On the predictability of Chinese stock returns
Kansas State University
College of Business Administration
Manhattan, KS 66506
Tel: (785) 532-6844
Fax: (785) 532-6822
Kenneth A. Kim
State University of New York at Buffalo
Jacobs Management Center
Buffalo, NY 14260
Tel: (716) 645-3266
Fax: (716) 645-3823
University of Iowa
Tippie College of Business
Iowa City, IA 52242-1994
Tel: (319) 335-3924
Fax: (319) 335-3690
University of Rhode Island
College of Business Administration
Kingston, RI 02881
Tel: (401) 874-7415
Fax : (401) 874-4312
On the predictability of Chinese stock returns
We examine stock return predictability in China. We take eighteen firm-specific variables that
have been documented to predict cross-sectional stock returns in the U.S. and examine their
relation with stock returns in China for the sample period from 1995 to 2007. We find relatively
weak predictability for Chinese stocks. Six (four) firm-specific variables have predictive power
for raw (risk-adjusted) returns in China. Tests on U.S. stock returns find more predictors can
explain cross-sectional stock return variation. We test two explanations for the cause of weak
returns predictability in China. First, perhaps return predictors in China are less heterogeneously
distributed than they are in the U.S. Second, stock prices are less informative in China than they
are in the U.S. We find support for both explanations.
On the predictability of Chinese stock returns
The stock market in China has grown rapidly in recent years. It plays an increasingly
active role in China’s economic growth, and is increasingly accessible to international
investors. Naturally, understanding the economic forces and individual firm characteristics
driving stock price movements in this market becomes an important issue. However, to date,
academic research in this area is still nascent. The purpose of this study is to provide a
systematic analysis on the relationship between firm characteristics and cross-sectional stock
returns in this emerging yet already quite large stock market.
The fact that stock returns can be predicted by various firm characteristics such as
size, book-to- market ratio, and past stock returns – which is generally termed “cross-sectional
return predictability” – has been well-documented for the U.S. market and many other equity
markets around the world. The cause of such predictability has also been long debated.
Unexplained by standard asset pricing models such as CAPM, cross-sectional return
predictability is viewed by many as prima facie evidence of market inefficiency. Meanwhile,
other researchers attribute predictability to unmeasured risk factors or rational dynamics in
conditional risk-return tradeoff. In this study, we take a large set of return-predictive firm
characteristic variables that are documented in the U.S. market, and examine whether these
variables can predict stock returns in the Chinese market. While our approach to document
return predictability is systematic, our goal is modest. We do not intend to provide a definite
answer on whether such predictability is a rational asset pricing effect or due to market
inefficiency (answering this question is an ambitious task that we leave for future studies).
However, some of our empirical results are informative in evaluating a subset of explanations.
Moreover, we believe our findings can serve as valuable basis for the further discussions on
the causes and explanations of stock return predictability in the emerging markets.
To the extent that return predictability is caused by investor misreaction to information
or irrational preferences, there are reasons to anticipate strong predictability in the Chinese
market. Several studies have documented strong behavioral biases among investors in this
market, such as over-confidence, disposition effect, representativeness bias, and herding; see,
e.g., Chen, Kim, Nofsinger, and Rui (2007), Feng and Seasholes (2005), and Shumway and
Wu (2006). In addition, stock trading is dominated by individual investors (e.g., Chen, Kim,
Nofsinger, and Rui, 2007), who are potentially subject to stronger biases relative to
sophisticated institutional investors. Further, short-sales are not allowed in China, making it
difficult for mispricing to be quickly arbitraged away. 1
There are several studies on the predictability of Chinese stock returns; however, these
studies typically focus on a small set of predictive variables. Wang and Xu (2004) find that
firm size explains cross-sectional differences in returns but that the book-to-market ratio does
not. Kang, Liu, and Ni (2002) document price momentum in the Chinese market while Wang
(2004) finds the contrary. In addition, Wang and Chin (2004) document the return predictive
power of trading volume. Eun and Huang (2007) further find that that although stock beta is
not priced in cross-sectional stock returns in the Chinese market, a few firm characteristics,
including size, book-to- market ratio, firm-specific risk, dividend-yield, and the existence of
offshore shares (B and H shares), can be used to predict returns. They argue that many of
these effects are related to market imperfections, and that “[g]iven imperfections, stocks are
priced rather rationally in China, despite the widespread perception to the contrary.”
We specifically identify 18 firm-specific variables, known to be predictive of cross-
sectional stock returns at the annual horizon in the U.S. market, and test their predictability on
Chinese stock returns. The variables we include in our study are as follows: (i) market
capitalization, book-to- market ratio, and momentum (i.e., past 12- month stock returns), which
are three return predictors most extensively scrutinized in the U.S. literature; (ii) three
conventional value indicators: earnings-to-price ratio, cash flow-to-price ratio, and sales
growth; (iii) two measures for earnings quality: accounting accruals and net operating assets;
(iv) measures of firms’ tangible and intangible investments, including capital expenditure,
research and development expenditure, and advertising expenditure; (v) asset growth and
change in gross profit margin, which summarize the firms’ growth in size and profit margins,
(vi) equity financing and debt financing for firms’ external financing activities, and finally,
(vii) idiosyncratic return volatility, trading turnover, and Amihud’s (2002) illiquidity ratio as
variables related to information uncertainty or liquidity of stocks.
We find that during the period from 1995 to 2007, almost all of the proposed
predictors predict the direction of subsequent one- year stock returns in a way consistent with
To the extent that stock return predictability is driven by rational asset pricing effect, the existence of such
predictability is interesting on its own; it suggests that there are similar risk factors or similar dynamics in risk-
return tradeoff in p lay in both the U.S. and Chinese markets.
their predictive patterns in the U.S. market. To be specific, after adjusting for the sign to make
each variable positively related to stock returns in the U.S. market, the return difference
between top and bottom decile portfolios sorted by each individual return predictor is mostly
positive in the Chinese market. Six of the 18 proposed predictors are statistically significant in
explaining the cross-sectional variation of subsequent returns. The variables with significant
return predictability include the book-to- market ratio, accruals, net operating assets, R&D
spending, asset growth, and illiquidity. When performing the same tests on U.S. stocks during
the same period, we find significant return spreads between top and bottom decile portfolios
for 10 out of 18 variables, suggesting a weaker anomaly effect in the Chinese market than in
the U.S. market. For robustness we examine the alphas of stock portfolios sorted on firm
characteristics, based on the Fama-French three factor model. Five out of the 18 variables
produce significant alphas for the top-bottom decile return spread in the Chinese market while
10 predictors produce significant alphas in the U.S. market. The overall message is that, stock
return predictability exists in the Chinese market, but the effect is weaker relative to the U.S.
What explains the weaker predictability of Chinese stock returns? To start with, we
first point out an intriguing contradiction of our empirical evidence with the conventional
perception. It is often considered that, because return predictability is an indication of market
inefficiency (conditional on behavioral explanations), in cross-country comparisons, the
market with higher return predictability should be the one with higher degree of market
inefficiency. This, however, is unlikely the case because the Chinese market is usually
perceived to be much less efficient.
One possible explanation we consider is that Chinese firms are more homogeneous in
the characteristics examined. If there is not much cross-sectional dispersion in a return-
predictive variable, the cross-sectional dispersion in stock return associated with the variable
will also be small. To examine this hypothesis, we compare the cross-sectional differences in
the return predictors themselves between the two markets. We find that indeed return
predictors are more homogeneous in China. However, we also find that the sensitivity of
stock returns to per unit cross-sectional difference in many of the variables is weaker in
China. We further confirm such results by using Fama-MacBeth (1973) cross-sectional
regressions. Therefore, lack of cross-sectional dispersion of firm characteristics is not the
whole reason for weaker return predictability in China.
We next consider a hypothesis that is opposite to conventional wisdom. That is, the
lack of return predictability in China is actually due to long- lasting market inefficiency. Under
the behavioral asset pricing framework, return predictability depends on two factors –
investors’ initial misreaction and subsequent price correction. If initial mispricing is large, but
stock price is persistently noisy, to the extent that mispricing is seldom corrected, then return
predictability will be weak. This is likely for an emerging market such as China, where stock
prices are very noisy and at the same time fundamental information is often not reliable. For
example, earnings management and accounting manipulation have been found to be prevalent
(e.g., Chen and Yuan, 2004; Haw, Qi, Wu, and Wu, 2005; Jian and Wong, 2004); there is also
evidence of rampant market manipulation (e.g., He, 1998; Shenzhen Stock Exchange, 2005).
To test this hypothesis, we resort to a well-documented measure of valuation
efficiency – stock return synchronicity, or the R-square of regressing individual stock returns
onto market returns. 2 A number of studies have concluded that return synchronicity is
inversely related to stock price informativeness; that is, lower return R2 is associated with
greater capitalization of firm-specific information (Roll 1988, Morck, Yeung, and Yu 2000,
and Durnev, Morck, Yeung, and Zarowin 2003). It is also well-known that in less developed
capital markets, R2 tends to be particularly high, suggesting low pricing efficiency.
Our empirical evidence is consistent with this hypothesis. During our sample period,
the average R2 in China is 0.46 while it is 0.12 in the U.S. market, indicating that stock prices
in the Chinese market incorporate firm-specific information to a substantially less degree.
Further, when we divide Chinese stocks into subsamples based on synchronicity, we find that
return predictability is stronger among stocks with low R2 , relative to stocks with high R2 .
Therefore, weaker predictability in China does not mean that Chinese stocks are more
rationally priced; rather, it is a symptom of persistent noisy valuation and persistent
mispricing. An interesting implication that one can generalize, at least in the context of
comparing international financial markets, is that there exists an intricate relationship between
Admittedly, our test here is indirect. An ideal direct test of the hypothesis would involve identify ing an
“intrinsic value” for a stock based on fundamental informat ion, and then examining the joint process of the
observed stock price and the intrinsic value. Ho wever, this approach is difficult to imp lement due to lack of
reliable fundamental information in the Chinese market.
return predictability and market efficiency. It would be naïve to label a market more efficient
simply because there is less return predictability.
The remainder of the paper is organized as the following. Section II provides a review
of the existing literature on stock return predictability. Section III discusses data and
methodology for constructing stock return predictors for both the U.S. and Chinese markets.
Section IV and V provide empirical results. Section VI concludes.
2. Brief discussion of stock return predictors
The literature on the relation between firm-specific variables and the cross-section of
expected stock returns is vast. Within this literature, we attempt to identify the most important
firm-specific return predictor variables. We end up with 18 such variables, and group them
into the following categories: (i) three return predictors most extensively scrutinized in the
U.S. literature: market capitalization, the book-to- market ratio, and momentum (i.e., past 12-
month stock returns); (ii) three conventional value indicators: earnings-to-price ratio, cash
flow-to-price ratio, and sales growth; (iii) two measures for earnings qua lity: accounting
accruals and net operating assets; (iv) measures of firms’ tangible and intangible investments,
including capital expenditure, research and development expenditure, and advertising
expenditure; (v) asset growth and change in gross profit margin, which summarize the firms’
growth in size and profit margins, (vi) equity financing and debt financing for firms’ external
financing activities, and finally, (vii) variables related to information asymmetry or liquidity
of stocks including idiosyncratic return volatility, trading turnover, and Amihud’s (2002)
illiquidity ratio. In this section, we offer a brief literature review of these variables.
