Individual Price Momentum

Published Date: 02 Nov 2017

Momentum is most commonly defined by the process of buying stocks with high returns in the recent past as it is believed that it will have higher future returns and selling stocks with low returns in the past. This particular process is believed to generate excess abnormal profits and is a true challenge to the weak form market efficiency. Fama (1998) had state the momentum strategies as an ‘open puzzle" and there is an extensive list of research of momentum profits. Nearly every academic had used multiple formations of momentum to provide evidence of the stock return predictability based on a variety of firm- specific variables and some of them had presented momentum in its various forms. Discussion of major momentum profits evidences are described below according to its firm characteristics.

2.1 Individual Price Momentum

DeBondt and Thaler (1985) were the first to identify about the momentum anomalies. They reported that that past winners (losers) ultimately become losers (winners) when investing in three to five years horizon. They proceeded with their research by comparing the performance of two groups of companies classified as extreme losers and extreme winners. For each year since 1933, they formed portfolios of the best and the worst performing stocks over the previous three years and then calculated the portfolios return over the five years following portfolio formation. They showed that over 3 to 5 years holding periods, stocks that performed poorly over the previous years achieved higher returns than stocks that performed well over the same period. To be more precise they concluded that "thirty- six months after portfolio formation, the losing stocks have earned about the 25 % more than the winners, even though the latter are significantly more risky". Hence they argued that the difference in returns was due to overreaction in security prices in the sense that extreme losers become too cheap and bounce back and on the other hand extreme winners become too expensive and earn lower subsequent returns.

The most celebrated study in the academic literature which documents the momentum phenomenon was by Jegadeesh and Titman (JT) (1993). They believed that the concept of buying stock with high returns (winners) and selling stock with low return (losers) exhibit momentum profits in the medium term. Jegadeesh and Titman had examined the American Stock Exchange (AMEX) for the period of 1965 to 1989, with formation and holding periods between 3 and 12 months. Then they ranked stocks in an ascending order based on their 3 to 12 month past returns. Based on these rankings, JT had formed ten equally weighted deciles portfolios. The portfolio with the highest return was called the "winners" decile and the portfolio with the lowest return was called the "losers" decile. In each overlapping period, the strategy was to buy the winner decile and sell the loser decile with the holding period of 3 to 12 months. Hence their finding had added a new angle to the above history by proving that past winners, on an intermediate horizon of 3 to 12 months continue to outperform past losers. Empirically they found that an equally weighted portfolio stocks which finances the purchase of the 10% highest performing stocks with the short sale of the 10% worst-performing stocks, yields returns of approximately 1% per month, when using six-month formation and holding periods.

Conrad and Kaul (1998) presented striking evidence suggesting that the momentum profits are attributable to the cross-sectional differences in expected returns rather than to any time- series dependence in returns. They argued that that if realized returns are strongly correlated to expected returns, then past winners (losers) that have higher (lower) returns tend to yield higher (lower) expected returns in the future. They investigated stocks listed on the NYSE and AMEX from periods 1962 to 1989 in a way to make their research comparable with the one by Jegadeesh and Titman (1993). They had examine momentum strategies for which the length of the formation and the holding periods are identical; ranging between 1 week and 36 months and their selection was based on based on the unconditional expected returns of individual stocks and they had separate the cross-section and time-series perspectives of the returns in an additive form. Hence with the exception of the 1 week/ 1 week strategy, they document that all other strategies are profitable up to and including the 18 month/ 18 months strategy. Thus, Conrad and Kaul confirmed the success of momentum profits on the medium horizons as documented by Jegadeesh and Titman (1993). However, empirical studies (e.g., DeBondt and Thaler, 1987; Grundy and Martin, 2001; Jegadeesh and Titman, 2001) conclude that the link between future price movements and past stock returns is unlikely to be explained by cross-sectional variation in mean stock returns and Jegadeesh and Titman(2002) show that Conrad and Kaul’s results suffer from small sample biases, and when these biases are corrected for in the tests, the variation in mean returns explains very little of the momentum profits.

Grundy and Martin (1998) studied momentum profits using the Fama-French three factor risk -adjusted returns model. They investigated the NYSE and AMEX stocks over the period of 1966 to 1995 and they documented profitability of more than 1.3 % per month using momentum strategies. However, Grundy and Martin (2001) showed that the Fama-French three-factor model cannot explain this price momentum effect. This is because given Fama and French’s (1996) finding; momentum phenomenon is recalcitrant to their factor model. Momentum profits to a large extent depend on the autocovariances and cross-autocovariances of stock return and Fama-French factors help explain the cross-section of expected returns, they may not be important in determining the (cross-) autocovariances of returns.

