still show the impact of the transaction fees months later.
To compute the daily percent commission rate, the researchers first obtained the average trade size for each stock by averaging the dollar volume of the trades within each day. They then took the average trade size and combined it with a commission schedule from CIGNA Financial Services to estimate the commission for an average trade. Daily percentage commission rates were determined by dividing the commission by the average trade size. The last step in the process was to aggregate the daily spread (the difference between the asking price and the selling price) and the commission rate during the month of earnings announcements to estimate the average transaction cost.
The researchers looked at earnings from 126,386 announcements culled from the quarterly trading data. They matched this with stock returns and transaction cost data between 1988 and 2005.
“We first looked at how big a response [in the stock price] there was at the earnings announcement,” Rusticus said. “For companies with low transaction costs, there should be a big response at the earnings announcement because it’s cheap to trade, so the whole response should come then. For companies where transaction costs are high, that’s where we would expect a small response at the earnings announcement and more returns in the subsequent months. That’s exactly what we found.”
The research does not rule out other factors that may be triggering the stock price drift, but it is well supported by the data.
“Our theory and empirical results do a good job at capturing patterns in the data. It doesn’t mean that our theory is the only explanation,” Rusticus adds. “Just as in physics there are all kinds of forces that work on things. In economics, the same thing applies.”