The size effect and the value effect are among the earliest proposed stock return
predictors (e.g., Banz, 1981; Basu, 1977; Reinganum, 1981). Smaller market capitalization
firms and firms with higher earnings-to-price ratios tend to earn higher returns. Similarly,
Fama and French (1992) show that firms with higher book-to-market ratios subsequently have
higher returns. Lakonishok, Shleifer, and Vishny (1994) report that future returns are
positively correlated with cash flow-to-price ratio and negatively correlated with past sales
growth. They suggest that investors overvalue firms’ past performances, so these variables are
often jointly regarded as value or contrarian indicators.
Jegadeesh and Titman (1993) show that stocks with better past returns subsequently
earn higher returns, at horizons ranging from 3 to 12 months. The buying of past winners and
the selling of past losers are widely known as momentum strategies. As shown by Fama and
French (1996), the momentum effect is not subsumed by the size or value effects. Several
behavioral models have been proposed to explain the momentum effect based on investor
underreaction or overreaction to information (Barberis, Shleifer, and Vishny, 1998; Daniel,
Hirshleifer, and Subrahmanyam, 1998; Hong and Stein, 1999) or irrational investor
preferences (Grinblatt and Han, 2005).
Sloan (1996) reports an accruals anomaly. Stocks with higher accounting accruals
tend to have lower future returns. His explanation for the anomaly is investors overvalue
firms with high accruals thinking that accruals will persist. Hirshleifer, Hou, Teoh, and
Zhang (2004) find that firms with higher net operating assets (NOA) have lower future returns,
suggesting that marginal investors do not understand that high NOA implies decreasing
returns to assets scale. These two anomalies are somewhat related as net operating assets is
the accumulation over time of the difference between net operating income and free cash
flows, plus capitalized investments. However, despite their potential relation, Hirshleifer, et
al. (2004) show the explanatory power of accruals and NOA do not subsume each other when
used jointly to predict returns. These two effects are often referred to as earnings quality
Titman, Wei, and Xie (2004), and Beneish, Lee, and Tarpley (2001) find that firms
with high capital investments subsequently have low returns. Titman, et al. (2004) attribute
this finding to an overinvestment tendency of corporate managers and investor underreaction
to information. In contrast to evidence on tangible capital investments, Chan, Lakonishok and
Sougiannis (2001) find that ratios of corporate R&D spending and advertising spending to
market capitalization are positively correlated with future returns. They point out that
intangible investments on R&D and advertising are expensed rather than capitalized in
accounting treatment (to the effect of depressing current earnings at the benefit of future
earnings) and that investors appear to misreact to this accounting effect. The effects of capital
expenditure, R&D, and advertising can be referred to as corporate investment anomalies.
A large number of corporate event studies have shown that future stock returns are
abnormally low in the years following initial public offerings (Ritter, 1991), seasoned equity
offerings (Loughran and Ritter, 1997), debt offerings (Spiess and Affleck-Graves, 1999), and
bank borrowings (Billett, Flannery, and Garfinkel, 2006). Conversely, future stock returns are
abnormally high following stock repurchases (Ikenberry, Lakonishok, and Vermaelen, 1995;
Lakonishok and Vermaelen, 1990). Bradshaw, Sloan and Richardson (2006) summarize
firms’ external financing activities into two variables: external equity financing and debt
financing. They report that these two variables are negatively correlated with future stock
returns, and they attribute this pattern to investor optimism and firms’ efforts to time the
market in raising capital. We categorize all of these variables and findings as financing-
Related to the corporate investment and financing anomalies is the asset growth
anomaly reported by Cooper, Gulen and Schill (2007). They find a strong inverse relation
between firm asset growth and future stock returns. They associate their finding to investor
misreaction to assets growth. Because firms in their early growth stages experience a pos itive
relation between assets growth and subsequent returns, investors might assume this relation
will persist into the future thus underappreciating that return on assets diminishes as asset size
Abarbanell and Bushee (1998) view changes in gross profit margins as part of a
fundamental analysis strategy. Greater changes in gross margins indicate an improvement in
the firm’s terms of trade, which, in turn, leads to higher expected operating performance and
future stock returns.
Finally, information uncertainty and stock liquidity have also been found to be related
to future returns. Ang, Hodrick, Xing, and Zhang (2006) show that stocks with high
idiosyncratic volatility risk have low subsequent returns. Datar, Naik, and Radcliffe (1998)
and Lee and Swaminathan (2000) show that stocks with high trading volume earn lower
future returns. Jiang, Yao, and Xu (2006) link both anomalies to adverse selection in
corporate disclosure and investor underreaction. In addition, Amihud (2002) finds that stock
liquidity inversely predicts stock returns.
From the above discussion, we identify 18 firm-specific return predictor variables.
We list these variables in Table 1. For those variables that have been posited to be negatively
related to subsequent returns in the U.S. market, we transform them by changing their signs
and we indicate this transformation by putting a “–” sign in front of variable. All of our
variables are posited to be positively related to future returns. We do not attempt to develop
unique ex ante hypotheses for each predictor on Chinese stock returns as our study is
exploratory in nature. Instead, we accept the hypotheses that have been developed for U.S.
markets as the default applicable hypotheses for Chinese markets.
[Insert Table 1 Here]
3. Data and variables construction
For data on Chinese firms, information on stock price, return, and trading volume, as
well as corporate financial statement information, is obtained from the PACAP-CCER China
database, a joint product of the PACAP Research Center at the University of Rhode Island
and SINOFIN Information Service Ltd. There are different share types in China. In this study,
we only include A-shares, which account for more than 85% of the tradable market value of
Chinese stocks at year-end of 2007. 3 Our sample covers financial statement data from 1994
to 2005 and stock return data from July 1995 to June 2007. 4 We eliminate banks, close-end
funds, real estate firms, and investment companies. Approximately 5% of Chinese stocks
have a price below ¥1. 5 To avoid potential market microstructure related issues in measuring
returns, we require that a stock have end-of-June price of no less than ¥1 to be included in our
sample for year t.6 Further, a stock must have available information on stock price, market
capitalization, and at least one valid stock return predictor in addition to firm size at the end of
June in order to be retained in our sample.
Stocks of a typical Chinese firm may consist of state shares (those owned by the central or local governments),
legal-entity shares (those held by domestic legal entities such as listed companies, state owned enterprises, and
banks), and tradable shares, with the restriction that state and legal-entity shares cannot be traded publicly.
Tradable shares are further classified into tradable A- and B-share classes. Tradable A-shares are ordinary shares
primarily made available to Chinese citizens and institutions, whereas B-shares are primarily made availab le to
foreign investors. A recent regulatory change makes A -shares available to a small g roup of qualified foreign
institutional investors. Since February 2001, B-shares are available to do mestic investors.
In our tests, return pred ictors are constructed using accounting information at the end of fiscal year t -1. The
holding period fo r stock portfolios in the subsequent year is fro m July of year t to June of year t +1. More
specifically, the return hold ing period corresponding to accounting informat ion reported at the end of fiscal year
2005 (December 2005 as December is the fiscal year end for all Ch inese firms) is fro m Ju ly of 2006 to June
The official abbreviation for the Chinese currency, the yuan, is CNY, but it is commonly abbreviated as RMB
(renmindi, which literally translates to “people’s currency”). The Latin ized sy mbol fo r the yuan is ¥.
Our results remain similar without this minimu m price threshold and when we set the minimu m price at ¥5.
Panel A of Table 2 presents summary statistics for the Chinese market. In June of
1994, a total of 287 stocks are traded in the Shanghai Stock Exchange and the Shenzhen Stock
Exchange, and the number substantially increases to 1,516 in June of 2007. In June of 2007,
the aggregate market capitalization of tradable A-shares is ¥5294.6 billion (equivalent to
USD756.37 billions using an exchange rate of ¥7 for USD1), about 40% of the aggregate
nominal market capitalization of ¥13,465 billion (USD 1,923.57 billion).
[Insert Table 2 Here]
For data on U.S. firms, information on stock price, return, and trading volume is from
CRSP. Information on corporate financial statements is from COMPUTSTAT. As with the
Chinese sample period, the U.S. sample period also covers 1994-2005 for financial statement
data and July 1995 to June 2007 for stock returns data. Our filters on U.S. data are as follows.
First, we select all common stocks traded in NYSE, AMEX, and NASDAQ at the end of June
in each year t. Second, we eliminate primes, close-end funds, real estate investment trusts,
American Depository Receipts, and foreign companies. Third, we require stocks to have a
minimum price of $1 at the end of June.
Panel B of Table 2 shows summary statistics for our U.S. sample. From Panels A and
B, we note some interesting differences between China and the U.S. First, we see that
Chinese stocks experience far more trading. During our sample period, the average annual
stock turnover ratios, defined as the ratio of trading volume to firm’s market capitalization,
are 503% in Shanghai Stock Exchange and 509% in Shenzhen Stock Exchange, whereas the
corresponding average annual turnovers are 133% for NYSE/AMEX and 177% for NASDAQ.
In addition, Chinese stocks are much smaller in size with respect to price: the time-series
average stock price in the Chinese market is ¥10.77 for A shares, which is roughly $1.5 when
using the exchange rate of ¥7 per US dollar, much smaller relative to the average U.S. stock
price of $27.75.
3.2. Construction of stock return predictor variables
Based on our discussion in section 2, we construct 18 stock return predictor variables.
Here, we briefly describe the construction of each of those variables. More detailed
information on all variables can be found in the Appendix.
1. Firm Size (SIZE)
SIZE is the natural logarithm of the firm’s market value of equity, which is the firm’s
market price multiplied by its common shares outstanding at the end of June of year t. For
our Chinese firms, only the outstanding A-shares are used in the market value computation.
2. B/P Ratio (B/P)
B/P is the book value of equity, taken as the ratio of the firm’s book value to its
market capitalization for the year-end at calendar year t – 1.
3. Price Momentum (MOM)
MOM is the cumulative return of a stock in month –12 through –1 preceding June of
year t. We skip one month between portfolio formation and holding period to avoid the
effects of bid–ask spread, price pressure, and any lagged reaction (Jagadeesh and Titman,
4. E/P Ratio (E/P)
E/P is the ratio of earnings for the fiscal year to the market capitalization at the year-
end of calendar year t – 1. The earnings measure for U.S. firms is earnings before
extraordinary items and depreciation, whereas it is net income for Chinese firms.
5. C/P Ratio (C/P)
C/P for U.S. firms is the sum of earnings before extraordinary items and depreciation
over the firm’s market capitalization at the fiscal year-end of year t – 1. This definition
follows Fama and French (1993). For Chinese firms, similarly, C/P is the sum of net income
and depreciation over its market capitalization at the fiscal year-end of year t – 1.
6. Sales Growth (SG)
SG is the sales revenue for the fiscal year ending in calendar year t – 1 over the sales
revenue from the fiscal year-end in year t – 2.
