Overreaction Hypothesis and Contrarian Strategy (the efficiency of financial markets) - Essay Example

2. Literature Review One of the most basic issues discussed in studies of financial markets is that of its efficiency. Financial market efficiencyimplies that prices reflect the intrinsic values of the stocks being bought and sold in the market, and that the development of the price of a stock over time is a random walk. If prices develop in a predictable manner, arbitrageurs would discern the trends, act on them, and make money at a rate that is above the normal market returns and their actions would quickly bring stock prices to their intrinsic values. This set of assumptions is defined by what is known as the Efficient Market Hypothesis (EMH) that was popularised by Fama (1970). On the other side of the academic spectrum are the proponents of the so-called behavioural finance who think that it is possible to beat the market through a careful analysis of historical price trends and financial reports (Shiller, 1990). They claim that the stock market has several anomalies and market inefficiencies that allow investors to earn higher returns compared to the market. Those who believe in the OR and CS hypotheses belong to this group. The OR hypothesis states that investors overreact to information, and that there are two ways by which investors exaggerate their reaction. In the face of bad news, for example, some investors think that the reality is worse and react over-pessimistically, while some think that the reality is not as bad as it seems and react over-optimistically. So while bad news can be factored in by rational investors according to EMH and their effect on the value of the stock can be calculated before these investors begin to do anything (buy, sell, or hold), some investors are claimed by behavioural finance proponents as acting in irrational ways, making decisions based on their overreaction to information. The effect of overreaction is a large decline in stock prices when pessimistic investors begin to think that the bad news is not true and that the reality is much worse than it really is. The opposite effect holds in the face of good news: investors may overreact and think that the reality is better, so they buy stocks in the market. This shows that some investors are biased in the way they interpret information, and this bias causes stock price anomalies that can be exploited by investors by using a contrarian strategy. The contrarian strategy is based on the contrarian hypothesis: one can make money and beat the market by doing the opposite of what the market is doing. If stocks of a company were being dumped because of bad news, the contrarian strategy would be to buy the stocks because market overreaction caused an anomalous decline in the price. And if the market is buying because of good news, then do the opposite: sell. By doing so, higher than market profits could be made. The first to present empirical evidence on the OR hypothesis were Rosenburg and Rudd (1982), but it was DeBondt and Thaler (1985; 1987) who confirmed the initial findings and came out with research on the topic that continue to be used to this day. Kahnemann and Tversky (1982) observed that people tend to exaggerate their reactions to unexpected events and proposed that this overreaction can affect the behaviour of stock market investors (Kahneman won the 2002 Nobel Prize in Economics for this). Therefore, if overreaction can be observed in the stock market, it would be possible to predict such behaviour in the future. The evidence in the OR hypothesis can be seen in the irregular price changes in the market, which goes against the predictions of supporters of EMH (Fama and Blume, 1966). Several academics have investigated the OR and CS hypotheses and found empirical evidence of their occurrence, like Brown and Harlow (1988), and Brown and Tinic (1988), and Zarowin (1989). Bremer and Sweeney (1991) showed evidence of OR and CS hypotheses in large companies in the U.S., while Lin (1988) and Chang, McLeavey, and Rhee (1995) found evidence in stock exchanges in Taiwan and Tokyo. Yang (1999) used monthly return data over a twenty-year period to examine how stock prices reflect available information in a biased manner by tending to either go abnormally high or unreasonably low and concluded that short-term overreaction and contrarian strategies in Taiwan are not as powerful as in other equity markets. Amongst possible reasons cited for this finding is the inclination of government authorities in emerging markets to protect big firms and the high volatility of stock returns in the TSE. Otchere and Chan (2003) examined the short-run overreaction phenomenon in the Hong Kong market using data from March 1996 to June 1998 during the pre- and post-Asian financial crisis period. There is evidence of overreaction in the market prior to the crisis with the phenomenon more pronounced for winners and losers, but that the abnormal profits obtained from exploiting the occurrence are economically insignificant after accounting for transaction costs. The authors also explored the possibility that results were affected by factors such as bid-ask bounce, the size effect, and the day-of-the-week effect. Ali Siad (2001) tested the Stock Exchange of Thailand (SET) for non-linear dynamics and concluded that the country’s equities markets suggest the presence of high order chaos that implies a complex relationship amongst many variables that make price movements unpredictable. Similar studies have been conducted to investigate the OR and CS hypotheses in other countries. Antoniou, Galariotis, and Spyrou (2005) found that, when January returns are excluded, contrarian profits in the