Asian Review of Financial Research Vol.31 No.1 pp.117-155
Rank Momentum Strategies in the Korean Stock Market
Key Words : Momentum,Rank,Salience,Underreaction,Mispricing
I demonstrate that the rank momentum strategy advanced by Chen, Chou, Ko, and Rhee (2016, Nonparametric momentum strategies, working paper, henceforth CCKR) is profitable in Korea. The strategy constructed on the basis of average past ranks of daily returns generates significant momentum profits for up to four years following portfolio formation. Further, I find evidence that the profitability is attributed to investor underreaction to non-salient information embedded in stock prices. Rank measures capture the non-salient component in stock prices largely neglected by investors, and therefore, the rank can predict the future changes in stock prices. The main findings of this study are as follows. First, rank momentum strategies generate significant returns for holding periods ranging from six months to four years subsequent to portfolio formation. In particular, the risk-adjusted return of the rank momentum strategy constructed by buying stocks with high average ranks and short selling those with low average ranks is 0.81% per month (t-statistics 2.90) when the long-short portfolio is held over six months. The profitability of rank momentum strategies is robust to controls of various cross-sectional effects such as size, book-to-market, illiquidity, and idiosyncratic volatility. Second, rank-based momentum strategies even work better than the price momentum strategies proposed by Jagadeesh and Titman (1993, Returns to buying winners and selling losers: implications for stock market efficiency, J. Finance 48, 65-91, henceforth JT). For example, over one year holding period following formation, the rank momentum earns an average monthly profit of 0.44%, while the profits of JT price momentum are even negative. The rank momentum strategies also outperform the 52-week high momentum strategies suggested by George and Hwang (2004, The 52-week high and momentum investing, J. Finance 59, 2145-2176). Third, rank momentum profits tend to fall for longer holding periods but are not reverted to become negative afterward in contrast with that of JT price momentum. The literature documents that investor overreaction leads to long-term reversals but investor underreaction does not. In this regard, the lack of reversals indicates that the rank momentum profits are not driven by the overreaction, but by the underreaction. Fourth, rank momentum profits primarily come from loser stocks' underperformance. Fama-MacBeth (1973, Risk, return, and equilibrium: Empirical tests, J. Polit. Econ. 81, 607-636) regressions results reveal the short-leg of the long-short strategy yields persistent negative returns over holding periods ranging from six months to two years, while the profit of long-leg is positive only for the first six months. Consistent with the earlier findings in my study, this result also indicates that the rank momentum is associated with stock mispricing, which has not been eliminated by arbitragers. In particular, the rank momentum is closely related to stock overpricing and the resulting reduction. Arbitragers face greater impediments when they short-sell an overpriced stock than when they purchase an underpriced stock. The difficulty of short-selling deters arbitrage that reduces the overpricing. Accordingly, the underreaction of loser stocks to bad news is more persistent than underreaction of winner stocks to good news, which leads to persistent rank momentum profits for loser stocks. Lastly, the profitability of the rank momentum is strong among stocks with weak salient features, where the salience is proxied by the presence of extreme price changes. This evidence indicates that the rank momentum is the investors' underreaction, rather than overreaction, is an underlying cause of the rank momentum. The literature including Bordalo, Gennaioli, and Shleifer (2013, Salience and asset prices, American Economic Review 103, 623-28) and CCKR asserts that investors tend to overreact to salient information and underreact to non-salient information. Weaker rank momentum in stocks with higher salient features lends support to the conjecture that the rank momentum profits are attributed to investor underreactions, not due to investor overreactions. Overall, our empirical findings for the Korean stock market is quite consistent with that for the U.S. stock market. As in CCKR, we provide ample evidence that rank captures the non-salient information embedded in stock prices, which is largely due to investors' underreaction, and the profitability of rank momentum strategy primarily comes from the underreaction to bad news among loser stocks. This study contributes the literature on the momentum phenomenon in the Korean stock market. I explore the sources of the profitability of rank momentum strategies by comparing it with that of the traditional JT momentum strategies. Moreover, this study sheds light on the understanding of investors' reactions to non-salient information in the Korean stock market. Investors tend to underreact to non-salient news, and the overpricing, which is due to the underreaction, and the resulting reduction produces momentum in stock prices.