7. Accruals (ACC)
Accounting accruals is the noncash component of earnings. Following Sloan (1996),
we estimate ACC as the change in noncash current assets less the change in current liabilities
(excluding debt in current liabilities and income tax payable) and less depreciation, during the
fiscal year ending in year t – 1, scaled by the average total assets at the beginning and end of
that fiscal year.
8. Net Operating Assets (NOA)
Following Hirshleifer et al. (2004), we estimate NOA as the difference between
operating assets and operating liabilities for the fiscal year ending in calendar year t – 1,
scaled by the average total assets at the beginning and end of that fiscal year.
9. Capital Expenditure (CAPEX)
Following Jegadeesh et al. (2004), CAPEX for U.S. firms is the capital expenditure of
the firm for the fiscal year ending in calendar year t – 1 over the average total assets at the
beginning and end of that fiscal year. For Chinese firms, the numerator of the CAPEX
variable is the change in net fixed assets plus the change in accumulated depreciation from the
fiscal year ending in calendar year t – 2 to the fiscal year ending in calendar year t – 1.
10. R&D Expenses (RD)
Following Chan et al. (2001), RD is the ratio of R&D expenditure over market
capitalization. R&D is not separately reported in China but is included in management
expenses. RD for Chinese firms is, thus, approximated by the ratio of management expenses
for the fiscal year ending in calendar year t – 1 to market capitalization at the end of year t –
11. Advertising Costs (ADV)
Following Chan et al. (2001), ADV is advertising expenses over market capitalization.
Advertising costs are not separately reported in China but are included in sales and marketing
expenses. ADV for Chinese firms is, thus, approximated by sales and marketing expenses for
the fiscal year ending in calendar year t – 1 over market capitalization at the end of year t – 1.
12. Assets Growth (AG)
Following Cooper et al. (2007), AG is measured as the percentage change in total
assets from the fiscal year ending in calendar year t – 2 to the fiscal year ending in calendar
year t – 1.
13. Change in Gross Profit Margin (∆GPM)
Consistent with Abarbanell and Bushee (1998), GPM is the difference between net
sales and cost of goods sold divided by net sales. ∆GPM is the percentage change in GPM
from the fiscal year ending in calendar year t – 2 to the fiscal year ending in calendar year t –
14. External Equity Financing (∆EQ)
Following Bradshaw et al. (2006), ∆EQ is the net cash received from the sale (and/or
purchase) of common and preferred stock less cash dividends paid for the fiscal year ending
in calendar year t – 1.
15. External Debt Financing (∆DT)
Following Bradshaw et al. (2006), ∆DT is the net cash received from the issuance (or
reduction) of debt for the fiscal year ending in calendar year t – 1.
16. Idiosyncratic Risk (STDR)
Similar to Ang et al. (2006), STDR is the standard deviation of the residuals in the
regression of daily stock return on daily value-weighted market return with five lags and five
leads for the period from month –12 through month –1 preceding June of year t.
17. Trading Turnover (TURN)
Following Jegadeesh, Kim, Krische, and Lee (2004), TURN is the percentile rank of
the average daily volume turnover in the 12 months preceding June of year t, where daily
volume turnover is the ratio of the number of shares traded each day to the number of s hares
outstanding at the end of the day. Because trading volume is measured differently in
NASDAQ than in NYSE and AMEX, a percentile ranking is performed separately for
NASDAQ and for NYSE/AMEX.
18. Illiquidity (ILLIQ)
Following Amihud (2002), ILLIQ is the percentile rank of the average daily ratio of
the absolute stock return to its dollar volume, across month –12 through month –1 preceding
June of year t. Again, as with the trading turnover measure, a percentile ranking is performed
separately for NASDAQ and for NYSE/AMEX.
Table 3 summarizes the cross-sectional distributions of the 18 stock return predictors,
for both our Chinese data (Panel A) and our U.S. data (Panel B), including 25 and 75
percentiles, mean, median, standard deviation, skewness, and kurtosis.
[Insert Table 3 Here]
Recall that among these 18 variables, 10 are posited to be negatively correlated with
future stock returns: SIZE, SG, ACC, NOA, CPX, AG, ∆EQ, ∆DT, STDR, and TURN. In the
rest of analysis, we add a negative sign in front of each of these variables (see Table 1) so that
the resulting variables positively predict returns.
4. Empirical results
4.1. Raw returns of sorted portfolios
We first examine the return-predictive performance of our 18 variables using sorted
decile portfolios. In June of each year t, we rank stocks into deciles based on each predictor
using year t – 1 financial statement data. We form equally weighted portfolios in each decile
and hold the positions from July of year t to June of t + 1. 7 That is, we test whether our
predictors can explain the cross-sectional variation in subsequent one- year returns. We first
calculate the cross-sectional average return in each portfolio and then compute the time-series
averages of the cross-sectional mean for each predictor. D10 and D1 deciles contain stocks
with the largest and smallest measures of each return predictor, respectively. For example, for
firm size (SIZE), D10 stocks contain the decile of the largest firms, and D1 contain the decile
of the smallest firms. For momentum (MOM), D10 contain firms with the largest past returns,
and so forth. Table 4, Panel A, reports the average portfolio returns for each decile for the
Chinese market and also the mean return difference (i.e., spread) between D10 and D1 stocks,
in Panel B we report the return spread between D10 and D1 stocks for the U.S. market, and in
Panel C we report the spread difference between China and U.S. markets.
[Insert Table 4 Here]
In looking at the second-to- last row in Panel A of Table 4, where mean differences
between D10 and D1 stocks are reported, we see that except for MOM, E/P and C/P
predictors, the signs of the mean differences for all other predictors are positive, consistent
with their hypothesized signs (recall that when hypothesized signs are negative we transform
those variables so that their hypothesized signs become positive). This observation is
somewhat surprising, as it suggests that variables that have been found to predict returns in
mature markets, like U.S. markets, are also potentially useful in predicting returns in
emerging markets such as China’s markets, confirming the argument in Rouwenhorst (1998,
1999) that similar return factors are present around the world.
The bottom row in Panel A of Table 4 presents t-statistic for the return difference
between D10 and D1 stocks. It shows that only five return predictors significantly predict
stock performance in the subsequent year. More specifically, reported in column (2), the top
B/P-sorted portfolio outperforms the bottom B/P-sorted portfolio by 0.74% per month (t =
We obtain consistent, slightly weaker, results when using value-weighted returns.
1.80), consistent with a finding in Eun and Huang (2007). In columns (7) and (8), we see that
high –ACC and NOA firms outperform low –ACC and -NOA firms by 0.35% and 0.56% per
month, significant at the 10% and 5% levels, respectively. In column (10), we see firms with
high R&D outperform those with low R&D by 0.87% per month (t=2.52). The statistically
significant results for ACC, –NOA and R&D is consistent with the interpretation that Chinese
investors are largely uninformed and/or un-sophisticated; they underreact to salient
information provided in firms’ financial statements regarding firms’ accruals, net operating
assets (accumulated accruals) and R&D expenses (related to firm intangible assets) (Sloan,
1996; Hirshleifer et al. 2004; Chan et al., 2004). Further, in columns (12), we see that firms
with high asset growth have higher subsequent returns. For Illiquidity rank-sorted portfolios,
D10 stocks outperform D1 stocks by 0.66% per month (t=1.65), indicating that Chinese
investors are willing to pay a significant premium for more liquid stocks.
While it is hard to provide a consistent story to justify why some predictors have
significant return predictive power in the Chinese market while others do not, interesting
points arise. First, trading on momentum, perhaps the most important stock characteristic
predicting future stock performance in the U.S. and in other developed markets, is
unprofitable in the Chinese market. The return difference between D10 and D1 stocks sorted
by prior 12- month momentum is 0.05% with a t-statistic of 0.16. This finding is consistent
with Wang (2004). Second, firm size, return standard deviations, and turnover do not
significantly predict stock performance in the subsequent year. 8
Panel B of Table 4 reports U.S. results. For brevity, we simply provide mean return
differences between D10 stocks and D1 stocks sorted by respective firm characteristics. Over
the same sample period 1994-2007, the return differences for all the return predictors are
positive, but some firm characteristics, such as firm size, R&D, return standard deviations,
and stock turnover are not reliably return predictive. That is, return differences for 11 of the
18 predictors are statistically significant. Overall, however, these results show that, relative to
the Chinese stock market, more predictors are statistically significant in explaining U.S.
We alternatively measure firm size using firm total market cap italizat ion, including market capitalization for A
and B shares and total book value of assets. We obtain similar results.
As the average annual turnover ratio is much higher in Ch ina than in U.S. , it is likely the included predictors
work better in shorter horizons. In untabulated analysis we examine this conjecture by constructing predictors
In Table 4, we also examine the return predictive power of the 18 firm-specific
variables jointly. We create a combined return predictor measure using the average ranking of
the 18 return predictors weighted by their sensitivities to subsequent returns. Specifically, the
combined predictor measure for stock j in year t is expressed as
COMBO t j Rank i ,jt / 18 (1)
where Ranki ,jt is the cross-sectional percentile rank of return predictor i for stock j. Stocks are
ranked into deciles in each year based on COMBO t j . That is, firms with the lowest and
highest COMBO measures are denoted as D1 and D10 stocks, respectively. Reported in the
last column of Panel A, Table 4, we see that the average return for COMBO-sorted D1 stocks
is 1.52% per month and the average return for COMBO-sorted D10 stocks is 2.24% per
month. The return difference between D10 and D1 is 0.72%, with a t-statistic of 1.98. This
result suggests that a combined rank measure of all predictor variables is effective in
predicting future stock returns in the Chinese market. For the U.S. market, the return
difference between D10 and D1 stocks for the COMBO-sorted decile portfolios is 0.89%
In Panel C, we report differences between D10-D1 return spreads between China and the
U.S. Most differences are negative, indicating that the return spread is larger in the U.S.
market than in the Chinese market. The differences in return spreads between China and U.S.
are statistically significant negative for 8 out of 18 predictors, including MOM, E/P, C/P, -SG,
ADV, -AG, -∆EQ, and -∆DT.
4.2. Alphas of sorted portfolios
An important question that needs to be raised whenever any pattern of stock return
predictability is found is whether the pattern is related to risk. To control for the risk–return
relation, we use the Fama–French (1993) three-factor model to obtain alphas for all decile
portfolios. Our analysis is motivated by Fama and French (1998), who show that the value
effects in international markets can be explained by their three- factor model. The three- factor
regression model takes the following form:
based on semi-annual data (fro m 1994 to 2001) and quarterly data (from 2002 to 2006), and analyzing stock
returns in subsequent 3 and 6 months . The results are similar to those reported in Table 4.
Rit RFt i bi RMRFt si SMBt hi HML t i ,t , (2)
where Ri ,t – RFt is the monthly returns of portfolio i in excess of the monthly risk free rate.
