When Investors Disagree

Until recently, the reason for those disagreements and their impact on trading remained unclear. However, Snehal Banerjee, an assistant professor of finance at the Kellogg School of Management, and Ilan Kremer, a professor at Stanford University, have developed a business model that, in Banerjee’s words, “can intuitively generate the patterns we see in the data.”

“Our model provides some implications of investor disagreement when investors interpret the same public information differently,” Banerjee explains. “Understanding why people disagree and how their disagreements change over time is an important part of understanding how markets function and, in particular, why there is so much trading in financial markets.”

Learning and Processing Information

The project stemmed from Banerjee’s interest in how investors learn and process information in financial markets, and how that affects prices and trading behavior. “As financial economists, we still don’t have a very good understanding of why people trade as much as they do,” he says. “Why people disagree and how this affects their trading is an important part of the story.”

Investors may react differently to the same piece of news for at least two reasons. They may possess private information about the firms in question, which leads them to update their valuations differently. Or they might interpret the news itself in different ways, because they have different models of the world. In general, these two channels are difficult to distinguish from each other. However, the new model suggests that investors can tell them apart by looking at the two channels’ predictions about patterns in trading volume and volatility.

The model builds on earlier studies by several researchers, but includes key differences. “While these models have improved our understanding of trading volume due to disagreement, they are restricted along some dimensions,” Banerjee explains. “We consider a more general model that relaxes some of these assumptions, and which lets us focus on the specific questions we wanted to look at.”

The model creates a comprehensive picture of the relationship between corporate announcements and trading. “We show that rare but major disagreements about the interpretation of public signals, like earnings announcements or news, can lead to clustering in volume and correlation between volume and volatility,” Banerjee says. “Trading volume jumps up at the date of the public news and then gradually returns to normal over time. This abnormal trading is also associated with higher volatility of prices.”

Three Specific Predictions

While Banerjee and Kremer’s research project was mainly theoretical, their model makes three specific predictions. The first forecasts a large amount of trading volume and volatile prices when investors disagree markedly about a piece of news. This translates into a higher persistence in trading volume and higher average returns.

The second prediction is more subtle. “We do not necessarily observe directly whether investors disagree a lot or a little,” Banerjee explains. “However, the model predicts that higher disagreement is associated with higher volatility. Hence, we can split the data into high- and low-volatility days which should correspond to high- and low-disagreement days. The model then predicts that persistence in volume is positively related to the volume on high-disagreement—that is, high-volatility—days, but negatively related on low-disagreement days.”

The two researchers base their third prediction on the frequency of disagreement among investors on days when companies announce their earnings. So those days, they predict, will see higher trading volume and volatility, and the volume will die down more slowly in the following days.

For an initial check of the predictions, the two researchers used data from the Center for Research in Security Prices. “We have some preliminary empirical results using daily returns and turnover from the CRSP dataset,” Banerjee says. “We document how unexpected trading volume is related to serial correlation in volume, return volatility, and average returns.”

Overall, Banerjee and Kremer conclude, “The differences of opinion framework [that underlies the model] is an interesting and promising alternative to the standard asymmetric information models. The view that agents may ‘agree to disagree’ not only seems plausible, but also appears better suited to address some of the empirical evidence involving trading volume.”

Related reading on Kellogg Insight

Agreeing to Disagree: Differing beliefs lead to price drift

Explaining Stock Price Drift: Trading costs lead the way

Jumps in the Market Make for Jumpy Investors: Short-lived market events can have long-term effects on the appetite for risk

Mysterious Profit: ‘Peso events’ drive returns to the carry trade