Athens Stock Exchange (ASE) were influenced more by firm-specific rather than common factors, with results reinforced by allowing for time variations in factor sensitivities. Bowman and Iverson (1998) discovered significant price reversals in the New Zealand Stock Market, especially in the case of large price declines and that evidence showed that reversals increase in magnitude as the initial price change increases. Gaunt (2000) found that evidence of price reversal where monthly portfolio rebalancing is employed in the Australian stock market, but the price reversal disappears when a buy and hold strategy is used. Further analysis revealed that the loser portfolio was dominated by small firms and that any abnormal returns are not exploitable given the lack of liquidity in small cap Australian stocks. These findings are inconsistent with those of research in the U.S., implying that the different time periods examined in these studies may have been the main cause. Mun, Vasconcellos, and Kish (2000) investigated the OR/CS hypothesis using a non-parametric methodology with a multi-factor asset pricing model within both the U.S. and Canadian stock markets. Experiments showed that, for the U.S., short-term and immediate-term contrarian portfolios yielded significant excess returns above the market, whilst for the Canadian market, the intermediate-term contrarian portfolio worked best. Although there was evidence of the OR hypothesis from the behaviour of stock prices, the next problem was how to explain it. Cox and Peterson (1994) attributed it to market liquidity and the bid-ask bounce following Amihud and Mendelson (1987). Other researchers showed that profits from contrarian strategies are not significant if return calculations are adjusted for seasonality (Jones, 1987; Brown and Harlow, 1988; Zarowin, 1989; Pettengill and Jordan, 1990), time frame used for return calculations (Ball, Kothari, and Wasley, firm size effects (Zarowin, 1990), use of bid instead of ask prices to calculate returns (Ball and Kothari, 1995), and risk change (Chan, 1988). Chopra, Lakonishok, and Ritter (1992) noted that while price reversals are evident, explaining the reasons behind such findings is not easy. This is the main reason why research on the OR and CS hypotheses continue to be made. Lee and Swaminathan (2000) showed evidence that past trading volume provides an important link between momentum and value strategies. Firms with high past turnover ratios earn lower future returns and have consistently more negative earnings surprises over the next eight quarters, and that past trading volume also predicts both the magnitude and persistence of price momentum, helping to reconcile intermediate-horizon “underreaction” and long-horizon “overreaction” effects. Similar studies in emerging markets provide insights into the behaviour of the Stock Exchange of Thailand (SET). There is some evidence on overreaction in the emerging stock markets. Da Costa (1994) found that the Brazilian stock market is over-reactive and that price reversals are asymmetric for a data set of Brazilian stocks. Harvey (1995) discovered that autocorrelation is much higher in emerging markets than in developed markets. He also suggests that the level of autocorrelation is directly associated with the size and the degree of concentration of the market. Higher autocorrelation would imply persistence in the sign of the returns, and consequently, predictability. Erb, Harvey, and Viskanta (1995) find that equity returns and volatility are predictable for a group of 40 countries by using credit risks obtained from Institutional Investor as the sole explanatory variable. Diamonte, Liew, and Stevens (1996) indicate that changes in political risk measures are capable of predicting the returns in emerging markets better than in developed markets. So far, we have seen that overreaction may be explained by firm-specific and external common factors, time boundaries or seasonality, investment strategies (portfolio rebalancing or buy and hold) or differentials in the bid and ask price, or market liquidity, that adopting a contrarian investing strategy has been found to give above market returns in most countries. There are a few exceptions like Taiwan due to factors that affect any or most of the factors that influence overreaction, and that there is sufficient evidence across countries and markets, both developed and emerging, that overreaction has been observed in the short, intermediate, and long-term. We can summarise the key points of the OR and CS hypotheses based on Shefrin (2000) as follows: First, investors overreact to both bad news and good news (De Bondt and Thaler) but overreaction leads past losers to become underpriced and past winners to become overpriced. Second, apparent overreaction of stock prices to information is about as common as underreaction (Fama). Third, various evidence show that there is apparent underreaction at short horizons between three and twelve months and apparent overreaction at long horizons for periods more than one year. Before we investigate the evidence of overreaction in the Thai stock market, we look at other theories that can help us understand this market anomaly: recency, overconfidence and biased self attribution, and the strength and weight of event information, after which we conclude this chapter with a review of the literature on our empirical calculations. De Bondt and Thaler (1985, 1987) attributed overreaction to the psychological phenomenon of recency, which states that people when assessing information tend to overweigh recent information compared with their prior belief. This means that investors who are not sure of the intrinsic value of a stock will be too