Looking at the international context, Rouwenhorst (1998) showed that momentum strategies examined by Jegadeesh and Titman (1993) are not confined to the United States market but instead it is also profitable outside the United States. He believed that international equity markets exhibit medium term return continuation. Hence, Rouwenhorst (1998) had replicated JT by examining 12 European countries for the period 1980 to 1995 and hereby forming relative strength portfolios. Therefore, Rouwenhorst concluded that "an internationally diversified portfolio of past Winners outperformed a portfolio of past Losers by about one percent per month". His finding was that return associated with momentum strategies are very close to the return reported by JT for the U.S. market, although that t –statistics computed for European stocks are 4.02 compared to the U.S. market t-statistic which is 3.07. In short for the six -month/six –month strategy, European market earns 1.16 % compared to U.S market with 0.95 %. Moreover, Rouwenhorst concluded that momentum anomaly is present in all countries and holds for both large and small firms, although for smaller firm it is stronger than larger firm. The European evidence is very closely similar to that finding for United States by Jegadeesh and Titman and hence because of this similarity the former also document that the returns on European momentum portfolios are significantly correlated with relative strength strategies in the United States.

2.2 Industry Momentum

Moskowitz and Grinblatt (1999) evaluated industry momentum as the source of much of the momentum trading profits over intermediate investment horizons (6 to 12 months). They formed value-weighted industry portfolios and ranked stocks based on past industry returns and found that high momentum industries outperform low momentum industries in the six-months after portfolio formation. To see the extent to which the industry return contributes to momentum profits, they had examined the performance of a "random industry" strategy. They replaced each firm in the winner and loser industries with other firms that are not in these industries, but have the same ranking period returns as the firms that they replace. The random industry portfolios have similar levels of past returns as the winner and loser industry portfolio. Thus, they concluded that that the profitability of a momentum strategy is attributable primarily to momentum in industry factors. They argued that when stocks from past winning industries are bought and stocks from past losing industries are sold, the strategy appears highly profitable after controlling for size, individual stock momentum, the cross sectional dispersion in mean returns and potential microstructure influences.

Several other studies had tested the claim made by Moskowitz and Grinblatt (1999). Bacmann, Dubois, and Isakov (2001) documented profitability of momentum strategies of G-7 countries. All the G-7 countries are profitably linked to industry momentum except for Japan. They also found that the profits of momentum strategies are driven by the cross-sectional dispersion of stock indexes especially during expansion periods. Moreover Swinkels (2002) result also indicates that there industry momentum exhibit in America and Europe except in Japan. However, Griffin and Karolyi (1998) had examined the industry momentum in international asset returns for global portfolio diversification strategies by using stock listed on the Dow Jones World Stock Index and found that their result confirm the findings of Heston and Rouwenhorst (1994) that less than 4% of the variation in country index returns can be explained by their industrial composition. Thus they fail to find the industry momentum claim by Moskowitz and Grinblatt (1999) in international markets.

2.4 Country Momentum

Country momentum can be defined as the process of buying stocks in countries that have performed well over a certain period and selling stocks in the countries that have underperformed. Richard (1997) tested the national stock market indices of 16 countries using a value weighted model. As a result, he found a momentum effect of 0.57 per cent per month at six month horizon is statistically insignificant. Thus he concludes that "there is no evidence that loser countries are riskier that winners countries either in terms of standard deviations, covariance with the world market or other risk factors". On the other hand Chan et al (2000) and Bhojraj and Swaminathan (2001) concluded that momentum on a country levels exist. Chan et al (2000) had used Data Stream market indices and a value weighted portfolio where they found significant excess momentum return of 0.46 per cent per month whereas Bhojraj and Swaminathan (2001) find significant excess return for sample of 38 countries they investigated.

Rouwenhorst (1999) in his study examined the size effect of twenty emerging markets using equally weighted portfolios from period 1982 to 1997. Though his analysis, he concluded that an international portfolio of small stocks outperformed international portfolio of large stocks by an average of 8.28 percent point annualized and value stock outperformed growth stock. Hence he found little evidence of country momentum.

