اثر تکانه بر سهام دارایی واقعی در چین : مدارک و شواهد از سطح عملکرد شرکت

This paper investigates the momentum effects under different firm performance levels for Chinese real estate stocks using quantile regression with a dummy variable estimator. This paper finds that regardless of the momentum horizon, the momentum effects are positive under high-performing individual stocks, but they are negative under low-performing individual stocks. While prior literature only finds that this asymmetric phenomenon appears under different market states, and the findings on different horizons are inconsistent. Furthermore, this paper finds that the positive (negative) momentum effect under high (low) firm performance levels is stronger than that under bullish (bearish) markets. This implies that superior (inferior) fundamental business performance and bullish (bearish) markets can cause the stock prices to go up (down); however, the effect of the former is stronger than that of the latter. Moreover, this paper finds that the relation between future returns and past turnover ratios is positively correlated under high-performing stocks, but negatively correlated under low-performing stocks. Based on the above findings, this paper regards past turnover ratios as a leading indicator of stock returns and suggests two profitable investment portfolios which are superior to the average returns of real estate stocks.

Studies on the momentum effect suggest that stock returns relate to their own lagged cumulative stock returns. The literature refers to this phenomenon as the momentum effect, whereas the lagged cumulative stock returns are called the momentum factors (Jegadeesh and Titman, 1993 and Jegadeesh and Titman, 2001). If the momentum effect is positive, investors can earn abnormal profits through a momentum strategy; namely, buying past winners and selling past losers (Jegadeesh and Titman, 1993). On the other hand, if the momentum effect is negative, investors can employ a contrarian strategy of selling past winners and buying past losers to earn profits (DeBondt and Thaler, 1985). There is a body of literature to explain the momentum effect. The first key factor of the momentum effect is the industry factor (Moskowitz and Grinblatt, 1999). After controlling for the industry factor, Stevenson (2002) studies real estate securities in 11 markets1 and finds a positive momentum effect. Notably, Stevenson (2002) and Lee and Kuo (2010) indicate that the horizon of the momentum effect on real estate securities is different than that on general securities. Jegadeesh and Titman, 1993 and Jegadeesh and Titman, 2001 and Conrad and Kaul (1998) indicate that the long-horizon (24–36 months) momentum effect is negative for general securities, while Stevenson (2002) finds the long-horizon momentum effect is positive for real estate securities. This difference has induced more studies to identify the momentum effect on real estate securities (Hung and Glascock, 2008 and Lee and Kuo, 2010). For example, Chui et al. (2003) finds that momentum factor is the main determinant of REITs returns for the US stock market. Moreover, Hung and Glascock (2008) investigate the REITs momentum effect under different market states and find that the REITs momentum effect is positive under up markets and negative under down markets. This finding is consistent with the conclusion of general stocks examined by Cooper et al. (2004). However, Lee and Kuo (2010) divide market states into bullish, range-bound, and bearish markets, and partly support Hung and Glascock's finding. However, Lee and Kuo find that the above asymmetric phenomenon only persists for a period of 3 to 6 months, which is shorter than Hung and Glascock's results. Although numerous studies investigate the momentum effect on real estate securities, most studies focus on REITs and clarify the reasons, such as firm size (Chui et al., 2003) and dividend distribution (Hung and Glascock, 2008), why the momentum effect differs across REITs and general securities. However, little is known about the reasons why the momentum effect differs across real estate stocks and general stocks. Hence, we will concentrate our attention on real estate stocks and focus us the Chinese market. China has become one of fastest economic growth countries in the last 10 years and was recognized as one of the BRICs2 by international investment banks (Goldman Sachs) in 2001. In addition, the Chinese real estate industry has also experienced the booming process.3 However, there are numerous unique features in the Chinese real estate industry, and these features have attracted the attention of both scholars and practitioners.4 First, people in China can only obtain a land-use-right, but cannot obtain land ownership. Second, the Chinese government often employs macroeconomic policies to influence the real estate industry. Third, China experiences opposite trends in its stock and real estate markets5 (Zhang and Fung, 2006). Fourth, China lacks REITs. At this stage, the China Securities Regulatory Commission (CSRC) has not approved the issuing of REITs, and there are few studies about the momentum effect in REITs. Furthermore, Eichholtz (1996) indicates that international diversification can reduce the risk of real estate stock portfolios even more than it can reduce the risk of general stock and bond portfolios. Remarkably, China is also one of the few countries whose stock markets were negatively correlated with the American stock market (Kang et al., 2002). From the viewpoint of international investors, China real estate stocks offer a good approach to indirectly invest in China's real estate market, considering the restrictions on direct investment in Chinese physical real estate by foreign investors. Hence, the research on the momentum effect in Chinese real estate stocks is not only interesting to finance scholars but also timely for investment practitioners (Lee and Kuo, 2010). The related literature about the momentum effect in Chinese real estate stocks is still limited, and the momentum effect in Chinese real estate stocks is unique compared to that in Chinese general stocks and other markets' real estate securities. First, to the best of our knowledge, Lee and Kuo (2010) is the only other investigation on the momentum effect in Chinese real estate stocks. Second, Lee and Kuo find that a positive momentum effect only occurs at 3 to 6 months; whereas Kang et al. (2002) demonstrate that a three- to 12-month momentum effect is positive for Chinese A-shares, and Stevenson (2002) indicates that a