Emerging Markets Finance and Trade

ISSN: 1540-496X (Print) 1558-0938 (Online) Journal homepage: http://www.tandfonline.com/loi/mree20

Investor Sentiment, Financial Report Quality

and Stock Price Crash Risk: Role of Short-Sales
Constraints

Yugang Yin & Rongfu Tian

To cite this article: Yugang Yin & Rongfu Tian (2015): Investor Sentiment, Financial Report
Quality and Stock Price Crash Risk: Role of Short-Sales Constraints, Emerging Markets Finance
and Trade, DOI: 10.1080/1540496X.2015.1093844

To link to this article: http://dx.doi.org/10.1080/1540496X.2015.1093844

Published online: 03 Nov 2015.

Submit your article to this journal

Article views: 37

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=mree20

Download by: [University of Nebraska, Lincoln] Date: 29 November 2015, At: 23:06
 Emerging Markets Finance & Trade, 1–18, 2015
Copyright © Taylor & Francis Group, LLC
ISSN: 1540-496X print/1558-0938 online
DOI: 10.1080/1540496X.2015.1093844

Investor Sentiment, Financial Report Quality and Stock

Price Crash Risk: Role of Short-Sales Constraints
Yugang Yin1 and Rongfu Tian2
1
School of Securities and Futures, Southwestern University of Finance and Economics, Chengdu,
Sichuan, China; 2Department of International Trade, Ningbo Dahongying University, Ningbo,
Zhejiang, China

ABSTRACT: We use firm-year observations of Chinese firms between 2003 and 2013 and empirically
investigate the association between investor sentiment and stock crash risk with respect to short-sales
constraint conditions. In addition, we also evaluate the incremental effect of financial reporting quality on
this association and the existence of such an association under market conditions. We find that investor
sentiment is positively associated with future stock price crash risk and poorer financial report quality and
short-sale constraint will strengthen this association. In consideration of the firm-level fundamental
information in stock prices and different market states, we find that lower fundamental information in
stock price and bull market state will strengthen the positive association between investor sentiment and
Emerging Markets Finance and Trade

future stock price crash risk. Our findings are robust with several robustness checks.
KEY WORDS: crash risk, financial report quality, investor sentiment, short-sales constraints

The asymmetry of volatility in stock returns (large price movements are more likely to be crashes
rather than increases) not only may cost the investors their wealth, but also does harm to the stability
and development of capital markets. Its striking characters are that stock markets crash in the absence
of negative information and contagiously across the entire class of stocks (Hong and Stein 2003).
Investigations on stock price crashes rise in the 1980s. The initial analyses are developed on the
framework of rational models with complete information aggregation, and the workhorse is volatility
feedback models, which explain the asymmetric changes in stock prices (Campbell and Hentschel
1992). But these models cannot fit the distinct facts of the crashes such as the stock prices dropping
systematically without pieces of information. Then, further studies focus on three aspects: the asym-
metric information, market fraction, and differences of investors’ opinions, indicated by the rational
models with incomplete information aggregation and behavioral finance. Rational models with
incomplete information aggregation implies that crash risk is the result of intensive release of private
information by insiders (Romer 1993; Cao, Coval, and Hirshlerfer 2002) and the push work by the
outsiders (Barlevy and Veronesi 2003). Marin and Olivier (2008) demonstrate that inside trade volume,
which is lagged from two months to twelve months, has a significantly positive relation with stock
crash risk, which is consistent with the conclusions of Romer (1993) and Cao, Coval, and Hirshlerfer
(2002), they show that inside traders release large amounts of news at one time, which increases the
crash risk. Grech and Mazur (2004) document that information efficiency and the degree of asymmetry
affects the crash risk. They show that lower information efficiency means a higher crash risk. Hutton,
Marcus, and Tehranian (2009) uses accruals to measure the transparency of financial statement and
find that the transparency is significantly negative with crash risk. Atilgan and Demirtas (2013)
investigate the relation between downside risk and expected returns on the aggregate stock market
in an international context and find that, for emerging markets, fixed effects panel data regressions
provide evidence for a significantly positive relationship between monthly expected market returns and
downside risk.