2.5 52-Weeks High Momentum

The momentum literature recently considers an alternative momentum strategy based on the proximity of a stock's price to its 52-week high price and only a few studies examine the profitability of the 52-week high momentum trading strategy. George and Hwang (GH) (2004) were the first to investigate that the ratio of a stock close’ price to its 52 week high price is a good predictor of future returns. Following evidences that stock return do not follow random walk, and that return are predictable, they were of viewed that with stock’s current price, readily available as piece of information –the 52 week high price-largely explains the profits from momentum investing . Therefore to test their theory, they used the U.S stocks from 1963 to 2001, created equally weighted portfolios and ranking the stock, from one nearest to their 52 week high price to the one that were furthest away. The top thirty percent began the winner portfolio and the bottom thirty percent the loser portfolio. Hence they found that by buying stock in the winner portfolio and the selling the stocks in the loser portfolio, they were able to produce positive abnormal returns. Moreover, George and Hwang had also compared the momentum strategies of Jegadeesh and Titman (1993) and Moskowitz and Grinblatt (1999) to a strategy based on the nearness of a stock’s price to its 52-week high. They found "that nearness to the 52-week high is a better predictor of future returns than are past returns, and that nearness to the 52-week high has predictive power whether or not stocks have experienced extreme past returns’. Hence, their result shows that price levels determine momentum effects more than past price changes.

Marshall and Cahan (2005) had examined the effectiveness of GH 52-week high momentum trading strategy in the Australian Market as they wanted to investigate whether this strategy persists in the stock outside United States. They tested the stock listed on the Australian Stock Exchange (ASX) based on the same holding strategy of six months with data from 1990 to 2003. Hence they had found that the 52 week high momentum strategy is very profitable on small and large stocks and on liquid and illiquid stocks. They also found that the risk adjusted returns are significant and the positive abnormal statistically significant returns are even higher than the price momentum of Jegadeesh and Titman (1993) and the industry momentum of Moskowitz and Grinblatt (1999). That is the Australian market 52 week price momentum had generated returns of 2.14 % per month, as compared with 0.59% and 0.16% for the price and industry momentums, respectively. However they had used stocks in the portfolio that were approved to be short sold on Australian market and they did not examine whether the 52-week high momentum profit reverses in the long run, thus providing no test on an important implication of the behavioral models.

On a more global scale, Du (2008) tested the 52 week high strategy in the international stock market indices using a sample of 18 developed countries that covers the period from 1969 t0 2004. Replicating the methodology of George and Hwang (2004), he proved that price levels dominate past returns in terms of predictive power than the JT (2003) when the 52wk high strategy is paired with the momentum strategy. This is consistent with the George and Hwang (2004) findings that is based on individual stocks. Therefore, when comparing the two individual strategies, Du provided empirical evidence that momentum performs better than the 52 week high strategy even after risk adjustment. However he contradicted George and Hwang 2004 by proving that the continuation of returns coexists with the eventual reversal of returns for both the momentum and 52wk high strategy. Hence he suggested that an overreaction can still occur when investors correct their initial bias.

On the other hand, Ching-Hua Yu and Yen-Chih Liu (2012) is one of the few studies that had compared the different momentum strategies’ performances. They had compared the 52-week high price momentum generated by George and Hwang (2004), the traditional momentum strategy of Jegadeesh and Titman (1993) and the earning momentum strategy by Chan, Jegadeesh and Lakonishok (1996) in which the latter documented that an unexpected high earning reported by a firm will outperformed an unexpected low earnings hereby resulting in price momentum. They had tested the robustness of these momentum strategies in an emerging market namely the Taiwan’s stock market for the time period of 1995 to 2009. Thus they had provide empirical evidence by doing a cross sectional regression analysis introduced by Fama- Macbeth (1973) and portfolio analysis for the comparison of the three momentum strategies and hence concluding that there is significant earnings momentum up to 12 months in the Taiwan Stock Market. However the price momentum and the 52 week high momentum do not exist in the Taiwan Stock Market as these had not reached statistical significance.

Moreover Liu et al (2010) also analysed the robustness of the 52 week high momentum strategy in the international stock market based on the methodology of George and Hwang (2004). He analyzed 16 markets out of which sample of data were taken in thirteen European stock markets and three in Asian stock markets. However, the sample period for the different countries varies because they required at least 50 stocks in each month and some of the countries do not have so much stocks data available. An analysis of the 52 week high momentum as well as a comparison with the Jegadeesh and Titman (JT)’ price momentum strategy was also made. As a result, they documented that the 52- week high momentum strategy is robust in the ten out of the sixteen stocks market for which nine of the European stock market showed statistically significant GH momentum profits with average monthly return of 6.0% to 1.0% and it also showed the presence of the JT momentum strategy. On the other hand, out of the Asian stock market, Hong Kong proved to be having the GH and JT momentum strategies, with average monthly return of 1 % and 0.68 % respectively. Moreover, they also proved that GH and JT momentum strategies are highly correlated with correlation of 0.75 in ten stock markets hereby denoting that both momentum strategies co- exist in the stock markets.

