«Do Momentum Strategies Work?: - Australian Evidence Michael E. Drew, Madhu Veeraraghavan and Min Ye School of Economics and Finance Queensland ...»
Do Momentum Strategies Work?:
- Australian Evidence
Michael E. Drew, Madhu Veeraraghavan and Min Ye
School of Economics and Finance
Queensland University of Technology
GPO Box 2434
Brisbane Queensland 4001 Australia
Department of Accounting and Finance
The University of Auckland Business School
Private Bag 92019 Auckland New Zealand
Address for Correspondence:
Dr. Madhu Veeraraghavan
Department of Accounting and Finance The University of Auckland Private Bag 92019 Auckland New Zealand Tel: 0064-9-373 7599 (Ext: 85172) Fax: 0064-9-373 7406 Email: M.Veeraraghavan@auckland.ac.nz Abstract This paper investigates the profitability of momentum investment strategy and the predictive power of trading volume for equities listed in the Australian Stock Exchange. Recent research finds that momentum and trading volume appear to predict subsequent returns in U.S. market and past volume helps to reconcile intermediate-horizon “under reaction” and long-horizon “overreaction” effects. However, bulk of the evidence on this important relationship between past returns and future returns is limited to U.S. portfolios.
This study provides an out of sample evidence by examining the relationship between “trading volume” (measured by the turnover ratio) and “momentum” strategies in an Australian setting.
We document a strong momentum effect for the Australian market during the period 1988 through 2002 and find that momentum plays an important role in providing information about stocks. We also find that past trading volume predicts both the magnitude and persistence of price momentum. In summary, our findings are consistent with the U.S. evidence.
1. Introduction An enormous body of empirical research over the past 15 to 20 years has demonstrated evidence against the prediction of the Sharpe (1964), Lintner (1965) and Black (1972) Capital Asset Pricing Model (CAPM). This evidence suggests that the cross-section of expected stock returns are not sufficiently explained by their beta, the systematic risk of CAPM. The results indicate that variables such as firm size (Banz, 1981), earnings yield (Basu, 1977), leverage (Bhandari, 1988), the firm’s book value of equity to its market value (Chan, Hame and Lakonishok, 1991) and momentum (Jegadeesh and Titman 1993, Lee and Swaminathan, 2000) and more recently idiosyncratic volatility (Malkiel and Xu 1997, 2000) adequately explain the cross-section of average stock returns. This paper extends the methodology of Lee and Swaminathan (2000) to the Australian market. The motivation comes fromthe fact that the bulk of existing research relates to the United States and there is very little evidence from markets outside the United States.
Jegadeesh and Titman (2001, p 699-700) state “The criticism that observed empirical regularities arise because of data mining is typically the hardest to address because empirical research in non-experimental settings is limited by data availability. Fortunately, with the passage of time, we now have nine additional years of data that enable us to perform out-of-sample tests as well as to assess the extent to which investors may have learned from the earlier return patterns”.
In this paper we provide out-of-sample evidence by investigating the momentum strategies for equities listed in the Australian Stock Exchange. In addition, our objective is also to provide academic researchers and investors with a greater breadth and depth of understanding of the anomalies discovered in the area of empirical finance. Most research on the profitability of momentum strategies is based on U.S. data, particularly for the NYSE stocks1. For instance, Jegadeesh and Titman (1993) report that strategies that buy past winners and sell past losers realize significant abnormal returns over the 1965 – 89 periods.
Conrad and Kaul (1998) argue that momentum profits arise because of cross-sectional differences in expected returns rather than time-series return patterns. Barberis et al. (1998), Daniel et al. (1998), and Hong and Stein (1990) present behavioral models which suggest that the post holding period returns of the momentum portfolio should be negative.
Jegadeesh and Titman (2001) find that the momentum profits in the eight years subsequent to their 1993 period are similar to the profits in the earlier period. Thus, they argue that momentum profits cannot be due to data snooping biases.
