momentum or contrarian strategy. The information used in the original and basic technical analysis is based on price. Lehmann (1990) points out that using contrarian strategy based on the return of previous week can earn profits on the following week. His explanation is that fundamental valuation does not change over a short time interval such as a week. Even though many scholars attribute Lehmann’s result to bid-ask bounce or other reasons, Cooper (1999) uses filter rule to filter out the noise, whose lagged return does not move down or up by a minimum amount, and document large and consistent profits. Moreover, Cooper incorporates lagged trading volume into filter. Blume, Easley, and O’Hara (1994) suggest in theory that traders who include trading volume into their analysis perform better than those
Cooper (1999) uses filter rule technical analysis and tries to illustrate the relationship between price and volume. He mainly cites two papers with
‧
and Wang (1993) and Wang (1994).
Campbell, Grossman, and Wang (1993) develop a model in which the risk-averse utility maximizers play a role like market maker, and the market information is symmetric. When some investors sell their stocks out of exogenous pressure, like liquidity, other risk-averse utility-maximized investors are willing to accommodate the selling pressure, but will require a higher expected return. The deviation from the original price, because of liquidity trading, is not due to the change of the fundamental valuation of the stocks. Thus, their model predicts that price changes with high volume will tend to be reversed in the future.
On the other hand, Wang (1994) indicates another scenario. There might be a different relationship between price and volume if the market is with information asymmetry. In this case, there are two types of investors:
informed and uninformed investors. Informed investors can trade for informational or noninformational purpose. Wang hypothesized when informed investors trade on their private information, price change, with high trading volume, tends to continue in the same way in the future.
Simply put, Campbell, Grossman, and Wang (1993) hypothesize that in an information symmetric market where investors trade for liquidity purpose, return tends to be reversed when there is high trading volume.
Instead, Wang (1994) hypothesizes that in an information asymmetric market where investors are trading on their internal information, return tends to continue until all the information is disclosed when there is high trading volume.
Based on these predictions, Cooper (1999) uses the stocks for the top
‧
300 largest market capitalization in the U.S. stock market between 1962 and 1993. He collected information on the return and volume change in the previous week for his filter rule. He found that “decreasing-volume stocks experience greater reversals” and “increasing-volume stocks exhibit weaker reversals and positive autocorrelation” [Cooper (1999, p.901)]. The reversals from decreasing-volume were explained by Cooper as “period of portfolio rebalancing for both informed and uninformed investors” [Cooper (1999, p.920)], while the weaker reversals and positive autocorrelation from increasing-volume corresponds to the hypothesis of Wang (1994) proposed.
This thesis mainly uses the same methodology of Cooper’s. The size and in history the development of Taiwan stock market are not as large and long as of U.S. stock market. A small market could be influenced easily by some traders when they trade for internal information. If a market is not fully developed, and if a market is not well regulated, internal information would be often exploited to trade. Accordingly, I presume that Taiwan stock market is much more information asymmetric than U.S stock market, and expect that the same pattern as Cooper (1999)’s in Taiwan’s stock market might be established as well.
However, there is a related article indicating another different role for trading volume. Gervais, Kaniel, Mingelgrin (2001) assume that volume reflects a given stock’s visibility and can be the price premium. Their assumption is based on the claim of Miller (1977) and Mayshar (1983) that the holders of a stock tend to be more optimistic about the stock’s prospect.
Gervais, Kaniel, Mingelgrin find that even with little or no price change, volume can be used to predict future price change, i.e. volume itself can be an exclusive predictor, does not need to be accompanied with return. A stock
‧
This is because the stocks experiencing unusually high trading volume and low return tend to exhibit weaker reversals and even positive autocorrelation, not appreciate, in his finding; and those experiencing low trading volume and low return tend to be reversed, not depreciate. Nevertheless, the window set in of Cooper’s finding is one week; and that in Gervais, Kaniel, Mingelgrin’s finding is more than one week, or even many months. The result might vary if Cooper prolongs the window. This motivates this study to extend the window and to see whether Cooper’s finding holds for the next few weeks.
Like Lehmann (1990), Kelley (2004) also discusses the contrarian strategy. Although he only uses price change as a filter to form his portfolio, buying previous week’s loser and selling previous week’s winner, he still found evidence for the contrarian: return of winner tends to be negative, while return of loser tends to be positive. In addition, he suggests that, for the first four weeks, extreme winner is predicted to turn to negative return, and extreme loser to positive return. After four weeks of time, the contrary would disappear, and original extreme return would continue; namely winner still wins whereas loser still loses, which lasts for almost a year.
When stock price is reversed, it means overreaction of the original return; but when the stock price stops reversal and continues its original price trend, that implies underreaction. It remains unclear if the overreaction Cooper found will turn out to be underreaction when the window is prolonged, another justification that the window is extended in the study.
‧
(2004), using slightly-changed filter rule to examine Taiwan Stock Exchange Capitalization Weighted Stock Index future contracts. His empirical evidence shows that price trend is predictable when “the noise” is filtered out, and the combination filters of price and volume display more predictability than those only with price change. Chen (2006) use filter rule on Taiwan Stock Exchange Capitalization Weighted Stock Index. He obtained similar result to that found in Cooper’s, though not so significant.A few important findings emerge from this study. First, in a relatively short time horizon, lagged return is a better return predictor than lagged volume. But, in a longer time horizon, return tends to be negative if the lagged volume extremely decreases. Second, return and volume change in two weeks of time are more powerful to predict the future return than that in one week. Third, when we use return and volume change in two days of time to predict future returns, we find that decreasing-volume stocks experience greater reversals and increasing-volume stocks experience reduced reversals. Fourth, when we use return and volume change in two weeks of time to predict future returns, we find that a simultaneous increase in return and volume change would be followed by an increase in return.
Fifth, Taiwan stock market is likely to be more dominated by liquidity trading than information trading in recent decade.
The remainder of this thesis is organized as follows. Chapter II describes the related theories and presents the hypothesis to be tested.
Chapter III explains empirical strategy and methodology to be used. Chapter IV gives empirical findings. Chapter V concludes.