رفتار قابل پیش بینی، سود، و توجه
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|10824||2007||21 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Empirical Finance, Volume 14, Issue 5, December 2007, Pages 590–610
Stocks in the Shanghai market that hit upper price limits typically exhibit three characteristics: high returns, high volumes, and news coverage. We show that these price limit events attract investors' attention. Attention-grabbing events lead active individual investors to buy stocks they have not previously owned. Consistent with lowering investor search costs, events that affect a few (many) stocks lead to increased (decreased) buying. Upper price limit events coincide with initial price increases followed by statistically significant price mean reversion over the following week. Rational traders (statistical arbitrageurs) profit in response to attention-based buying. Smart traders accumulate shares on date t, sell shares on date t + 1, and earn a daily average profit of 1.16%. We show the amount they invest predicts the degree of attention-based buying by individual investors. We end by decomposing individual investor trades in order to estimate losses attributable to behavioral biases.
Recent work in financial economics suggests that individual investors have limited attention and processing capabilities.2 Such traits become particularly apparent when studying the investment choices of active individual investors. When deciding which stock to purchase, individuals face a daunting search problem that is exacerbated by the thousands of stocks to choose from. Behavioral theories predict that attention-grabbing events help to narrow the universe of stocks an individual might research. This narrowed universe of stocks is called the “consideration set”. In a world with limited short selling, newly considered stocks (even those stocks with poor prospects) do not induce investors to initiate short positions.3 Most individual investors hold only three or four stocks in their portfolios, so they have narrow consideration sets when it comes to deciding which stock to sell next. Thus, attention-grabbing events lead to predictable behavior—individual investors become net buyers of stocks that catch their attention. Barber and Odean (2005) are the first to comprehensively study individual trading behavior in the presence of attention-grabbing events. They argue that abnormal trading volume, extreme returns, and news can all be thought of as attention-grabbing events. Their empirical analysis shows that each of these three types of events is indeed linked to aggregate net buying by individual investors.4 In this paper we study trading behavior and stocks that catch individual investors' attention. First, we examine whether attention-grabbing events lead to predictable buying behavior by individual investors. We test whether attention-grabbing events (on date t) are linked with net individual trade imbalances the following trading day (on date t + 1) by computing a net buy–sell imbalance measure on date t + 1 for each event. This measure is the same one as that used in Barber and Odean (2005) and our results provide an out-of-sample confirmation of their results. Second, the link between attention and investor behavior is predicated on substantial search costs faced by individual investors. If few events happen simultaneously, search costs are reduced, and the consideration set is considerably narrowed. We test this hypothesis by measuring individual imbalances on date t + 1 as a function of the number of contemporaneous events on date t. We expect more positive (less positive) buy–sell imbalances on days following few (many) contemporaneous events. Third, attention-grabbing events help individual investors narrow the set of stocks under consideration. After an event, an investor's consideration set may contain stocks the investor has not previously owned. If upper price limit events catch the attention of individual investors, there should be more first-time buys of a particular stock the day following an attention-grabbing event compared to other days. Therefore, we test whether price limit events cause investors to consider, and ultimately purchase, stocks they have not previously owned. Fourth, behavioral biases do not exist in a vacuum—especially when biases coincide with asset price movements. We hypothesize that there exist traders who rationally try to profit by trading against individuals who exhibit traits such as attention-based buying. Throughout the paper we refer to this group as “smart traders” or “statistical arbitrageurs”. Showing that smart traders profit by exploiting a group such as individuals is difficult. It requires data that are not available in many situations and also requires making a direct link between the trading behaviors of the two groups of investors. We outline a profitable strategy based on publicly available information that takes advantage of individual investor trading around upper price limit events. Smart traders accumulate shares during upper price events by buying from individuals who are (presumably) willing to sell for a gain (consistent with the disposition effect.) The smart traders then sell the following day to another group of individuals who are eager to buy (consistent with attention-grabbing events.) We identify accounts that execute such a strategy and describe the trading behavior in a level of detail previously not possible. Rather than studying a number of different types of attention-grabbing events, we choose to study a single type of event that combines several important features: upper price limit events on the Shanghai Stock Exchange.5 The exchange is an electronic limit order book system that allows us to know when a trade is placed, when a trade is executed, which account placed the trade, which account was on the other side of the trade, and where these accounts are located. Like many markets in the world, the Shanghai Stock Exchange imposes daily price movement limits. Studying stocks that hit their upper price limits has a number of benefits. First, an upper price limit event incorporates three characteristics previously associated with attention-grabbing events as in Barber and Odean (2005): i) return is high, ii) volume is high, and iii) the event generates news. Therefore our analysis can be