Portfolio return Ri ,t in each month is the equally weighted stock returns. In our setting,
stocks are ranked into deciles in each month by a specific return predictor and i refers to the
ith decile portfolio. RMRFt is the market return in excess of the risk free rate; SMBt and HMLt
are the monthly size and book-to-market factors, respectively. Regressions are performed
using average monthly returns of each decile-sorted predictor variable during the entire
We follow procedures provided in Fama and French (1993) to compute SMB and
HML factors using Chinese stock return data. As discussed previously, although Chinese
public companies issue multiple shares, we focus on A-shares. The market value of stocks is
taken by multiplying A-shares’ closing prices to their shares outstanding. Book value per
share is book value divided by totals shares issued by the firm. We construct size and B/P
portfolios in June of each year t. In particular, we sort all stocks into small and big size
groups based on the median market capitalization of all stocks in June of year t. We
independently sort all stocks into low, median, and high B/P groups based on the 30% and
70% cutoff points of the book-to-market ratio of all stocks. Six size-B/P portfolios are
defined as the intersections of the two size and three B/P groups. The monthly value-
weighted average return on each portfolio is then computed. SMB is the difference, in each
month, between the simple average of returns on the three small-stock portfolios (S/L, S/M,
and S/H) and the simple average of returns on the three big-stock portfolios (B/L, B/M, and
B/H). Similarly, HML is the difference, in each month, between the simple average of returns
on the two high-B/P portfolios (S/H and B/H) and the average of returns on the two low-B/P
portfolios (S/L and B/L).
Following Wang (2004) and Kang, et al. (2002), we use the monthly yield of the
three- month household deposit interest rate in China as the risk- free rate. We calculate the
monthly market returns in China as the value-weighted average monthly returns for all A-
shares traded in the Shanghai and Shenzhen stock exchanges. The risk-adjusted results are
reported in Table 5.
[Insert Table 5 Here]
Table 5 results are consistent with that in Table 4. As reported in Panel A of Table 5,
we see that four return predictors are statistically significant after we control for risk. The
alpha differences between D10 and D1 stocks are significant for -NOA, RD, –AG, -STDR in
the Chinese market. Panel B of Table 5, which reports on U.S. stock returns, shows that, after
making risk adjustments, 10 predictors are statistically significant in explaining subsequent
returns. Consistent with our earlier results on raw returns, after risk adjustments, we find that
some firm-specific variables are able to predict returns in China, but more variables are able
to predict returns in the U.S. For the COMBO measure, the return spread is 0.79% per month
(t=2.42) in the Chinese market while it is 1.07% per month (t=2.94) in the U.S. market. Panel
C reports the difference between three- factor adjusted alphas of stocks in China and the U.S.
Different from the procedure used in the first two panels, here we no longer compute alphas
for stock portfolio deciles. Instead we estimate three- factor adjusted alpha for each stock in
each month based on the following two-step procedure. First, we estimate factor loadings by
regressing prior 12- month stock returns onto corresponding factor premiums. Second, we
subtract the expected returns using the estimated loading from stock returns in the current
month to obtain the three-factor adjusted alpha for the stock. With alphas of individual stocks,
we next compute alpha differences between predictor-sorted D10 and D1 portfolios and then
calculate the differences in differences between China and U.S. in each month. Finally, we
average the monthly differences over time to obtain the numbers reported in Panel C. The
differences in alpha spreads between China and U.S. are statistically significant negative for
10 out of 18 predictors.
4.3. Cross-sectional regressions
In this section, we further check the robustness of our decile-sorted portfolio results by
performing cross-sectional regressions of monthly stock returns on return predictors. We use
the Fama–MacBeth (1973) procedure to compute the time-series averages of the coefficients
on each of our predictors. Panel A of Table 6 reports results when we perform univariate
regressions. The dependent variable is the monthly stock return and the explanatory variable
is each individual stock return predictor.
[Insert Table 6 Here]
From Table 6, Panel A, we see that the coefficients on B/P, -NOA, RD, -AG, and
ILLIQ are significantly different from zero, generally validating the sorted-portfolio evidence
reported in Table 4. Also confirming prior results, reported in Panel B of Table 6, we see that
8 firm specific variables have significant coefficients in the return regression in the U.S.
We also conduct two multivariate regression tests, where monthly returns are
regressed on either all 18 predictor variables or on the first 6 principal components. The two
main advantages to estimating a multivariate regression is that we can identify the marginal
effect of each variable on its ability to predict returns and we can estimate the predictive
ability of all of these variables jointly rather than in isolation (i.e., the effect of each variable
can be evaluated while holding constant other variables). For stocks with missing
observations for one or more stock return predictors, we replace the missing predic tor with the
annual cross-sectional median value for that predictor to take advantage of the large cross-
sectional data. The results of multivariable regressions are reported in Panel C of Table 6.
We find few variables have significantly positive coefficients. Multivariable regression
results using U.S. data is reported in Panel D of Table 6.
The average adjusted R2 s from the monthly multivariate regressions, which are also
reported in Panels C and D, provide a way to quantify the overall return predictability in each
market. As it turns out, the average adjusted R2 for the multivariate regression model is 9%
for the Chinese market, while it is 6% in the U.S. market. If independent variables in a
regression model are correlated, then a multicollinearity problem may inflate the R2 s. To rule
out this explanation, we identify the first six principal components of the 18 return predictors,
and we use these six principal component factors as joint regressors. The time-series average
R2 of this principle component-based regression is 5% for both markets. It is interesting to
see that despite there being fewer statistically significant return predictors in China, the
overall explanatory power of firm-specific variables is similar in these two markets as
revealed by their similar R2 s.
4.4. What explains weak return predictability in China?
4.4.1 Homogeneity in Return Predictors
So far, our results suggest that return predictors have much less power in predicting
future stock returns in China than in the U.S. One possibility for this finding is that return
predictors in China are less heterogeneously distributed than they are in the U.S.; that is, (i)
the difference between average values of a return predictor for D10 and D1 stocks sorted by
the predictor is small in China and (ii) return spreads between D10 and D1 stocks are
accordingly smaller in China than in the U.S. market. To examine this possibility, we
compare the spread of D10 and D1 stocks in the predictors themselves in China and U.S. The
results are reported in Panel A of Table 7.
[Insert Table 7 Here]
The first row in Panel A reports the spread in return predictors between D10 and D1
stocks in China and the second row reports the spread for the U.S. Note that numbers
reported in Panel A are not sign adjusted, but return predictors themselves. The differences in
the spreads between China and U.S., and associated t-statistics, are also shown in Panel A.
The differences in the spreads between China and U.S. are negative for 15 out of the 18 return
predictors, suggesting return predictors are generally more homogeneously distributed in
China than in U.S. While SIZE, -NOA, RD, -AG, -∆EQ, and –STDR spreads in China are
significantly lower than that in U.S., 9 out of the 18 differences in spreads are not
significantly different, though negative, between the two countries. A further interesting point
is that the average spread for ACC is larger in China (0.49) than in U.S. (0.36), significant at
the 1 percent level. This finding is consistent with the reality that earnings management is
more rampant in China, leading to lower quality of Chinese firms. However, and perhaps just
as puzzling, as reported in Table 4, the accrual return predictability is only marginally
significant in the Chinese market.
We further assess the impact of the variation of return predictors by estimating
standardized stock return spreads between D10 and D1 portfolios sorted by each of the 18
return predictors for the Chinese and U.S. markets. The standardized return spread of a return
predictor is the time-series average of the stock return spreads between D10 and D1 portfolios
scaled by the predictor spreads of that market in corresponding years. The essential
assumption behind the examination of heterogeneity of return predictors across markets is that
return spread per unit of predictor spread holds constant. Therefore our specific hypothesis
here is that the standardized return spreads across D10 and D1 deciles sorted by return
predictors are identical in these two markets.
The results, reported in Panel B of Table 7, show that standardized return spreads in
China and U.S. exhibit patterns similar to those for unstandardized return spreads reported in
Table 4. Differences in standardized return spreads between China and US are negative in 14
predictors. Among them, China has significantly lower standardized return spreads in MOM,
E/P, C/P, -SG, -ACC, ADV, -AG, and –∆DT. That is, the standardized return spreads is also
lower in the Chinese market. Taken together, Table 7 suggests the weaker return predictability
in China is not merely due to there being less variation in the predictors themselves.
4.4.2 Return synchronicity
Despite that return predictability is often considered an indication of market
inefficiency, it is unlikely that weaker predictability in the Chinese market is due to higher
efficiency. We therefore explore another possible reason for low return predictability -- stock
prices are persistently uninformative in the Chinese market. Under the behavioral asset
pricing framework, stock return predictability depends on both initial mispricing and
subsequent price correction. Given initial mispricing, if stock prices are persistently noisy,
and correction never takes place, return predictability could still be weak. It is likely that there
is large (initial) mispricing in the Chinese stock market. However, it is also likely that stock
prices are so noisy that they seldom converge to the fundamental values.
To test this, we resort to a measure well-known in the literature – return synchronicity,
or R2 from regressing individual stock returns on market returns. Price synchronicity is a
measure of informativeness of stock prices. Low synchronicity is associated with greater
incorporation of firm-specific information on individual returns (Roll, 1988; Morck, Yeung,
and Yu, 2000; and Durnev, et al., 2003), thus higher price informativeness.
We test the difference in return synchronicity between the two stock markets.
Following Mock et al (2000), we perform market model regressions for all the stocks in the
Chinese and U.S. markets by regressing weekly stock returns onto weekly market returns in
each year (July of year t-1 to June of year t) during the period of 1994 to 2005 and obtain the
R-square statistics of each stock in each year. The weekly stock return is the compounded
daily return for a calendar week. We compound daily CRSP value-weighted returns to get
weekly market returns. Within each market, we first average the synchronicity measure across
stocks in each year then aggregate the measure over time. We find that the average market
model R2 in China 0.46, while it is 0.12 in the U.S. The much higher R2 in China supports the
view that stock prices are generally much less informative in China.
We then use a double-sort procedure to test whether there is a positive link between
price informativeness and return predictability. In each year, we first sort Chinese stocks into
three groups based on synchronicity. Then, within each synchronicity group, we sort stocks
into equal-weighted decile portfolios based on each of the 18 predictors. Finally, for each
predictor, we calculate the time-series average of the standardized return spreads (return
spread divided by spread in the predictor itself) between the top and bottom portfolios within
each synchronicity group. Under our hypothesis, the standardized return spread should be
higher among stocks with lower synchronicity. The results are reported in Table 8.
[Insert Table 8 Here]
From Table 8, we see that for 14 out of 18 predictors, the standardized return spread is
higher in the lowest synchronicity group relative to the highest synchronicity group. The
exceptions are MOM, -SG, -CPX, and ∆GPM. Out of the 14 predictors, 11 work significantly
better predicting returns in low synchronicity groups related to high synchronicity groups, as
inferred by the t-statistics on the difference in standardized return spreads between lowest and
highest synchronicity groups. These variables include B/P, E/P, C/P, -ACC, -NOA, RD, -
∆EQ, -∆DT, -STDR, -TURN, and ILLIQ.
To summarize, return predictability is stronger in groups where pricing
informativeness is higher. To some extent this explains the weaker return predictability in
Using Chinese stock returns data, we examine the predictive power of 18 firm- specific
variables that have been hypothesized to predict U.S. stock returns. Using portfolio-sorted
returns and regression analyses, we find that almost all firm-specific variables predict
subsequent one-year returns in their hypothesized direction. However, only 6 firm-specific
variables are statistically significant in their ability to predict raw returns, and 4 firm-specific
variables are statistically significant in their ability to predict risk-adjusted returns. We
conduct the same tests on U.S. stock returns from the same sample period and find that more
predictors are statistically significant in explaining subsequent cross-sectional stock return
variation, indicating that the stock return predictability is weaker in China than it is in the U.S.
