Information Flows and Group Investing Behavior

In April and July, we looked at the psychology of individual investors. This month, we will begin to look at the way information passes between investors, and how this gives rise to group behavior. Next month we will look at how the structure and institutions of the markets themselves affect behavior and returns. In October, we will bring all these strands of thinking together to see if they can help us better understand the wide variations we observe in index performance, along with the patterns that seem to recur across a range of markets.

Individual investors don’t operate in a vacuum. Rather, they interact everyday with other investors, either directly by trading with them or indirectly by observing the consequences of their actions (for example, changes in the price of a stock, its trading volume, etc.).

At the same time, investors are dealing with information about the stocks they own and the ones they would like to buy or sell short. Some of this information costs them very little to obtain (e.g., yesterday’s prices and volumes, or prices relative to an index or moving average), while the cost (in time or money) to obtain other information can be quite high (e.g., buying a research report from Multex, or interviewing customers to check on the relative performance of a company’s products). On top of this, not all information available to an investor is an equally reliable guide to a company’s future stock price – either because it may not be accurate, or because the way other investors will interpret it may not be clear. Finally, investors differ in terms of their perceptions of their own abilities relative to others to accurately interpret the information they have. For example, while Venus Williams certainly wouldn’t defer to Morgan Stanley’s internet analyst when it comes to choosing a tennis racquet, she probably would when it came to buying a technology stock.

These starting points give rise to a number of interesting phenomena in financial markets. Let’s look at the most interesting of these: bubbles and crashes, which are both examples of the phenomena known as "herding". We’ll use an example to show how these play out.

Let’s start with ten people, one of whom is a well known analyst at a famous Wall Street Investment Bank, one of whom is your cousin Al who regularly drones on at family parties about his great record "in the market", and one of whom is your friend Lisa who has a PhD. in software engineering. Further assume that the other seven people (including you) are regular Janes and Joes. On Monday morning, following the release of strong quarterly results by the company, Well Known Analyst announces a strong buy recommendation on XYZ.com, a firm you’ve never heard about before. All of the regular folks hear about this (it is public information), and so take an interest in XYZ. A couple of them buy solely on the strength of the analyst’s recommendation, but everyone else holds back because they are cynical about sell side analysts. On Tuesday, Lisa tells you that she has bought some XYZ, and its price went up ten percent. At the same time, you are sitting at Starbucks and overhear two skateboarders with multiple tattoos and body piercings strongly criticizing XYZ’s website.

That night, you mull over your decision. The analyst recommended it. And Lisa (who at the very least knows more about software than you do, and is therefore better able to make sense of the public information about XYZ, even if she doesn’t have any private information about the company) bought it, and earned a quick ten percent profit (assuming minimal trading costs). On the other hand, you also have private information (from Starbucks), that isn’t positive. On balance, you decide that the weight of the public information (the analyst recommendation and the financial information disclosed by the company), and what you infer from the actions of others about the information they have (Lisa’s and the others’ purchases, and the rise in the stock price), together with the results of those actions (the stock is up ten percent) more than offsets the weight of your private information, and you decide to buy.

On Wednesday, the stock price is up another ten percent. That night Cousin Al sees that XYZ has been among the market’s best performers for the past two days. As you’ve suspected all along, Al has neither the time nor the inclination to do any in-depth research on the company, so he has no private information about it. Instead, he looks at the analyst recommendation, the company’s public financial information, and recent trading volume and price changes in the stock, infers that other people must have private information that is positive, and decides to invest.

On Thursday, three things happen. Two other people besides Al invest in XYZ, and its stock pops another twenty percent. Al calls you to brag about the latest hot stock he has found, and urges you to get in before it’s too late (you change the subject). Later, while eating at your favorite restaurant, you overhear another diner tell his date that he is the head of sales at XYZ and is worried that they may lose a big new sale they had been counting on getting. As a result, earnings might be fifty percent less than expected. When you get home, you think about what is happening. On the one hand, a lot of relatively uninformed people seem to be piling into XYZ – you know for sure that Al can’t have any private positive information about the company that would help him determine that it was undervalued. Nope, Al just bought because XYZ was heading up. You realize that the herd has begun to move, and a bubble may be developing (if the stock price is above the company’s fundamental value, then it is a bubble; however, due to uncertainty about the fundamentals – that sale they may or may not lose, for example – you can’t be sure yet). Moreover, you know that some other investors probably realize this too. You conclude that the risk of holding XYZ has probably increased. On the other hand, you don’t want to be the first one to sell if the stock is still going up (you can just imagine what Al would say if he finds out you got out too soon). Trading off all these considerations, you decide to stay invested in XYZ only if its price gains get larger, to compensate you for what you perceive is the increased risk of holding a potential bubble stock.

On Friday, that is exactly what happens. XYZ finished the day up thirty percent, as two more people buy XYZ stock in a very thin market (nobody else wants to be the first one to leave either). On Saturday, however, you are back at your favorite restaurant, and see the XYZ sales manager sitting at the bar working on his resume, and overhear him telling the bartender that he’s looking for a new job because things are falling apart at the (unnamed) company where he works. On Sunday, you realize that you now have a piece of private information (which, because of the way you obtained it, is legally not "inside" information) that provides a very clear signal about the worsening state of affairs at XYZ.