optimistic about its value when the firm is winning, i.e., its price is going up, and too pessimistic when it is losing, i.e., its price is going down. Recency therefore results in a time lag before stock prices reflect intrinsic value. Stock market investors and traders form beliefs about the future value of stocks, leading to an overvaluation of winning firms and undervaluation of losing firms, as predicated by the OR hypothesis. Daniel, Hirshleifer, and Subrahmanyam (1998) attempted to integrate underreaction and overreaction using psychological evidence about individual behaviour based on psychological findings that people are generally overconfident and subject to biased self- attribution. Experts such as investment analysts tend to be more overconfident compared to inexperienced investors and therefore take on more risks than what may be prudent. Biased self attribution concerns the way investors adjust their beliefs when public information is disclosed: confidence increases if the information agrees with prior private beliefs, but it does not decrease if the information is contradictory to prior private beliefs. When public information does not confirm private information, they tend to attribute the result to bad luck rather than to ignorance or lack of ability. Therefore, investors tend to overreact to private information, giving rise to short-term price overreaction. When the private information contains good news, the share price will rise too high, but if the information contains bad news, the share price will tend to fall too much, and when public disclosure confirms the private information, a further overreaction takes place. However, when subsequent public disclosures eventually show that the share price is too high or too low, the result is a price adjustment in the right direction although the adjustment to the rational price will be much slower than the previous overreaction, leading to the long-term reversals discovered by Lakonishok, Shleifer, and Vishny (1994). Barberis, Shleifer, and Vishny (1998) observed that signals have both strength and weight. Strength relates to the size of the signal whilst weight relates to how much importance should be placed on it. Underreaction tends to occur in signals of low strength and high weight, whilst overreaction would tend to occur in signals of high strength and low weight. Investors do not give enough emphasis to the weight of a signal, focusing more on its strength. The manner by which the OR and CS hypotheses are manifested and measured in stock markets is through changes in the stock price or in the mean of market returns, which EMH predicts reflects all available publicly available information. At the same time, the weak form of the EMH states that it is not possible to predict stock prices based on past information (Fama, 1970). As we have already seen, De Bondt and Thaler (1985) challenged this, showing through empirical evidence that after a year or so, the returns of stocks that are undervalued now would outperform by 25 percent the returns of stocks that are currently overvalued, leading to the conclusion that it is possible to predict future winners by investing in today’s losers. In lieu of stock prices, market indices may be used as shown by Ratner and Leal (1999) who tested for overreaction in emerging equities markets in Latin America and Asia. They observed market overreaction in some of the emerging markets, but the evidence for the majority of the emerging markets is contrary to the market overreaction hypothesis. Using a logit analysis explained some of the large one-day movements in emerging markets, particularly in Asia including Thailand, but they find that given the generally insignificant abnormal returns following a large one-day movement, it is unlikely that a short-term trading strategy based on market overreaction would be profitable. As Antoniou et al. (2005) pointed out, investor overreaction to information implies that price reversals may be predictable from past information, a fact that directly contradicts the weak form of EMH. Citing De Bondt and Thaler (1987), they mention evidence consistent with the OR hypothesis showing that excess return for losers in the test period are negatively related to both long-term and short-term formation period performances, and that this effect cannot be attributed to changes in risk as measured by models such as the Capital Asset Pricing Model nor to the size effect. Computing for the return of a stock before and after an event and observing the relationship between the two sets of data can therefore determine evidence of overreaction. From various literature already reviewed, we have seen that if the movement of the price of a stock exhibits a marked change after the event compared to its price movements prior to the same event, what is understood by the related terms price reversal or reversal of mean market returns, this can be seen as evidence of investor or market overreaction that would allow contrarian profits to be earned. In developing their model, Antoniou et al. (2005) build on previous studies by Lo and MacKinlay (1990) and Jegadeesh and Titman (1995) that employ a specific portfolio strategy, weekly re-balancing of portfolios, and estimation of profits. We discuss these further below. In the next section (3) of this paper, we explain our research data and methodology. Section 4 discusses the empirical results of our analysis before we conclude in Section 5 our findings to our research problem of whether the OR and CS hypotheses was observed in the SET during the period under investigation. Lee, C.M. and Swaminathan, B. (2000) Price momentum and trading volume. Journal of Finance, 55 (5), 2017-2069. Read More