We test two explanations for the cause of weak return predictability in China. One
possible explanation is that return predictors are more homogeneously distributed in the
Chinese market than they are in the U.S. market. While we find evidence of greater
homogeneity in the Chinese market, we also find the stock return sensitivity to return
predictors is lower in the Chinese market than in the U.S. Another possible explanation for
low return predictability in China is that there is high price inefficiency in China. We find
that the market model R2 , a measure of stock price uninformativeness, is much higher for
China than for the U.S. Further, across Chinese stocks, when comparing the return spread for
high and low R2 stocks, we find that more predictors work better for the bottom R2 stocks
than for top R2 stocks. This evidence suggests that weak return predictability is related to low
price informativeness. Our results also question the validity of the conventional perception of
the positive relationship between return predictability and market efficiency, at least in the
context of cross-country comparison.
Amihud, Y., 2002, Illiquidity and stock returns: Cross-section and time-series effects, Journal
of Financial Markets 5, 31–56.
Ang, A., R.J. Hodrick, Y. Xing, and X. Zhang, 2006, The cross-section of volatility and
expected returns, Journal of Finance 61, 259–299.
Abarbanell, J. and B. Bushee, 1998, Abnormal stock returns to a fundamental analysis
strategy, Accounting Review 73, 19-45
Banz, R.W., 1981, The relationship between return and market value of common stocks,
Journal of Financial Economics 9, 3–18.
Basu, S., 1977, Investment performance of common stocks in relation to their price-earnings
ratios: A test of the efficient market hypothesis, Journal of Finance 32, 663–682.
Barberis, N., A. Shleifer, and R. Vishny, 1998, A model of investor sentiment, Journal of
Financial Economics 49, 307–343.
Beneish, M.D., C.M. Lee, and R.L. Tarpley, 2001, Contextual fundamental analysis through
the prediction of extreme returns, Review of Accounting Studies 6, 65–189.
Billett, M.T., M.J. Flannery, and J.A. Garfinkel, 2006, Are bank loans special? Evidence on
the post-announcement performance of bank borrowers, Journal of Financial and
Qualitative Analysis 41, 733-751.
Bradshaw, M.T., S.A. Richardson, and R.G. Sloan, 2006, The relation between corporate
financing activities, analysts’ forecasts and stock returns, Journal of Accounting and
Economics 42, 53–58.
Chan, L., J. Lakonishok, and T. Sougiannis, 2001, The stock market valuation of research and
development expenditures, Journal of Finance 56, 2431–2456.
Chen, G., K.A. Kim, J.R. Nofsinger, and O.M. Rui, 2007, Trading performance, disposition
effect, overconfidence, representativeness bias, and experience of emerging market
investors, Journal of Behavioral Decision Making, in press.
Chen, K.C.W. and H. Yuan, 2004, Earnings management and capital resource allocation:
Evidence from China’s accounting-based regulation of rights issues, Accounting
Review 79, 645–665.
Collins, D.S., S.P. Kothari, and J. Rayburn, 1987, Firm size and the information content of
prices with respect to earnings, Journal of Accounting and Economics 9, 111-138.
Cooper, M.J., H. Gulen, and M.J. Schill, 2008, Asset growth and the cross-section of stock
returns, Journal of Finance 63(4), 1609-1651
Daniel, K.D., D. Hirshleifer, and A. Subrahamanym, 1998, Investor psychology and security
market under- and over-reaction, Journal of Finance 53, 1839–1886.
Datar, V.T., N.Y. Naik, and R. Radcliffe, 1998, Liquidity and stock returns: An alternative
test, Journal of Financial Markets 1, 203–219.
Durnev, A., R. Morck, B. Yeung, and P. Zarowin, 2003, Does greater firm-specific return
variation mean more or less informed stock pricing?, Journal of Accounting Research
Eun, C.S. and W. Huang, 2007, Asset pricing in China’s domestic stock markets: Is there a
logic?, Pacific-Basin Finance Journal 15, 452-480
Fama, E.F. and K.R. French, 1992, The cross-section of expected stock returns, Journal of
Finance 47, 427–465.
Fama, E.F. and K.R. French, 1993, Common risk factors in the returns on stocks and bonds,
Journal of Financial Economics 33, 3–56.
Fama, E.F. and K.R. French, 1996, Multifactor explanations of asset pricing anomalies,
Journal of Finance 51, 55–84.
Fama, E.F. and K.R. French, 1998, Value versus growth: The international evidence, Journal
of Finance 53, 1975–1999.
Fama, E.F. and J. MacBeth, 1973, Risk, return, and equilibrium: Empirical tests, Journal of
Political Economy 81, 607-636.
Feng F. and M. Seasholes, 2005, Do Investor Sophistication and Trading Experience
Eliminate Behavioral Biases in Financial Markets? Review of Finance 9, 305-351.
Grinblatt, M. and B. Han, 2005, Prospect theory, mental accounting, and momentum, Journal
of Financial Economics 78, 311–339.
Haw, I. D. Qi, D. Wu, and W. Wu, 2005. Market consequences of earnings management in
response to security regulations in China, Contemporary Accounting Research 22, 95–
He, X., 1998, Stock market manipulation behavior and regulatory suggestions, Soft Science in
Hirshleifer, D., K. Hou, S. Teoh, and Y. Zhang, 2004, Do investors overvalue firms with
bloated balance sheets? Journal of Accounting and Economics 38, 297–331.
Hong, H. and J. Stein, 1999, A unified theory of underreaction, momentum trading and
overreaction in asset markets, Journal of Finance 54, 2143–2184.
Ikenberry, D., J. Lakonishok, and T. Vermaelen, 1995, Market underreaction to open market
share repurchases, Journal of Financial Economics 39, 181–208.
Jegadeesh, N., J. Kim, S. Krische, and C. Lee, 2004, Analyzing the analysts: When do
recommendations add value? Journal of Finance 59, 1083–1124.
Jegadeesh, N. and S. Titman, 1993, Returns to buying winners and selling losers: Implications
for stock market efficiency, Journal of Finance 48, 65–91.
Jiang, G., T. Yao, and D. Xu, 2006, The information content of idiosyncratic volatility,
Journal of Financial and Quantitative Analysis, forthcoming
Jian, M. and T.J. Wong, 2004, Earnings management and tunneling through related party
transactions: Evidence from Chinese corporate groups, Working paper, Chinese
University of Hong Kong.
Kang, J., M. Liu, and S. Ni, 2002, Contrarian and momentum strategies in the China stock
market: 1993–2000, Pacific-Basin Finance Journal 10, 243–265.
Lakonishok, J., A. Shleifer, and R.W. Vishny, 1994, Contrarian investment, extrapolation,
and risk, Journal of Finance 49, 1541–1578.
Lakonishok, J. and T. Vermaelen, 1990, Anomalous price behavior around repurchase tender
offers, Journal of Finance 45, 455–477.
Lee C. and B. Swaminathan, 2000, Price momentum and trading volume, Journal of Finance
Loughran, T. and J. Ritter, 1997, The operating performance of firms conducting seasoned
equity offerings, Journal of Finance 52, 1823-1850.
Morck, R, B. Yeung, and W. Yu, 2000, The information content of stock marke ts: shy do
emerging markets have synchronous stock price movements? Journal of Financial
Economics 59, 215–60.
Reinganum, M., 1981, Misspecification of capital asset pricing: Empirical anomalies based on
earnings’ yields and market values, Journal of Financial Economics 9, 19–46.
Ritter, J., 1991, The long-run performance of initial public offerings, Journal of Finance 46,
Roll, R., 1988, R2 , Journal of Finance 43, 541–66.
Rouwenhorst, K.G., 1998, International momentum strategies, Journal of Finance 53, 267-
Rouwenhorst, K.G., 1999, Local return factors and turnover in emerging stock markets,
Journal of Finance 54, 1439-1464.
Shengzhen Stock Exchange, 2005, The major illegal behaviors in the Chinese stock market:
Case study on insider trading and price manipulation. Shengzhen.
Shumway, T. and G. Wu, 2006, Does disposition drive momentum? Working Paper,
University of Michigan and University of Houston.
Sloan, R., 1996, Do stock prices fully reflect information in accruals and cash flows about
future earnings? Accounting Review 71, 289–315.
Spiess, D.K. and J. Affleck-Graves, 1999, The long-run performance of stock returns
following debt offerings, Journal of Financial Economics 54, 45–73.
Titman S., J. Wei, and F. Xie, 2004, Capital investments and stock returns, Journal of
Financial and Quantitative Analysis 39, 677–700.
Wang, C and S. Chin, 2004, Profitability of return and volume-based investment strategies in
China’s stock market, Pacific-Basin Finance Journal 12, 541–564.
Wang, C., 2004, Relative strength strategies in China’s stock market: 1994–2000, Pacific-
Basin Finance Journal 12, 159–177.
Wang, F. and Y. Xu, 2004, What determines Chinese stock returns? Financial Analysts
Journal 60, 65–77.
Constructing Stock Return Predictors
We provide the variable name, description, co mputation details for the eighteen return predictors used in this paper. Predict ors are constructed in June of year t.
For all the financial statement items, year t refers to the fiscal year ending in the calendar year t. For the U.S. data, we provide data item nu mbers in the CRSP
and COMPUSTAT datasets.
Vari able Descripti on Computation Detail: China Computation Detail: United States
1. SIZE Natural logarith m of ln(closing price at the end of June mu ltip lied by A shares ln(closing price at the end of June [CRSP] mu ltip lied by
market cap italization outstanding at the end of June) common shares outstanding at the end of June[CRSP])
2. B/P Book to price BVt -1 , where BV o f A Shares = Book Value of BVt -1 , where BV is the book value of
MVt -1 MVt -1 in December
Equity*(Nu mber of outstanding A Shares in stockholders’ equity [D216], plus balance sheet deferred taxes
December/Total Shares Outstanding); MVt-1 is the market and investment tax cred it [D35, if availab le], minus the book
value of A shares at the end of year t-1. value of preferred tax [D56, D10, D130, in that order].
3. M OM Cu mulat ive market- im 112 (1+monthly return t ) – im 112 (1 + market
m m im 112 (1+monthly return t ) – im 112 (1 + market monthly
adjusted return for the
preceding 12 months monthly return t ), where m is June of year t return t ), where m is June of year t
4. E/ P Earnings to price Net Income t -1 where M Vt-1 is market capitalization of A Earnings before Extraordin ary Item t -1 [D18]
MVt -1 MVt -1 [D24 * D25]
shares at the end of year t – 1
5. C/P Cash flow to price Net Income t -1 Depreciati on t -1 where M Vt-1 is market Earnings before Extraordin ary Item t -1 [D18] Depreciati on t -1 [D14]
MVt -1 , MVt -1 [D24 * D25]
capitalizat ion of A shares at the end of year t – 1
6. SG Sales growth Salest 1 Salest 1 [ D12 ]
Salest 2 Salest 2 [ D12 ]
7. A CC Accruals [(CA – CASH– SI) t–1 – (CL – STD – LTDC – [(CA – CASH)t–1 – (CL – STD – TP)t–1 – DEPt–1 ]/ATAt–
TP)t–1 – DEPt–1 ]/ATAt–1 , where CA is the change in 1 , where CA is the change in current assets [D4] fro m
current assets from previous fiscal year; Cash is the previous fiscal year; Cash is the change in cash/cash
change in cash, SI is the change in net short term equivalents [D1]; CL is the change in current liabilities [D5];
investment; CL is the change in current liab ilities; STD STD is the change in debt included in current liabilities
is the change in short-term debt; LTDC is the change in [D34]; TP is the change in income taxes payable [D71]; DEP
long term debt included in current liabilit ies; TP is the is depreciation and amort ization expense [D14]; and ATA t-1 is
change in taxes payable; DEP is change in accumu lative the average of the beginning and ending total assets [D6] o f the
depreciation; and ATAt-1 is the average of the beginning reporting year t-1.
and ending total assets (TA)of the reporting year t-1.