viewed as an extension of the Barber and Odean (2005) study. After the Shanghai market closes for the day, a stock that has hit its daily limit is featured and discussed on investment-related television programs such as “China Business News”. Investors often watch television programs at the end of the trading day in order to get information before the next trading day. Second, using upper price limit events provides a fairly clean separation of event days and post-event days. The separation is possible because news and investment programs that report upper price limit events are typically aired after the close of the market. Our choice of market and event type also has some costs associated with the benefits mentioned in the previous paragraph. Once a stock hits its upper price limit, the Shanghai Stock Exchange allows trading to continue, but transaction prices may not exceed the limit. Thus, the time series of transaction prices are probably censored. The censoring is not a problem when measuring individual investor net buy–sell imbalances; however it does become an issue when estimating economic losses borne by individual investors. We address censored prices by decomposing losses into two portions. The first portion is attributed to the disposition effect (i.e., investors who sell a stock once it hits its limit even though the price is likely to rise the next day). The second portion comes from attention-based buying (i.e., buying a stock only to see its price mean-revert downward over the next five days). Our empirical analyses produce several important findings. We confirm existing results by showing that individuals have positive net buy–sell imbalances following attention-grabbing events. Further, individual investor net buy–sell imbalances are more positive (less positive) when a few (many) stocks hit their upper price limits on the same day. We find that attention-grabbing events induce individual investors to buy stocks they have not previously owned. In other words, the number of first-time buys for a given stock is significantly higher following an attention-grabbing event than during a typical trading day. We confirm that attention-grabbing events coincide with statistically significant mean reversion in prices. The day after an attention-grabbing event, individual investors are net buyers and prices appear to be “pushed” upward. Between dates t + 1 and t + 6 prices mean-revert back to pre-event levels. Our study has interesting findings regarding the so-called “smart traders” who accumulate shares during date t in anticipation of individual buying demand on date t + 1. The smart traders sell out these accumulated shares on date t + 1 and earn an average daily profit of 1.16% (0.71% net of transaction costs). The smart traders build large positions during times when there are a limited number of attention-grabbing events and during events with high trading volumes. The size of the smart traders' positions in a given stock on date t predict the level of individual investor net buying on date t + 1, suggesting that they are able to anticipate buying by individual investors. Individual investors' behavioral biases are costly. A portion of individual losses comes from selling stock on date t even though prices rise (on average) the following day. The desire to sell shares during a day with large price increases (therefore most likely at a gain) can be thought of as a form of the disposition effect or a simple misunderstanding of censored prices. The remaining portion of individual losses comes from buying stock on date t + 1 because prices mean-revert (on average) over the following five days. This second source of individual loss is of particular interest to us since we can attribute these losses to attention-based buying. Using our transaction data, we estimate that individual lose 1.46% over one day due to selling too early (disposition) and 0.88% over five days due to attention-based buying. Losses are naturally related to the smart trader profits, but are not the result of adding-up constraints, as there are other participants operating in the market. The rest of the paper is organized as follows. Section 2 describes our unique datasets. In Section 3 we study individual investor trading around upper price limit events documenting and testing for attention-based buying. In Section 4 we show that stock prices experience significant mean-reversion following attention-grabbing events. The focus of Section 5 is on the rational response to attention-based buying. In Section 6, we decompose individual investor losses into a portion attributable to selling and a portion attributable to attention-based buying, and Section 7 includes the conclusion.
نتیجه گیری انگلیسی
This paper studies the set of links between attention-grabbing events, predictable behavior by individual investors, transitory price movements, and the rational response of statistical arbitrageurs. We show that attention-grabbing events lead active individual investors to be net buyers of stocks. Moreover, these events lead investors who have not previously owned a stock to consider and ultimately purchase the stock. We argue and present evidence that not all attention-grabbing events lead to predictable behavior. When many events happen simultaneously, search costs are not reduced, the consideration set is not narrowed, and we do not see attention-based buying. Our paper shows that the buying coincides with transitory price movements. Stock prices temporarily rise following attention-grabbing events before mean-reverting to pre-event levels over the next five days. More importantly, we hypothesize that behavioral biases do not exist in a vacuum—especially when a bias is linked to asset price movements. To test our hypothesis, we study the high-frequency trading strategy of a group of statistical arbitrageurs or “smart traders” who are active around attention-grabbing events. The smart traders earn one-day profits of 1.16% by trading against individuals. We end by estimating the losses suffered by individual investors. Individual investors who currently holds a company's shares sell as prices increase during upper price limit events, but lose out on 1.46% of future price increases. Individuals who buy shares following attention-grabbing events lose 0.88% as prices mean-revert over the next five days.