three- to 60-month momentum effect is positive for real estate securities in 11 markets. Thus, the momentum effect in Chinese real estate stocks is unique and gives rise to the research motivation of this paper. The second important factor for the momentum effect is the horizon. Jegadeesh and Titman, 1993 and Jegadeesh and Titman, 2001 and Conrad and Kaul (1998) indicate that the short- (1 month) and long-horizon (24–36 months) momentum effects are negative, but the medium-horizon (3–12 months) momentum effects are positive in the US stock market. Daniel et al. (1998) and Hong and Stein (1999) propose two leading behavioral models to explain different momentum effects of different horizons. The former is based on investors' overconfidence, but the latter is based on investors' underreaction and overreaction.6 Similarly, both indicate that the horizon of the momentum effect is related to investor sentiment. Furthermore, the third important factor for the momentum effect is market state. Cooper et al. (2004) indicates that different market states will cause different momentum effects. Cooper et al. regards up and down markets as different market states. They find that investors are overly optimistic to new information during up markets, and thus, cause market overreaction. Therefore, the momentum effect is positive on stock returns during up markets but has no effect during down markets. As a result, Cooper et al. suggests that after considering the different market states, the empirical results of the momentum effect become more robust. Furthermore, Cooper et al. applies positive or negative previous average returns of the market index to discriminate up or down markets,7 and up or down markets also mean positive or negative market performance levels. Thus, the research of Cooper et al. can be regarded as a concept that divides different market performance levels to identify different momentum effects. Moreover, Lee and Kuo (2010) investigate the momentum effect of real estate stocks by dividing market states into bullish, range-bound, and bearish markets8 and find an asymmetric phenomenon of the momentum effect, which is a positive momentum effect under bullish markets but a negative under bearish markets. However, compared to the methods of Cooper et al. (2004) and Lee and Kuo (2010) for dividing different market performance levels, the first purpose of this paper is to further extend the prior literature by dividing firm performance levels9 to identify the momentum effects. On comparison between dividing market and firm performance levels, the former represents the response of aggregate investor sentiment in different market conditions, while the latter represents the response of individual investor sentiment in different firm performing levels. For the method of dividing market performance levels, since some investors do not have profits (losses) under bullish (bearish) markets, the aggregate investor sentiment may not show significance in this case. On contrary, for the method of dividing firm performance levels, investors will have high probability profits (losses) under high- (low-) firm performing levels and the individual investor sentiment would show significance. Hence, dividing firm performance levels may have a higher capability of representing individual investor sentiment than dividing market states. This paper expects that the different “firm performance levels” will result in differences in individual investor's sentiment; individual investors will apply different investment strategies, which will ultimately cause different momentum effects. Moreover, Grinblatt and Han (2005) attempt to employ the prospect theory to explain the momentum effect. According to the prospect theory provided by Kahneman and Tversky (1979), investors with capital gains or capital losses will have different ways to dispose of their assets. The investors with capital gains can be regarded as winners; they are risk-averse and tend to sell assets with capital gains. Those with capital losses can be regarded as losers; they are risk-seeking and tend to hold the asset with capital losses. For the sample in this paper, the excess returns on high- and low-performing individual stocks are 22.04% and − 16.85%, respectively. Thus, the investors with high-performing individual stocks are likely to be winners, those with low-performing individual stocks are similar to losers, and they tend to have different ways of disposing of their assets. Due to different disposing methods, this paper expects the momentum effect to be different under high- and low-performing individual stocks. Thus, this paper tests the first hypothesis: Hypothesis 1. For real estate stocks, the momentum effects are positive under high-performing individual stocks but are negative under low-performing individual stocks. If Hypothesis 1 is accepted, this implies that there is an asymmetric phenomenon of positive and negative momentum effects under high- and low-performing individual stocks. This is similar to Lee and Kuo's (2010) asymmetric phenomenon of positive and negative momentum effects under bullish and bearish markets. As mentioned before, market performance levels and firm performance levels both affect investor sentiment, but this paper is more interested in another issue that which one has a stronger effect on investor sentiment. Hence, the second purpose of this paper is to further identify whether the positive (negative) momentum effect under high (low) firm performance level is stronger than that under a bullish (bearish) market. Thus, the second and third hypotheses are as follows: Hypothesis 2. For real estate stocks, the positive momentum effect under high firm performance levels is stronger than that under bullish markets. Hypothesis 3. For real estate stocks, the negative momentum effect under low firm performance levels is stronger than that under bullish markets. If Hypothesis 2 and Hypothesis 3 are accepted, this implies that the effect of a firm's operative performance is stronger than that of market states. In other words, superior (inferior) fundamental business performance and bullish (bearish) markets can cause the stock prices to go up (down), but the effect of the former is stronger than that of the latter. Furthermore, to investigate the momentum effect under high- and low-performing individual stocks, this paper applies the quantile regression method (QR) provided by Koenker and Bassett (1978).10 QR possesses some comparative advantages with the estimating sample.11 Additionally, for dealing with the