The following from ‘a reader’, via e-mail to vreaa 29 Dec 2012, who writes: “The following is from a CFA level 3 Behavioral Finance textbook that a friend of mine found at a garage sale. Not too many people have access to the book so it might be of some interest. Nothing really too new, but it’s the only time I’ve seen bubbles and crashes written about in a textbook, they are not discussed in university level textbooks as far as I have seen. Provides a lot of insight into the biases that come into play with bubbles and the summary discusses how the crash unfolds. I sounds to me like we are at the early stages of such a crash.”
Bubbles and Crashes
Stock market bubbles and crashes present a challenge to the concept of market efficiency. Periods of significant overvaluation or undervaluation can persist for more than one year, rather than rapidly correcting to fair value. The efficient market hypothesis implies the absence of such bubbles. The frequent emergence of bubbles in history was documented in Extraordinary Popular Delusions and the Madness of Crowds (Mackay 1841). The book captures the concept of extremes of sentiment and apparent mass irrationality. Bubbles and crashes appear to be panics of buying and selling. A continuous rise in an asset price is fuelled by investors’ expectations of further increase; asset prices become decoupled from economic fundamentals.
A more objective modern definition specifies periods when a price index for an asset class trades more than two standard deviations outside its historic trend. Statistically, if returns are normally distributed, such periods should not represent more than 5 percent of total observations. However, for some stock markets and asset classes, these extremes of valuation account for more than 10 percent.
Bubbles and crashes are, respectively, periods of unusual positive or negative asset returns because of prices varying considerably from or reverting to their intrinsic value. Typically, during these periods, price changes are the main component of returns. Bubbles typically develop more slowly relative to crashes, which can be rapid. This asymmetry points to a difference in the behavioral factors involved. A crash would typically be a fall of 30 percent or more in asset prices in a period of several months. Some bubbles and crashes will reflect rapid changes in economic prospects that investors failed to anticipate. The global oil price shock of the 1970s and the Japanese asset price bubbles of the late 1980s, in which real estate and stock prices rose dramatically, would be examples. Initially in a bubble, some participants may view the trading and prices as a rational response – for example, to easy monetary conditions or a liquidity squeeze – but this view is typically followed by doubts about whether prices reflect fundamental values.
These bubbles have been observed in most decades and in a wide range of asset classes. Recent examples are the technology bubble of 1999-2000 and the residential property boom of 2005-2007, evident in a range of economies globally including the United States, the United Kingdom, and Australia. They appear to be periods of collective irrationality, but can be analyzed in more detail as representing some specific behavioral characteristics of individuals. Behavioral finance does not yet provide a full explanation for such market behavior, but a number of specific cognitive biases and emotional biases prevalent during such periods can be identified.
First, it should be noted that there can also be rational explanations for some bubbles. Rational investors may expect a future crash but not know its exact timing. For periods of time, there may not be effective arbitrage because of the cost of selling short, unwillingness of investors to bear extended losses, or simply unavailability of suitable instruments. These were considerations in the technology and real estate bubbles. Investment managers incentivized on, or accountable for, short-term performance may even rationalize their participation in the bubble in terms of commercial or career risk.
The extent to which investors may rationalize their behavior during bubbles is evident in Exhibit 12 [Investor Behavior in Bubbles]. Both managers appear to have misunderstood risks and exhibited the illusion of control bias. The manager of Fund A believed he could exit a bubble profitably by selling near the top. The manager of Fund B may not have recognized the potential scale of a bubble, or client perspectives on a period of relative underperformance while not participating in the bubble.
Consider the differing behavior of two managers of major hedge funds during the technology stock bubble of 1998-2000:
The manager of Hedge Fund A was asked why he did not get out of internet stocks earlier even though he knew by December 1999 that technology stocks were overvalued. “We thought it was the eighth inning, and it was the ninth. I did not think the NASDAQ composite would go down 33 percent in 15 days.” Faced with losses, and despite a previous strong 12-year record, he resigned as Hedge Fund A’s manager in April 2000.
The manager of Hedge Fund B refused to invest in technology stocks in 1998 and 1999 because he thought they were overvalued. After strong performance over 17 years, Hedge Fund B was dissolved in 2000 because its returns could not keep up with the returns generated by technology stocks.
In bubbles, investors often exhibit symptoms of overconfidence, overtrading, underestimation of risks, failure to diversify, and rejection of contradictory information. With overconfidence, investors are more active and trading volume increases, thus lowering their expected profits. For overconfident investors (active traders), studies have shown that returns are less than returns to either less active traders or the market while risk is higher (Barber and Odean 2000). At the market level, volatility also often increases in a market with overconfident traders.
The overconfidence and excessive trading that contribute to a bubble are linked to confirmation bias and self-attribution bias. In a rising market, sales of stocks from a portfolio will typically be profitable, even if winners are being sold too soon. Investors can have faulty learning models that bias their understanding of this profit to take personal credit for success. This behavior is also related to hindsight bias, in which individuals can reconstruct prior beliefs and deceive themselves that they are correct more often that they truly are. This bias creates the feeling of “I knew it all along.” Selling for a gain appears to validate a good decision in an original purchase and may confer a sense of pride in locking in the profit. This generates overconfidence that can lead to poor decisions. Regret aversion can also encourage investors to participate in a bubble, believing they are “missing out” on profit opportunities as stocks continue to appreciate.
Overconfidence involves an illusion of knowledge. Investors would be better off not trading on all the available information, which includes noise or non-relevant information. Asset prices provide a mix of information, both facts and the mood of the crowd. But in a stock market bubble, noise trading increases and overall trading volumes are high. Noise trading is buying and selling activity conduceted in the absence of meaningful new information, and is often based on the flow of irrelevant information. A manager increasing trading activity in a rising stock market can misinterpret the profitability of activity, believing it is trading skill rather than market direction delivering profits.
The disposition effect recognizes that investors are more willing to sell winners, which can encourage excess trading. There can also be a confirmation bias to select news that supports an existing decision or investment. Indeed, search processes may focus almost exclusively on finding additional confirmatory information. Investors may be uncomfortable with contradictory information and reject it. Investors can also have a bias to buy stocks that attract their attention, and pay more attention to the market when it is rising. For short-term traders who may derive entertainment from the market, monitoring rising stock prices is more entertaining and instills more pride. Entertainment and pride are emotional effects.
As a bubble unwinds, there can be under-reaction that can be caused by anchoring when investors do not update their beliefs sufficiently. The early stages of unwinding a bubble can involve investors in cognitive dissonance, ignoring losses, and attempting to rationalize flawed decisions. As a bubble unwinds, investors may initially be unwilling to accept losses. In crashes, the disposition effect encourages investors to hold on to losers and postpone regret. This response can initially cause an under-reaction to bad news, but a later capitulation and acceleration of share price decline. This situation will only apply to stocks already held by investors, with hedge funds that can sell stock short being more inclined to react first to bad news in a downturn. In crashes, there may be belief that short sellers know more and have superior information or analysis.