2.6 Theoretical Explanations

It can be noted that there are various empirical evidences from different studies of abnormal profits using momentum profits strategies but however there is no single explanation of why momentum works. Therefore, most formal attempt to explain this predictability of stock returns utilizes findings in behavioral finance.

Hong and Stein (1999) provided a seminal model of information diffusion in relation to changing pricing accuracy and hence based their explanation on the behavior of different sets of traders and how they interact with each other. They model a market populated by two types of traders: "newswatchers" and "momentum traders" which leads to underreaction at short horizons and overreaction at long horizons. Hong and Stein (1999) assumed that newswatchers make their forecast based on actions that affect the fundamentals alone whereas momentum traders use only simple univariate models based on past price movements to forecasts future price movements. Other assumption they made is that the news is fed slowly to newswatchers allowing the market to adjust slowly to new announcements and this very adjustments made are only underreactions rather than overreactions to the new information. On the other hand, by this price movements, momentum traders react in a way which create a flurry of trading that ends with an overreaction to the announcement.

George and Hwang (2004) examine the 52-week high because they believed that their models predict that traders are slow to react, or overreact to good news. They were of viewed that if a stock whose price is at or near its 52-week high is a stock for which good news has recently arrived. Hence using the 52 week as a benchmark and through their analysis, they came up by concluding that on nearness to the 52-week high price as a result of positive announcement traders will be reluctant to buy but the new information is eventually incorporated into the price creating a continuation and with a negative announcement the price are drove away from its 52-week high price as traders are initially unwilling to sell the stock at prices that are as low as the information implies. The information eventually is assimilated and the price falls. Therefore George and Huang compared this scenario to a stock that is neither close nor far from its 52-week high and stating that "at prices that are neither near nor far from the 52-week high, priors adjust more quickly and there is no pronounced predictability when information arrives".

Baberis, Shleifer and Vishny (1998) presented a research on the investor form belief. They uncovered two regularities that is the underreaction of stock prices to news such as earnings announcements, and overreaction of stock prices to a series of good or bad news. They created their one investor and one asset theory basing themselves on the study of Tversky and Kahneman (1974) on the tendency of experimental subjects to view events as typical or representative of some specific class and to ignore the laws of probability in the process. In their model, the earning of asset follow a random walk but the behavior a firm’s earnings moves between two "states" or "regimes".

They were of viewed that the asset can be of 2 states; mean reverting and trend focused that is are likely to rise further after an increase. Hence, the investor observes earnings, and uses this information to update his beliefs about which state he is in. That is if the earnings increases and again increases, it shows that the stock is in trend, but if the earning increases and then decreases the investor would believe it was a mean reverting state. Hence using this theory, investors are able to predict the returns of stocks.

Daniel, Hirshleifer, and Subrahmanyam (1997) also constructed a model of investor sentiment based on psychological biases namely "investors overconfidence" about the precision of private information and "biased self-attribution" – a rule by which investors essentially believe "heads I win, tails it’s chance". In other words it can be defined as the tendency of investors to attribute positive outcomes to skill and negative outcomes to bad luck. Their theory also state that investors overreact to private information signals and underreact to public information signals. Hence, these particular behaviors cause the market to overreact in the short term and correct themselves in later periods.

De Long, Shleifer, Summers and Waldman (1990a) and Shleifer and Vishny (1997) presented a research stating that arbitrage opportunities is limited because movement in investor sentiment are in part unpredictable. Hence, this leads to arbitrageurs betting against mispricing run the risk atleast in the short run and moving the prices further away from their true fundamental values. As a result "noise trader risk" can lose money in the short run and in case when arbitrageurs are risk adverse and run the risk of losing funds under management when performance is poor, the risk of deepening mispricing reduces the size of the position they take. Therefore, both papers concluded that miss pricing can exist and arbitrage fails to eliminate the mispricing completely. Moreover, they also added that investor sentiment affects security prices in equilibrium.

Within the extant literature, it can be noted that there is a wide ample of evidences on the profitability of momentum strategy in most countries’ stock market but only few studies on the profitability of the GH 52-week high momentum trading strategy. Therefore, the purpose of this paper will contribute to the above literature by providing a detail analysis and empirical evidence of the economic feasibility of the 52-week high momentum trading strategy. For this paper, we will examine the model developed by George and Hwang (2004) and an analysis will be made so as to know if the 52 –week high price is better predictor of future returns than the price momentum of Jegadeesh and Titman (1993).