individual effects of different individual stocks, this paper further modifies the quantile regression method as a quantile regression with a dummy variable estimator (QR-DVE). However, the disadvantage of QR-DVE is the need to estimate many fixed effects, and it has the problem of the inflated dummy coefficient estimates. Hence, this paper also implements Koenker's (2004) methods to avoid to estimating numerous fixed effects and to correct for the inflated dummy coefficient estimates. Notably, QR-DVE uses individual excess returns to divide firm performance levels, but firm excess returns are lagging indicators. Although the empirical results of the momentum effect generated by QR-DVE have the advantage of robustness, in theory, the application of the results is difficult in practice. Fortunately, past literature often employs the turnover ratio as the proxy for investor sentiment and finds that past turnover ratios12 (or trading volume) can predict both the magnitude and persistence of price momentum (Subrahmanyam, 2005). Hence, past turnover ratios may be the fourth important factor for the momentum effect. Subrahmanyam (2005) finds that past turnover ratios are positively related to future returns for high-performing individual stocks but are negatively related to returns for low-performing individual stocks. Thus, this paper tries to test the fourth hypothesis as follows: Hypothesis 4. For real estate stocks, past turnover ratios are positively related to future returns for high-performing individual stocks but are negatively related to returns for low-performing individual stocks. According to the above analysis, this paper summarizes that the industry factor, horizons, market states, and turnover ratios (e.g., investment sentiment) are important factors for determining the momentum effect.13 This paper specifically investigates the momentum effect for Chinese real estate stocks by controlling the industry factor. Next, this paper tries to identify the asymmetric momentum effects under different firm performance levels and compares the horizon of asymmetric momentum effects under different market and firm performance levels. Finally, this paper also identifies the relation between past turnover ratios and future returns under different firm performance levels, and regards the past turnover ratios as indicators to design applicable investment portfolios for investors. Prior literature often regards the momentum horizon as the main reason for different momentum effects (Conrad and Kaul, 1998), while this paper finds that the key factor may be differences in firm performance levels (e.g., differences in investor sentiment). In addition, this paper further finds two factors (e.g., firm performance levels and their corresponding turnover ratios) that substantially influence investment performance. The above findings also highlight this paper's contribution to the literature. The layout of the remainder of this paper is as follows. Section 2 explains the models and methodology, Section 3 presents a description and analysis of the data, Section 4 displays the empirical results and analysis, and Section 5 provides the conclusion and suggestions.

This paper investigates the momentum effects under different firm performance levels for Chinese real estate stocks. The main findings are as follows. First, the momentum effects are not consistent under different firm performance levels for Chinese real estate stocks. Regardless of the momentum horizon, the momentum effects are positive under high-performing individual stocks, but they are negative under low-performing individual stocks. Lee and Kuo (2010) find that the above-mentioned different momentum effects only exist for the short-horizon between bullish and bearish markets, but the same momentum effects apply to the long-horizon under both bullish and bearish markets. However, this paper further finds that the horizon of positive momentum effects under higher-performing stocks is longer than that under a bullish market. Second, this paper finds that the positive (negative) momentum effect under high (low) firm performance levels is stronger than that under a bullish (bearish) market. This implies that superior (inferior) fundamental business performance and a bullish (bearish) market can cause the stock prices to go up (down), but the effect of the former is stronger than that of the latter. For the application of our findings, this paper uses past turnover ratios as a leading indicator to design two profitable investment portfolios. As the relation between past turnover ratios and future returns becomes positively correlated, which means the firm performance levels locate on high-performing individual stocks, this paper suggests a portfolio of buying past high turnover ratio winners. As the relation between past turnover ratios and future returns becomes negatively correlated, this paper suggests a portfolio of buying past low turnover ratio losers. After examining the profits of the portfolios, this paper proves that the performance of the two portfolios is superior to the average returns of real estate stocks. Prior literature often regards the momentum horizon as the main reason for different momentum effects (Conrad and Kaul, 1998), while this paper shows that the key factor may be differences in firm performance levels, i.e., differences in investor sentiment. If investors apply a momentum trading strategy to China's real estate stocks, they should consider carefully the firm performance levels and their corresponding turnover ratios since both factors will influence their investment performances. In addition, there are some research limitations that apply to this paper. First, this paper only tests 65 Chinese real estate stocks, and the small sample size is an important limitation. Second, this paper specifically studies the momentum effect for real estate firm stocks but does not further discuss the empirical results of the control variables. Finally, this paper provides some avenues for future research. First, this paper only tests the momentum effect of Chinese real estate stocks by dividing firm performance levels, while future research could test the real estate industry or REITs in other stock markets. Second, this paper employs the market index to divide different market states, whereas future research could use the real estate industry index or housing cycle to divide different market states. Third, the location of a real estate firm may also be an important factor for testing the momentum effect on real estate stocks.