Vari able Descripti on Computation Details: China Computation Details: United States
8. NOA Net operating assets [Operating Assets – Operating Liabilities]t–1 /ATAt–1 , where [Operating Assets – Operating Liabilities]t–1 /ATAt–1 , where
operating assets = TA – cash – short-term investment; operating assets = TA [D6] – cash and short-term investment
operating liabilities = total asset – STD – LTDC – LTD – [D1]; operating liab ilities = total asset – STD – LTD – MI – PS
MI – CE; STD = short-term debt; LTDC = long-term debt – CE; STD = debt included in current liab ilities [D34], LTD =
included in current liab ilit ies; LTD = total long-term debt; long-term debt [D9]; M I = minority interests [D38]; PS =
MI = minority interests; CE = common equity; and ATA = preferred stocks [D130]; CE = co mmon equity [D60]; and
the average of the beginning and ending total assets of the ATA = the average of the beginning and ending total assets
reporting year. The values of short-term debt, taxes [D6] of the reporting year. The values of short-term debt, taxes
payable, long-term debt, and minority interest are set as payable, long-term debt, minority interest, and preferred stock
zero if they are missing. are set as zero if they are missing.
9. CAPEX Capital expenditure to CapExt 1 , where Cap Ex is the change of net fixed assets CapExt 1 , where Cap Ex is capital expenditure [D128] in
total assets ATAt 1 ATAt 1
in fiscal year t – 1 p lus the change in accumulated year t-1, and ATA is the average of the beginning and ending
depreciations in year t –1, and ATA is the average of the total assets (TA, [D6]) of the reporting year t-1.
beginning and ending total assets (TA) of the reporting
10. RD Research and SGAE t 1 , where SGA E is the selling, general and R & Dt 1 , where R&D is research and development expenses
Develop ment MVt 1 MV t 1
expenses to market
administrative expenses at the end of year t-1; M Vt-1 is [D46]; market value is market capitalizat ion at the end of year t-
value of equity
market cap italization of A shares at the end of year t-1 1
11. ADV Advertising expenses Sales Expenses t -1 , where M V is market capitalization of ADVt 1 , where ADV advertising expenses [D45] in year t –
to sales MVt 1 MVt 1
A shares at the end of year t – 1. 1; M Vt-1 is market capitalization at the end of year t – 1.
12. A G Change in total assets TAt 1 TAt 2 , where TA is total assets. TAt 1 TAt 2 , where TA is total assets [D6].
TAt 2 TAt 2
13. ∆GPM Change in gross profit St 1 CGt 1 St 2 CGt 2 , where St-1 and CG are net sales St 1 CGt 1 St 2 CGt 2 , where St-1 and CGt-1 are net sales
margin St 1 St 2 St 1 St 2
and the cost of goods sold in year t-1. [D12] and cost of goods sold [D41] in year t-1.
14. ∆EQ Net cash flow CTt 1 CS t 1 , where ∆CT is the change in common CREt 1 CPEt 2 DIV t 1 , where CRE is cash received fro m the
received fro m ATAt 2 ATAt 2
external equity stock, ∆CS is the change in capital surplus, and ATA is the sale of common and preferred stock [D108], CPE is cash paid
financing average of the beginning and ending total assets of the for purchase of common and preferred stock [D115], DIV is
reporting year. cash dividends paid [D127], and ATA is the average of the
beginning and ending total assets [D6] of the report ing year.
Vari able Descripti on Computation Detail: China Computation Detail: United States
15. ∆DT Net cash flow LTD t 1 LTN t 2 STDt 1 , where ∆LTD is the change in CRDt 1 CPDt 2 CDt 1 , where CRD is cash received fro m the
received fro m ATAt 2 ATAt 2
external debt long-term debt, ∆LTN is the change in long term note, issuance debt [D111], CPD is cash paid for reduction of debt
financing ∆STD is the change in total short term debt, and ATA is [D114], ∆CD is change in current debt [D301], and ATA is the
the average of the beginning and ending total assets of the average of the beginning and ending total assets [D6] of the
reporting year. reporting year.
16. STDR Idiosyncratic risk STDR is the standard deviation of the error term in daily STDR is the standard deviation of the error term in daily data
data market model regression: market model reg ression:
k 5 k 5
Ri , i
k ,i * R m , k i ,t , where Ri,τ is daily stock Ri , i
k ,i * R m , k i ,t , where Ri,τ is daily stock
return in the 12 months preceding June of each year; Rm,τ+k return in the 12 months preceding June of each year, R m,τ+k is
is the value-weighted average of the Shanghai Stock the value-weighted CRSP market daily returns fro m 5 lags and
Exchange (SHSE) and Shenzhen Stock Exchanges (SZSE) 5 leads.
value-weighted daily index return fro m 5 lags and 5 leads.
t 1 t 1
17. TURN Average daily volu me
turnover Daily Volume/Shares Outstanding
Daily Volume/Shares Outstanding
Percentile rank in Percentile rank in
SHSE and SZSE, respectively, where n is number of days NASDA Q and NYSE/AM EX, respectively, where n is number
available for 12 months preceding the end of June in each of days available for 12 months preceding the end of June in
year. each year.
t 1 t 1
18. ILLIQ Amihud illiquidity
i, t | / VOLDi ,t
i, t | / VOLDi ,t
Percentile rank in SHSE and SZSE, Percentile rank in NASDAQ and
respectively, where |Ri,τ | is the return on stock i on NYSE/AMEX, respectively, where |Ri,τ | is the return on stock i
day within 12 months preceding June of each year, on day within 12 months preceding June of each year,
VOLDi , is the respective daily volu me in RMB, and n is VOLDi , is the respective daily volu me in dollars, and n is
number of days available for 12 months preceding the end number of days available for 12 months preceding the end of
of June in each year. June in each year.
Hypothesized Signs of Stock Return Predictors
For each stock market predictor (i.e., variable), we test its relation to subsequent stock returns. For variables that are
hypothesized to be negatively correlated with subsequent stock returns in the U.S. market, we adjust them so they
can positively predict returns. For examp le, because firm size (SIZE) and subsequent returns are posited to have a
negative correlation, we transform the SIZE variable to be: –SIZE. A brief description of each variable is provided.
Vari able Descripti on
-SIZE -Firm size (market capitalization)
B/P Book-to-price ratio
MOM Momentum (past returns)
E/P Earnings-to-price ratio
C/P Cash flow-to-price ratio
-SG -Sales growth fro m prior year
-ACC -Accruals-to-total assets ratio
-NOA -Net operating assets-to-total assets ratio
-CPX -Cap ital expenditures to total assets
RD Research and development expenditures -to-market value of equity ratio
ADV Advertising expenditures-to-market value of equity ratio
-A G -Assets growth fro m prior year
ΔGPM Change in gross profit margin
-ΔEQ -Net cash flo w received fro m external equity financing
-ΔDT -Net cash flo w received fro m debt financing
-STDR -Idiosyncratic risk
-TURN -Trading volu me turnover
This table reports market characteristics in China and the U.S. in June of the sample years fro m 1994 to 2007. Panel
A is for the Chinese market and Panel B is for the U. S. market. N is the number of listed firms. Tu rnover is annual
trading volume scaled by year-end shares outstanding. Price is the stock price in June of the samp le year. For the
Chinese market, it is respectively for A-, B-, and H-share price. Market values for A-, B- and H-shares are the
closing prices mult iplied by shares outstanding in June. For the Chinese market, the total market value is the sum of
the market value of A-, B-, H-shares, and non-floating shares, where non-floating share price is set at the price of the
A-shares. We use equal weight for the year-end cross-sectional averages.
Panel A: The Chinese Market
N T urnover Price Market Value (RMB billions)
T otal SHSE SZ SE SHSE SZ SE A B H A B H T otal
1994 287 169 118 12.12 7.74 5.52 4.08 2.28 56.48 12.98 8.56 475.50
1995 311 184 127 5.04 2.96 6.92 2.89 2.33 80.35 12.28 13.58 322.08
1996 514 287 227 7.99 14.40 7.65 2.95 1.69 148.17 16.20 12.04 634.68
1997 720 372 348 6.75 8.00 12.76 4.38 2.61 459.83 39.60 14.72 1597.72
1998 825 425 400 4.90 4.66 12.54 2.14 1.19 588.54 23.04 9.61 2018.66
1999 923 471 452 4.51 4.25 13.85 2.95 1.81 880.56 33.43 21.77 2325.03
2000 1060 559 501 5.25 5.21 15.49 2.72 1.38 1283.87 31.52 14.15 3595.05
2001 1139 637 502 2.43 2.06 16.90 9.88 2.42 1747.55 113.16 25.38 4999.48
2002 1206 705 501 2.18 2.06 12.19 6.21 1.79 1455.05 79.75 61.65 4320.93
2003 1266 772 494 2.37 2.11 8.97 4.55 2.22 1262.28 64.64 85.93 4010.72
2004 1362 831 531 3.36 3.25 7.03 3.87 3.31 1179.52 63.67 147.83 4880.63
2005 1365 829 536 3.51 4.15 4.80 3.04 3.49 934.71 56.64 188.52 3693.71
2006 1417 835 582 6.28 6.44 6.70 3.71 4.65 1561.23 75.13 1355.59 4352.54
2007 1516 852 664 10.87 10.79 14.85 8.13 8.00 5294.60 176.38 3101.86 13464.99
T otal/Average 1584 876 708 5.03 5.09 10.77 4.48 3.63 -- -- -- --
Panel B. The U.S. market
N T urnover Price Market Value (US$ billions)
T otal AMEX NASDAQ AMEX NASDAQ (US$)
1994 8084 3718 4366 0.78 1.38 16.75 4959.53
1995 8242 3723 4519 0.84 1.53 19.06 6052.35
1996 8765 3819 4946 0.99 1.66 21.37 7734.33
1997 9058 3860 5198 1.02 1.62 24.04 9922.80
1998 9014 3857 5157 1.42 1.81 28.32 12856.57
1999 8455 3706 4749 1.03 2.31 26.87 15279.38
2000 8284 3533 4751 1.18 2.32 25.79 17898.83
2001 7663 3361 4302 1.49 1.70 26.32 15053.20
2002 7140 3299 3841 1.64 1.44 25.96 12656.96
2003 6690 3235 3455 1.40 1.85 28.61 12756.17
2004 6603 3280 3323 1.58 2.41 35.15 15597.49
2005 6651 3382 3269 1.70 2.16 35.48 16906.57
2006 6645 3421 3224 2.08 2.10 38.08 18497.99
2007 6749 3578 3171 3.26 2.26 44.29 21835.03
T otal/Average 15486 6477 9009 1.33 1.77 27.75 --
Summary Statistics for Stock Return Predictors
This table presents the time-series averages of cross sectional averaged statistics of return predicto rs for China and
the U.S. market for our sample period. Details of these variables are provided in the Appendix. Equal weight is used
to compute cross-sectional averages.