Chordia and Shivakumar (2002) argue that profits to momentum strategies can be explained by a set of lagged macroeconomic variables and payoffs to momentum strategies disappear once stock returns are adjusted for their predictability based on these macroeconomic variables. There is also some evidence on momentum suggesting that bulk of the observed Jegadeesh and Titman (1993) find that when stocks are selected based on their past six month return and held for six months they realize a return of over 12 percent per annum.
momentum in U.S. individual stock returns is an industry effect. (See, for example, Moskowitz and Grinblatt (1999) and O’Neal (2000)).
Rouwenhorst (1998) and Dijk and Huibers (2002) provide evidence for European price momentum in the intermediate-horizon. Chan et al. (2000) examine the momentum effect based on individual stock market indices in 23 countries. They also find statistically significant evidence of momentum profits. However, Hameed and Kusnadi (2002) suggest that the factors that contribute to the momentum phenomenon in U.S. are not prevalent in the Asian Markets.
It is a well accepted fact that trading volume plays a minor role in conventional models of asset prices. However, recent research shows that past trading volume provides an important link between “momentum” and “value” strategies (Lee and Swaminathen, 2000;
Connolly and Stivers 2003). Several papers suggest that past trading volume may provide valuable information about a security. For instance, Lamoureux and Lastrapes (1994) examine the ability of volume data to shed light on the source of persistence in stock-return volatility. They find that the dynamics of daily return variance are due solely to daily persistence in the latent speed of arrival of information to the market, which leads to similar dynamics in the level of trading volume. Blume et al. (1994) present a model in which traders can learn valuable information about a security by observing both past price and past volume information. However, their model does not specify the nature of the information that might be derived from past volume.
In a similar vein, Lee and Swaminathan (2000) find that the effect of momentum appears more pronounced among high-volume stocks than low-volume stocks. They show that trading volume is only weakly correlated with traditional liquidity proxies and that the volume effect is robust to various risk adjustments. However, Scott et al. (2003) propose that the predicting power of the price momentum and trading volume is a result of the under reaction of investors to earnings news – an effect that is most pronounced for high-growth companies. Wang (1994) suggests that the dynamic relation between volume and returns varies depending upon the motive for trading by the “ informed investors”.
Wang (1994) also states that momentum in consecutive returns is likely if the primary motive for the informed investors’ trading in the former period has better information about the stock’s fundamentals. Conversely, a reversal is likely if the primary motive for the informed investors’ trading in the former period is changes in their outside investment opportunities.
In light of these findings, this paper attempts to examine the relationship between “trading volume” (measured by the turnover ratio) and “momentum” strategies for Australian equities.
The study of markets outside the US is interesting since our objective is to provide an out of sample evidence on the interaction between momentum strategies and future returns. Our findings show a strong momentum effect in Australian market during the period 1988 through 2002 and it plays an important role in providing information about stocks. Consistent with Lee and Swaminathan (2000), when conditional on past returns we find that low volume stocks generally do better than high volume stocks over the next 12 months. The price momentum premium is higher in median and low volume stocks. The remainder of this paper is organised as follows. The next section deals with the data and portfolio aggregation procedures. Section 3 presents the findings while section 4 concludes the paper.
2. Data and Methods Our sample consists of stocks listed on the Australia Stock Exchange during the period June 1988 through May 2002 with at least a year of data prior to the portfolio formation date.
Monthly stock and market returns, number of shares outstanding and traded and value of shares traded are obtained from the database maintained by the Securities Industry Research Center of Asia Pacific (SIRCA). We divide the sample into two periods because of data limitations. The first period is from June 1988 to May 1995. In this period we only have returns data for our sample. There are 296 eligible stocks in this period. This sample provides some useful information about the momentum effect.
