واکنش شدید در بازار سهام استرالیا: 1974-1997

Previous assessment of overreaction in the Australian equity market by Brailsford [Brailsford, T., 1992. A test for the winner–loser anomaly in the Australian equity market: 1958–87, Journal of Business Finance and Accounting, 19 (2) 225–241] and Allen and Prince [Allen, D.E., Prince, R., 1995. The winner/loser hypothesis: Some preliminary Australian evidence on the impact of changing risk. Applied Economics Letters 2, 280–283] finds no evidence of performance reversal in loser portfolios and no significant difference between the test period performance of winner and loser portfolios. This result is not consistent with evidence from overseas markets and warrants further examination. This study finds evidence of price reversal where monthly portfolio rebalancing is employed but the price reversal disappears when a buy and hold strategy is used. Further analysis reveals that the loser portfolio is dominated by small firms and that any abnormal returns are not exploitable given the lack of liquidity in small capitalisation Australian stocks. It is possible that the lack of consistency between Australian and US research can be explained by the different time periods examined in these studies.

In their seminal work, DeBondt and Thaler (1985) report that a portfolio of US stocks which perform worst (losers) over an initial 3-year period (rank or portfolio formation period) tend to perform best in the subsequent 3-year period (test period). A similar performance reversal is evident for the rank period winner portfolio, which goes on to perform worst in the subsequent test period. This suggests that stock market investors overreact, that excessive optimism or pessimism causes prices to be driven too high or too low from their fundamental values, and that the overreaction is corrected in a subsequent period. It also suggests an easily implemented profitable trading strategy of buying losers and selling winners and has important implications for the validity of the efficient market hypothesis (EMH) which asserts that all publicly available information is incorporated into asset prices. Chan (1988) argues that DeBondt and Thaler (DT) fail to control for time-varying risk, and when properly controlled the overreaction disappears. Ball and Kothari (1989) make a similar claim. However, DeBondt and Thaler (1987) and Chopra et al. (1992) provide evidence that differential risk cannot explain the performance reversal of winner and loser firms. Zarowin (1990) claims that firm size can explain this overreaction. He argues that losers tend to be smaller than winners and when size is controlled there is no significant difference in test period performance. However, Chopra et al. (1992) find that the overreaction persists after controlling for size as do Albert and Henderson (1995) after correcting potential biases in Zarowin's methodology. Using UK data, Clare and Thomas (1995) conclude that the difference in performance between the loser and winner portfolios is probably due to the size effect. Dissanaike (1997) also uses UK data and finds in favour of the overreaction hypothesis after limiting his study to the larger listed companies. Conrad and Kaul (1993) assert that the overreaction observed in this type of study is due to the process of cumulating single period returns over long periods where these single period returns contain errors caused by bid–ask spread bias and infrequent trading. However, Loughran and Ritter (1996) dispute the methodology employed by Conrad and Kaul and show that their conclusions are not valid after correcting the methodology. Despite the passage of time and several methodological refinements, the conclusions of DT using the basic methodology still appear to hold. While the bulk of research on this issue has been undertaken using US data, there have been a handful of applications in other markets. For example, Clare and Thomas (1995) examine the UK market and find evidence of overreaction, but conclude that this can be explained by the small firm effect. However, Dissanaike (1997) finds strong evidence of overreaction amongst the larger companies listed on the UK exchange. DaCosta (1994) presents evidence of overreaction in stocks listed on the exchange in Brazil as do Leung and Li (1998) in the case of the Hong Kong stock market. Kryzanowski and Zhang (1992) study the Canadian market and find continuation behaviour for 12- and 24-month test periods, but evidence of reversals for 36- to 120-month test periods though the performance of the arbitrage portfolio is only significant at the 10% level. The results are even weaker when risk is controlled. There is also evidence of overreaction over much shorter periods than that studied here. Chang et al. (1995) report a price reversal in the Japanese stock market for portfolios formed on 1-month returns. Bowman and Iverson (1998) reach a similar conclusion using New Zealand stocks based on single-week returns. When the basic DT methodology is employed within and across markets, there appears to be a consistent finding of overreaction and reversal. One apparent exception is the Australian market where both Brailsford (1992) and Allen and Prince (1995) using the basic DT methodology reports no reversal in performance by the loser portfolio and no significant difference in the test period performance between the loser and winner portfolio. As an exception, the Australian market warrants further examination. This study makes a number of contributions towards resolving this apparent anomaly. First, it employs an unbiased approach to calculating returns. Dissanaike (1994) asserts that much previous research employs an invalid or unrealistic method of calculating returns which results in errors in portfolio formation and measurement of test period performance. Second, this study reports and analyses the test period performance of all intermediate portfolios, which is rarely undertaken by authors. Third, this study reports risk-adjusted portfolio returns, which Brailsford argues as problematical. While Allen and Prince do report risk-adjusted returns, their requirement that firms be listed for a minimum of 9 years introduces a survivorship bias and possibly a bias towards larger companies which may adversely affect their results. Fourth, this study explores the role of size in the performance of the winner and loser portfolios. Fifth, this study extends the data set used by Brailsford a further 10 years from 1988 to 1997, and extends the data set used by Allen and Prince another 6 years from 1992 to 1997.

Using an extended Australian data set and a rebalancing approach to calculating market-adjusted test period portfolio returns, this study finds evidence of performance reversal for both the rank period loser and winner portfolios and positive abnormal returns for the arbitrage (loser–winner) portfolio. However, consistent with Brailsford (1992), this result largely disappears when a buy and hold strategy is employed. After an appropriate adjustment for risk, the performance reversal experienced by the loser portfolio using the rebalancing approach is considerably reduced. Despite this, a significant positive abnormal return to the arbitrage portfolio is still evident. However, this again disappears when a buy and hold approach is used. Further analysis indicates that the loser portfolio has an average capitalisation significantly smaller than the remainder of the data set and that the loser portfolio is dominated by small firms. While the results presented here suggest that a small firm effect may be driving the positive abnormal returns to the arbitrage portfolio, further research is required to disentangle the small firm effect from any overreaction effect. Regardless of the underlying source of the abnormal returns, the results here suggest that these abnormal returns are not exploitable in the Australian equity market due to the lack of liquidity associated with small capitalisation stocks and the transaction costs associated with monthly portfolio rebalancing. While this paper has gone some way to explaining the results of earlier Australian research by considering the impact of alternative methodologies on the visibility and genuineness of price reversals in the Australian equity market, there remains a fundamental difference between Australian and US studies of this phenomena. That is, when using the basic DeBondt and Thaler (1985) methodology where no adjustments are made for risk, size or other factors, US studies consistently demonstrate price reversal in the winner and loser portfolios whereas Australian studies do not. It is possible that this may at least partly be explained by the relatively short and recent period used in Australian studies. While the data set used here covers the period from 1974 to 1997, comprehensive US studies go back to 1926. Chen and Sauer (1997) find that price reversal is not consistent over that long period of time when raw returns are analysed. Dividing the time period into four sub-periods (pre-war, 1940s–1950s, pre-energy crisis, post-energy crisis), they find evidence of significant price reversals in the pre-war and pre-energy crisis periods but not in the other two periods. The post-energy crisis period (1971–1992), for which Chen and Sauer found little evidence of price reversal, corresponds closely with the period studied here using Australian data. When viewed against the results of Chen and Sauer for the corresponding period, the Australian and US results based on raw returns are quite consistent.