Panel A: Time-series average of cross-sectional distribution of stock return predictors in China
25% Mean Median 75% Std. Dev. Skewness Kurtosis
SIZE(Mil RM B) 442.477 986.909 674.261 1105.336 1244.608 8.140 108.236
B/P 0.701 1.365 1.022 1.535 1.419 5.640 65.871
MOM -0.054 0.259 0.160 0.450 0.555 3.814 52.734
E/P 0.045 0.096 0.088 0.150 0.214 -4.130 69.194
C/P 0.064 0.161 0.118 0.217 0.348 1.679 61.232
SG -0.024 0.366 0.132 0.358 2.273 13.346 292.576
ACC -0.057 0.003 0.004 0.063 0.137 -0.632 11.333
NOA 0.590 0.687 0.705 0.802 0.178 -1.382 13.181
CPX 0.008 0.074 0.047 0.120 0.140 1.514 20.441
RD 0.038 0.121 0.070 0.132 0.182 6.594 82.099
ADV 0.009 0.055 0.023 0.060 0.098 7.419 96.267
AG 0.026 0.196 0.129 0.287 0.334 4.134 38.287
ΔGPM -0.097 -0.013 -0.007 0.085 0.697 0.599 104.805
ΔEQ 0.000 0.040 0.008 0.048 0.097 0.610 40.086
ΔDT -0.012 0.042 0.026 0.088 0.113 0.537 8.368
STDR 0.016 0.019 0.019 0.022 0.006 0.740 4.681
TURN 0.335 1.425 0.733 1.503 7.881 12.111 16.923
ILLIQ( *10-7 ) 3.771 13.052 9.415 16.674 17.116 6.320 131.524
Panel B: Time-series average of cross-sectional distribution of stock return predictors in United States
SIZE(Mil USD) 94.032 2272.371 288.693 972.618 11370.977 14.652 296.271
B/P 0.291 0.793 0.509 0.788 14.548 1.386 993.820
MOM -0.088 0.115 0.107 0.307 0.419 0.362 3.624
E/P 0.016 0.099 0.049 0.075 5.921 -15.330 1270.846
C/P 0.042 0.188 0.081 0.125 7.699 -4.370 1126.556
SG 0.022 0.494 0.122 0.296 13.419 39.575 2091.711
ACC -0.069 -0.026 -0.028 0.012 0.105 0.526 25.047
NOA 0.325 0.546 0.592 0.756 0.296 -0.301 0.369
CPX 0.011 0.226 0.036 0.082 7.868 37.804 2014.799
RD 0.142 0.349 0.235 0.344 1.754 34.712 1617.333
ADV 0.006 0.048 0.017 0.047 0.109 7.901 106.583
AG 0.014 0.465 0.108 0.288 4.115 29.451 1328.820
ΔGPM -0.025 -0.008 -0.002 0.020 0.352 -1.532 213.056
ΔEQ -0.015 0.051 -0.001 0.012 0.231 4.462 35.806
ΔDT -0.015 0.017 0.000 0.025 0.127 2.430 26.952
STDR 0.015 0.026 0.023 0.034 0.015 2.303 19.852
TURN 0.283 0.443 0.398 0.561 0.228 1.510 3.957
ILLIQ ( *10-7 ) 0.473 9.362 2.343 9.930 34.453 3.012 53.385
Returns to Decile Portfolios Sorted on Stock Return Predictors
This table reports the average monthly returns of stock portfolio deciles in China and the U.S. Panel A shows the decile retu rns, as well and the return spread
between D10 and D1 stocks, sorted by each of the 18 return pred ictors and by a comb ined return predictor (COMBO) of all 18 variab les for Ch ina market. Panel
B shows the return spread between D10 and D1 stocks for the U.S. market. Panel C shows the difference in the return spread between the Ch ina market and the
U.S. market. In June of each year t, stocks are sorted into deciles based on each predictor, where D10 represents the highest decile of the signed predictor, and D1
is the lowest. Decile portfolios are held unchanged from July of year t to June of year t+1. The 18 predictors are defined in the Appendix. COM BO is the average
percentile rank of all 18 return predictors weighted by their sensitivities to subsequent stock returns. To be included in the analysis, stock price is restricted to be
no less than ¥1 fo r Ch ina market and $1 for the U.S. market in June o f each year. The accounting data is fro m fiscal year 1994 to 2005 and stock return data is
fro m Ju ly 1995 to June 2007.
Panel A: China
-SIZE B/P MOM E/P C/P -SG -ACC -NOA -CPX RD ADV -AG ∆GPM -∆EQ -∆DT -ST DR -T URN ILLIQ COMBO
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19)
D1 1.84 1.49 1.74 2.27 2.59 1.97 1.93 1.83 1.83 1.42 1.35 1.67 1.76 1.81 2.00 1.67 1.78 1.67 1.52
2 2.13 1.57 2.16 2.10 1.97 1.98 1.84 2.34 1.74 2.01 0.91 1.99 1.88 1.94 1.90 1.97 1.98 1.81 1.61
3 1.91 1.96 2.06 1.92 2.01 2.15 2.13 1.97 2.08 1.90 1.27 1.84 1.82 2.27 1.95 2.14 1.99 2.05 2.00
4 1.76 2.07 2.27 1.88 2.02 2.17 1.93 2.12 2.01 1.74 1.21 2.06 2.19 2.39 1.98 2.19 2.06 2.15 2.05
5 1.99 2.29 2.46 2.06 1.75 2.13 2.06 1.99 1.99 2.27 1.38 2.26 1.95 2.20 2.04 2.18 2.25 2.11 1.89
6 2.04 2.39 2.21 2.02 2.01 2.18 2.37 1.99 2.22 2.31 2.06 2.08 2.08 2.27 2.07 2.19 2.29 1.97 2.36
7 2.23 2.46 2.27 2.10 2.25 2.24 1.98 1.88 2.43 2.18 1.39 2.11 2.56 1.50 2.29 2.18 2.14 2.23 2.37
8 2.20 2.25 2.04 2.11 2.00 2.01 2.19 2.15 2.03 2.46 1.48 2.15 2.56 2.23 2.22 1.99 2.07 2.17 2.28
9 2.23 2.24 1.94 2.15 2.01 2.23 2.32 2.22 2.19 2.24 1.67 2.35 2.33 3.84 2.23 1.88 2.29 2.36 2.43
D10 2.43 2.23 1.79 2.26 2.27 2.06 2.28 2.38 2.33 2.48 1.71 2.33 2.07 2.05 2.20 2.26 2.03 2.33 2.24
10–1 0.59 0.74 -0.05 -0.01 -0.33 0.09 0.35 0.56 0.50 0.87 0.17 0.66 0.31 0.24 0.20 0.59 0.25 0.66 0.72
t-stat 0.95 1.80 -0.16 -0.01 -0.54 0.18 1.62 2.58 1.14 2.52 1.02 1.72 0.82 0.57 0.34 1.35 0.97 1.65 1.98
Panel B: US
10–1 0.22 1.02 0.65 0.91 1.11 0.87 0.73 0.92 0.25 0.37 1.08 1.31 0.09 0.94 0.63 0.28 0.62 0.42 0.89
t-stat 0.79 2.04 1.61 1.97 1.95 2.36 2.30 4.37 0.39 0.51 2.04 3.44 0.59 1.75 3.25 1.01 0.98 1.19 2.88
Panel C: China – US
D10-D1 0.37 -0.38 -0.70 -0.92 -1.44 -0.78 -0.38 -0.36 0.25 0.50 -1.02 -0.65 0.22 -0.70 -0.43 0.11 -0.36 0.24 -0.17
t-stat 0.52 -0.62 -1.84 -1.65 -1.67 -1.71 -1.49 -1.16 0.33 0.52 -2.01 -1.70 0.82 -1.86 -2.40 0.12 -0.55 0.37 -0.25
Al phas to Decile Portfolios Sorted on Stock Return Predictors
This table reports the average monthly three-factor alphas of stock portfolio deciles in China and the U.S. Panel A shows the decile returns, as well and the return spread
between D10 and D1 stocks, sorted by each of the 18 return predictors and by a combined return predictor (COMBO) of all 18 variables for China market. Panel B shows
the alpha spreads between D10 and D1 stocks for the U.S. market. Panel C shows the difference in the alpha spreads between the China market and the U.S. market. In
June of each year t, stocks are sorted into deciles based on each predictor, where D10 represents the highest decile of the s igned predictor, and D1 is the lowest. Decile
portfolios are held unchanged from July of year t to June of year t+1. The 18 predictors are defined in the Appendix. COMBO is the average percentile rank of all 18
return predictors weighted by their sensitivities to subsequent stock returns. The risk-adjusted return is the intercept from time-series monthly regressions of decile
portfolio returns on the three factors: RMRFt is the market return in excess of the risk free rate; SMBt and HMLt are the monthly returns on size and book-to-market
factors, respectively, computed following Fama-French (1993) procedure. To be included in the analysis, stock price is restricted to be no less than ¥1 for China market
and $1 for the U.S. market in June of each year. The accounting data is from fiscal year 1994 to 2005 and stock return data is from July 1995 to June 2007.