The second period is from June 1995 to May 2002. This sample has available information on past returns, trading volume, market capitalization, and stock prices. It is used to test the relationship between trading volume and “momentum” strategies. Trading volume is defined as the average monthly turnover in percentage during the portfolio formation period, where monthly turnover is the ratio of the number of shares traded each month to the number of shares outstanding at the end of the month. There are 165 stocks that match the criteria.
We follow the approach of Lee and Swaminathan (2000) in constructing our portfolios. At the beginning of each month, all eligible stocks are ranked independently on the basis of past returns and past trading volume. The stocks are then assigned to one of five portfolios based on returns over the previous J months (where J=3, 6, 9 or 12) and one of three portfolios based on trading volume over the same period. All stocks are ranked in ascending order, so the top quintile based on the past return is the loser quintile and the bottom is the winner quintile. Similarly, the top treble based on the past volume consists of low volume stocks and the bottom consists of high volume stocks. The intersections resulting from the two independent rankings result in to 15 momentum-volume portfolios. In each month the strategy buys the winner portfolio and sells the loser portfolio in each volume group. We focus our attention on monthly returns of extreme winner and losers over the next K months (where K = 3, 6, 9 or 12) and next 5 years.
Similar to Jegadeesh and Titman (1993) and Lee and Swaminathan (2000), the monthly return for a K-month holding period is based on an equal-weighted average of portfolio returns from strategies implemented in the current month and the previous K – 1 months.
For example, the monthly return for a three-month holding period is based on an equalweighted average of portfolio returns from this month’s strategy, last month’s strategy, and strategy from two months ago.
3. Findings In this section the results for price momentum and volume-based price momentum strategies and the information content of trading volume are presented. Recall that our main objective is to examine the relationship between “trading volume” (measured by the turnover ratio) and “momentum” strategies in an Australian setting.
3.1 Momentum strategies At the beginning of each month, we rank all eligible stocks independently on the basis of past returns. The stocks are then assigned to one of five portfolios based on returns over the previous J months (J=3, 6, 9 or 12). In each month we long the winner portfolio and short the loser portfolio. This section reports the returns of the portfolio strategies described above over the two periods: from 1988 to 1995 and from 1995 to 2002. We report results for the top quintile portfolio of extreme losers (R1), the bottom quintile of extreme winners (R5), and one intermediate portfolio (R3). Our findings show a clear momentum effect in both sample periods.
Table I reports the results for the first sample period. Columns 4 through 7 report equalweighted average monthly returns over the next K months (K=3, 6, 9, 12). In addition, for each portfolio formation period (J) and holding period (K), we report the mean return from a dollar-neutral strategy of buying the extreme winners and selling the extreme losers (R5 – R1). For example, with a six-month portfolio formation period (J=6), past winners gain an average of 1.37 percent per month over the next six months (K=6). Past losers lose an average of 1.51 percent per month over the same time period. The difference between R5 and R1 is 2.88 percent per month. The differences in average monthly returns between R5 and R1 are significantly positive in all (J, K) combinations.
The last five columns of Table I report the annual event-time returns for each portfolio for five 12-month periods following the portfolio formation date. We find that momentum effect last until four years when portfolios are formed on past three months returns. The R5 – R1 portfolio yields a statistically significant positive return until year four. For other portfolios, we document a reversal in the sense that R5 – R1 is not statistically significant beyond year one (where J=6, 9 and 12). Nearly all R5 – R1 portfolios yield statistically insignificant returns, except for portfolio based on past 6-month returns. We also find that when portfolios are based on past 12-month returns, R5 – R 1 returns are negative post year one but significant in years four and five. Exhibit 1 shows this pattern graphically. Portfolio based on past 12month returns becomes loser in year 2. However, portfolio based on past 3-month returns is still a winner, although the abnormal return is much smaller and not statistically significant.
Our findings in this respect are consistent with Lee and Swaminathan (2000) who find that “the longer the estimation period for past returns, the more imminent the future price reversals” (Lee and Swaminathan 2000, 2026).