On Monday, you not only sell your shares in XYZ, but also sell more shares short at a price that is fifty percent under the current market (taking on the risk that XYZ’s price will keep rising and you will lose money in spite of the bad news you expect them to announce). The stock finishes the day fifteen percent up, as current investors (including, loudly, Cousin Al) add to their holdings.

On Tuesday, the short sale data for XYZ stock becomes publicly available, and the company announces that its vice president of sales has resigned. A few investors who were already nervous about a possible bubble begin to sell, and the stock price heads south. Under pressure from the investment bankers at his firm (who are trying to convince XYZ to acquire ABC in an all stock deal), Famous Analyst reiterates his buy, all the while feverishly making calls to his contacts at the company to try to figure out what is going on. At the end of the day, Famous Analyst checks with his trading desk and finds that the short sellers don’t appear to have covered their position yet (that is, they have yet to buy the stock they have committed to deliver via their short sale), even though the stock is down fifteen percent on the day. That night, Soon-to-be-Infamous Analyst has that sinking feeling that somebody out there knows more than he does, and what they know isn’t good. On Wednesday morning he downgrades XYZ.

At this point, the same phenomenon seen at baseball games and rock concerts kicks in. We’re all familiar with it – nobody wants to be the first one to stand up, but once a few people do, everyone else soon follows. Wednesday becomes famous in the short history of XYZ as the day the stock lost sixty percent of its value and employees’ options sunk under the waves.

Let’s summarize what this example tells us about financial markets:

• Investors differ in their access to information (about both companies and about other investors’ information and opinions). Some people only have access to public information (about companies or about other investors’ actions), while others have access to private information too.
  
  
• Publicly available information is asymmetric, because the prices of short sales are not disclosed, while the prices of long purchases are. In other words, the positive information about a stock is always in the market (and the price), but some of the negative information is not (until the price starts falling).
  
  
• Similarly, sell side analysts typically underreact to negative news and overreact to positive news about the companies they follow. In other words, they are systematically over-optimistic. This is logical, given their need to maintain preferential access to the company’s executives (e.g., to sell investment banking business and to develop accurate earnings forecasts), who tend not to like "hold" recommendations on their stock.
  
  
• The quality of private information is usually poor (that is, it rarely provides a clear signal to buy or sell).
  
  
• Investors differ in their ability to interpret the meaning of the public and private information they possess. Another way of saying this is that investors use a wide variety of models to convert the information they have into buy/sell recommendations. For this same reason, investors are prone to copying the decisions of those people whom they regard as having more relevant expertise than they do.
  
  
• Two conditions can set off herding. The first is when many investors follow the same public signal. For example, this could be the recommendation of a famous analyst (often known as a "lead steer" because of his or her apparent ability to get the herd moving) or the announcement that a company has significantly exceeded its expected earnings. The second is when many investors decide to buy or sell based on the belief that the private information they have about a company (e.g., "Charlie was just laid off from there") is more than offset by opposing public information they infer from the transactions they observe in the company’s stock ("everybody is buying it and getting rich, so they must know something I don’t").
  
  
• Bubbles usually begin slowly, because a typical investor must accumulate a substantial amount of information to change his or her views about a stock. Once that view has changed, however, herding often accelerates because people tend to look for and overweight information that confirms their opinions and screen out and underweight information that threatens them.
  
  
• Bubbles (upside herding) are inherently fragile, because as more people realize they are in one, and realize that others probably also know the same thing, the rate of price increase has to keep on increasing to compensate for the increasing risk and keep them from selling. At some point, this process has to come to an end.
  
  
• Bubbles tend to pop either when new public information is released (e.g., a company misses an earnings target or is downgraded by a highly respected analyst) or when relatively unambiguous new private information is received, and percolates among investors directly (via communication) and indirectly (via what people infer from other investors’ trades).
  
  
• Crashes (downside herding) tend to develop quickly for two reasons. First, information held by short sellers is only disclosed through their trading (or absence of it) when the market is falling (e.g., "XYZ fell through its support level today, and investors really began to bail out"). In other words, just as available public information tends to be asymmetrically positive when a stock is going up in price, the opposite is true when its price is falling. Second, for most people fear seems to be a more strongly felt (and contagious) emotion than hope (or even envy). More specifically, price falls tend to trigger changes in investors’ beliefs about the likely actions of other investors faster than prices rises do. For this reason, most of the largest one-day percentage changes in index values have been on the downside rather than the upside.
  
  
• Statistically, herding causes the distribution of returns in financial markets to deviate from the normal distribution. Instead, a typical distribution of financial returns tends to be negatively skewed (big moves are more likely on the downside) as well as "fat tailed" (big moves in either direction are more likely than they would be in the case of a normal distribution), and its volatility (standard deviation) tends to change ("cluster") over time rather than remaining constant.

| Indexes based on Growth and Value | Information Flows and Group Investing Behavior | Recommended Portfolio Performance |