Panel A: China
-SIZE B/P MOM E/P C/P -SG -ACC -NOA -CPX RD ADV -AG ∆GPM -∆EQ -∆DT -ST DR -T URN ILLIQ COMBO
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19)
D1 -0.87 -1.32 -1.40 -1.15 -1.18 -0.91 -1.17 -1.18 -0.92 -1.51 -0.86 -1.11 -1.23 -1.00 -0.98 -1.50 -1.20 -1.09 -1.38
2 -0.85 -1.26 -0.95 -1.46 -1.64 -0.77 -1.30 -0.89 -0.85 -1.16 -1.31 -0.80 -1.02 -1.21 -1.20 -1.20 -0.95 -1.13 -1.30
3 -1.10 -1.15 -0.87 -1.33 -1.22 -0.83 -1.21 -1.28 -1.01 -1.31 -1.06 -1.05 -0.95 -1.14 -1.16 -1.01 -1.10 -1.10 -0.97
4 -1.26 -1.03 -0.98 -1.38 -1.19 -0.92 -1.17 -1.09 -0.96 -1.34 -1.05 -1.12 -0.76 -0.92 -1.10 -0.87 -1.12 -1.00 -1.14
5 -1.07 -0.96 -1.16 -1.22 -1.27 -0.94 -1.26 -1.07 -1.11 -0.95 -1.17 -0.90 -0.92 -1.14 -1.24 -0.90 -0.95 -1.20 -1.13
6 -1.14 -1.02 -1.27 -0.97 -1.07 -1.06 -1.06 -1.17 -0.91 -0.81 -1.00 -1.20 -1.15 -1.04 -0.90 -1.09 -0.82 -1.17 -0.95
7 -1.04 -0.95 -0.92 -0.93 -0.72 -0.94 -0.88 -1.06 -0.98 -1.00 -0.75 -1.06 -1.00 -1.05 -0.90 -0.87 -1.04 -0.94 -0.73
8 -1.04 -0.92 -0.86 -0.65 -0.85 -1.30 -0.75 -1.02 -1.24 -0.82 -0.79 -0.98 -0.97 -0.66 -1.10 -0.97 -0.99 -1.07 -0.88
9 -0.91 -1.00 -0.85 -0.64 -0.71 -1.31 -0.68 -0.92 -1.16 -0.85 -0.58 -0.96 -1.05 -1.32 -0.81 -0.86 -0.86 -0.70 -0.78
D10 -0.59 -0.80 -0.88 -0.70 -0.75 -1.26 -0.96 -0.73 -1.26 -0.67 -0.58 -0.65 -1.42 -1.20 -1.04 -0.64 -1.08 -0.77 -0.59
10–1 0.28 0.52 0.52 0.45 0.43 -0.35 0.21 0.45 -0.34 0.79 0.23 0.45 -0.19 -0.20 -0.06 0.86 0.12 0.32 0.79
t-stat 0.85 1.59 1.14 1.23 1.16 -1.18 0.90 2.05 -1.12 2.48 1.27 2.19 -0.50 -0.88 -0.35 2.72 0.57 1.58 2.42
Panel B: US
10–1 0.10 1.31 0.86 1.30 1.56 1.08 0.32 0.85 0.58 0.19 1.25 1.51 -0.06 1.46 0.67 0.85 0.87 0.40 1.07
t-stat 0.33 2.45 1.62 2.65 2.59 2.99 1.59 3.77 0.86 0.21 2.20 3.71 -0.37 2.30 5.45 0.96 1.30 1.04 2.94
Panel C: China – US
D10-D1 0.15 -0.70 -0.41 -0.97 -1.08 -1.55 -0.08 -0.37 -0.82 0.45 -1.07 -1.04 -0.23 -1.75 -0.65 0.05 -0.70 0.12 -0.35
t-stat 0.25 -1.76 -0.72 -2.14 -2.25 -2.16 -0.36 -0.79 -1.87 0.93 -2.18 -2.07 -0.49 -2.31 -1.69 0.13 -1.81 0.33 -1.46
This table reports the time -series averages of the coefficients in the regression of monthly stock return on stock return pred ictors in Ch ina and U.S. markets.
Panel A and B show the result of univariate regressions for Ch ina market and U.S. markets. Panel C and D show the result of a jo int regression including all the
18 p redictors for both markets. AdjR2 is the adjusted R2 for mult ivariate regressions using all 18 predictors and AdjR2 * is the adjusted R2 for multivariate
regression using the first 6 principal co mponents. If a stock return predictor is missing, it is replaced with the annual cro ss-sectional median. To be included in
the analysis, stock price is restricted to be no less than ¥1 for China market and $1 for the U.S. market in June of each year. The accounting data is from fiscal
year 1994 to 2005 and stock return data is fro m July 1995 to June 2007.
-SIZE B/P MOM E/P C/P -SG -ACC -NOA -CPX RD ADV -AG ∆GPM -∆EQ -∆DT -ST DR -T URN ILLIQ AdjR2 AdjR2 *
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
Panel A. Univariate regressions for the Chinese market
Coeff 0.17 0.17 0.35 0.45 0.26 0.08 0.59 0.61 1.10 1.46 1.07 0.61 1.00 0.74 0.27 21.37 0.32 0.61 -- --
t-stat 0.67 2.12 0.67 0.55 0.54 0.64 1.14 1.86 1.30 2.22 0.81 1.77 1.42 0.84 0.61 1.10 1.63 1.83 -- --
Panel B. Univariate regressions for the U.S. market
Coeff 0.03 0.62 0.44 2.04 2.62 0.30 1.34 0.73 0.48 -0.31 1.68 0.27 0.03 0.80 1.86 4.80 0.61 0.48 -- --
t-stat 0.61 2.22 1.74 1.41 1.95 1.61 2.47 3.18 0.88 -0.56 1.79 2.20 0.08 1.16 5.84 0.31 0.90 1.17 -- --
Panel C. Multivariate regressions for the Chinese market
Coeff 0.34 0.01 0.30 0.18 0.16 -0.09 0.60 0.27 0.28 0.20 0.64 0.43 1.07 -0.91 -0.90 20.60 0.00 0.00 0.09 0.05
t-stat 1.47 0.15 0.77 0.20 0.39 -0.99 1.67 0.88 0.43 0.38 0.46 1.50 1.78 -1.27 -1.44 1.72 1.28 -0.26 -- --
Panel D. Multivariate regressions for the U.S. market
Coeff -0.04 0.33 0.33 -0.76 2.05 0.01 0.35 0.81 -0.24 -0.01 -0.26 -0.01 -0.25 0.62 1.20 -4.72 0.00 0.00 0.06 0.05
t-stat -0.70 3.06 1.37 -1.12 2.75 0.12 0.91 5.17 -0.67 -0.03 -0.26 -0.13 -0.91 2.44 4.30 -0.40 0.20 0.88 -- --
Standardized Return S preads
This table reports the time-series average of standardized stock return spreads between D10 and D1 portfolios sorted by return predictors for the China and U.S. markets.
The standardized return spread (D10-D1) of a return predictor is the stock return spreads scaled by the predictor spreads in a corresponding year. The difference of the
standardized return spread between the China and U.S. markets is the difference in the standardized return spreads in these t wo markets. Decile portfolios are formed in
the end of June and held unchanged from July of year t to June of year t+1 for each stock decile. To be included in the analysis, stock price is restricted to be no less
than ¥1 for Ch ina market and $1 for the U.S. market in June of each year. The accounting data is fro m fiscal year 1994 to 2005 and stock return data is fro m July
1995 to June 2007. The t-statistics are included at the bottom of each Panel.
-SIZE B/P MOM E/P C/P -SG -ACC -NOA -CPX RD ADV -AG ∆GPM -∆EQ -∆DT -ST DR -T URN ILLIQ
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
Panel A: Spreads of Average Return Predictors between D10 and D1 Stocks
China 2.53 3.98 1.50 0.80 1.17 3.54 0.49 0.63 0.47 0.58 0.39 1.02 0.81 0.28 0.41 0.02 0.90 0.90
US 6.21 4.33 1.54 1.36 1.67 4.05 0.36 0.99 1.84 1.41 0.27 3.72 0.60 0.71 0.43 0.05 0.90 0.90
China-US -3.68 -0.35 -0.04 -0.56 -0.50 -0.51 0.13 -0.36 -1.38 -0.88 0.12 -2.70 0.20 -0.43 -0.02 -0.03 0.00 0.00
(t-stat) -25.97 -0.17 -0.18 -0.74 -0.48 -0.38 6.79 -6.81 -1.38 -7.30 1.14 -2.65 0.97 -5.61 -0.67 -7.59 -0.11 -0.29
Panel B: Standardized Return Spreads between D10 and D1 Stocks
China 0.16 0.23 -0.01 0.30 0.02 -0.10 0.07 0.79 0.90 1.58 0.79 0.52 0.45 -0.59 0.00 9.82 0.77 0.51
(t-stat) 0.70 2.27 -0.04 0.35 0.03 -0.60 0.14 1.94 0.87 2.41 1.28 1.06 0.91 -0.62 0.00 0.44 1.05 1.26
US 0.03 0.48 0.43 1.36 1.48 0.29 1.18 1.04 0.52 -0.10 4.21 1.58 -0.22 1.30 1.66 5.84 0.24 0.51
(t-stat) 0.76 1.82 1.63 1.59 1.60 2.12 2.22 4.89 0.52 -0.19 2.08 3.50 -0.36 1.45 3.02 0.32 0.81 0.95
China-US 0.13 -0.25 -0.44 -1.06 -1.46 -0.39 -1.11 -0.25 0.38 1.77 -4.46 -1.07 0.67 -1.89 -1.65 3.98 -0.52 -0.01
(t-stat) 0.52 -0.89 -1.72 -1.85 -1.65 -1.87 -1.85 -0.54 0.27 1.99 -2.47 -2.13 0.89 -1.56 -2.41 0.14 -0.69 0.00
Standardized Return S preads across Stocks Sorted by Synchronicity
We break down firms into three groups based on the R2 from the market model regression: ri ,t t t rm ,t i ,t , where ri,t are weekly stock returns and rm,t
are weekly market returns in each year (July of year t-1 to June of year t) during the period of 1994 to 2005. This table reports the time -series average of
standardized stock return spreads between D10 and D1 portfo lios sorted by return predictors for the top, middle, and bottom R2 groups in the Chinese market.
The standardized return spread (D10-D1) of a return p redictor is the stock return spreads scaled by the predictor spreads in a corresponding year. Decile
portfolios are formed in the end of June and held unchanged from July of year t to June of year t+1 for each stock decile. To be included in the analysis, stock
price is restricted to be no less than ¥1 for China market in June of each year. The accounting data is fro m fiscal year 1994 to 2005 and stock return data is fro m
July 1995 to June 2007. The t-statistics are included.
-SIZE B/P MOM E/P C/P -SG -ACC -NOA -CPX RD ADV -AG ∆GPM -∆EQ -∆DT -ST DR -T URN ILLIQ
R Rank (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
High 0.34 -0.05 0.54 -0.57 -0.47 0.06 0.21 -0.27 1.29 1.29 -0.05 0.71 0.76 -1.12 -0.52 -0.22 -0.82 0.55
t-stat 1.42 -0.38 1.03 -0.70 -0.91 0.28 0.35 -0.49 1.56 1.78 -0.07 1.46 0.95 -0.90 -0.65 -0.91 -2.29 0.99
M id 0.52 0.26 0.20 0.56 0.17 0.05 0.70 1.20 0.74 1.79 0.31 0.75 0.69 0.50 0.13 20.46 0.46 0.68
t-stat 0.74 1.97 0.75 0.81 0.24 0.54 0.50 1.58 0.92 1.86 0.25 0.30 0.56 0.42 0.23 0.95 1.26 0.71
Low 0.83 0.44 0.12 1.25 0.53 -0.18 0.89 1.60 0.63 2.33 0.91 0.91 0.62 1.00 0.61 0.65 0.90 0.82
t-stat 0.88 3.17 0.26 1.14 0.72 -1.12 0.24 2.16 0.40 2.36 0.56 0.05 0.55 0.45 0.67 1.38 1.69 1.16
High-Low -0.49 -0.49 0.41 -1.81 -1.00 0.23 -0.68 -1.87 0.66 -1.04 -0.96 -0.21 0.14 -2.11 -1.13 -0.87 -1.72 -0.27
t-stat -0.50 -3.35 1.10 -2.90 -2.64 1.30 -1.73 -2.87 0.47 -1.94 -1.62 -0.46 0.12 -1.94 -1.95 -1.90 -3.09 -1.65