Address correspondence to Rongfu Tian, 899 Xueyuan Road, Yinzhou, Ningbo, Zhenjiang Province, China
315175. E-mail: tianrongfu@yahoo.co.jp
 2 Y. YIN AND R. TIAN

Behavioral finance explains the crash risk from two aspects: investor sentiment and differences of
investors’ opinions. Case and Shiller (1989) compares the sentiment of investors and economic
fundamentals before the crash in 1987 with those after the stock crash. His finding is that the factor
accounting for the crash is investor sentiment rather than the economic fundamentals. The economic
fundamental did not change and the sentiment dropped sharply, which was overoptimistic before the
stock crash and became pessimistic after the crash. Hong and Stein (2003) develop a new model to
grip with the characteristics, which were not explained by volatility feedback models. The new model
has two assumptions, differences of investors’ opinions due to overconfidence and short-sales con-
straints. Griffin, Nardari, and Stulz (2007) also supports their theory. They declare six months lagged
turnovers have significant positive coefficients of the regression with the negative skewness of stock
returns. Brooks and Katsaris (2005) divide the market states into three states: slowing up, skyrocket-
ing, and bubble burst, and then build a three state model of transformation. They find that excess trade
gets a significantly positive relation with the bubble burst risk.
This article develops the theory on the framework of behavioral finance and put another two
factors, information efficiency and short-sales constraint of rational models with incomplete
information aggregation, into the analysis. We study how the differences of sentiment caused by
information efficiency constraints affect the potential of a stock crash in China. We conclude that
the core variable of the behavioral finance is investor sentiment, which comes from investors’
expectations of asset prices. This article treats the sentiment as an endogenous variable and
Emerging Markets Finance and Trade

explores a new transmission mechanism in which the information efficiency affects the expecta-
tions and then influences the sentiment and stock price. The information efficiency can be
interpreted as the quality of financial reports or the fundamental information of companies in
stock price. According to Hong and Stein (2003), with short-sale constraint, the stock price reveals
the expectations of optimistic investors rather than pessimistic investors. Thus the bearish private
information gets hidden and accumulates. When the markets go down, such bearish information
pours into the market and raises the crash risk. Without short-sale constraint, investors would make
security financing trades to release the bearish information, and then different types of investors
learn about the negative information. Stock prices go down slowly, predicting the lower crash risk.
So the presence of constraints would affect the actual influences on the stock price from sentiment.
This article develops a theoretical model to demonstrate the negative relation between sentiment
and stock return with the short-sale constraint, and then in the empirical analysis we set the crash
risk variables following Chen, Hong, and Stein (2001) and Kim et al. (2011a, 2011b). As the
theoretical model implied, investor sentiment is positively associated with future stock price crash
risk and poorer financial report quality and short-sale constraint will strengthen this association.
Lower fundamental information in stock price and bull market state will also strengthen the
positive association between investor sentiment and future stock price crash risk.
The contribution of this article is our interact combination of rational models with incomplete
information aggregation and behavioral finance. We explore a new transmission mechanism, indicat-
ing the inherent relation between incomplete information and investor sentiment in explaining the
stock crash.

Model Design and Hypotheses

Delong et al. (1990) present a model to study how the noise traders affect the stock return in the
market, which contains rational investors and noise investors, and demonstrate that the wrong beliefs
of noise traders actually affect the stock return. The idea gets supported by the following empirical
work. The assumptions in our model are close to those in the model of Delong et al. (1990).
1. The market includes two different assets with the same dividends r. One is safe assets, and the
other is risk asset u. The price of s is standardized as 1. The price of risk asset is supposed to be
pt, and the total supply of risk asset is 1.
 ROLE OF SHORT-SALES CONSTRAINTS 3

2. The market contains two kinds of investors, noise traders, and rational investors. Noise traders
trade according to their expectations. At the time t, the misjudge on the price is quantized as the
deviation to the expected price
of asset u in the future ρt . It is an independent and distributed
normal random variable,ρt ～N ρ
t ; σ 2ρt . ρ
t is the mean of expected price of risk asset, which
imply the degree of the sentiment of noisy traders. The rational investors have sophisticated
expectations about the dividends and distribution of risky assets. Their demand of risk assets is
dependent on the arbitrage opportunity between the spot price and the inherent value. Rational
investors exploit the noise traders’ mispricing by taking the adverse strategies. The proportion of
noise traders is β, and the rational investors are weighted as 1–β. Over the life, the two types of
investors have two life periods: young age t and old age t + 1. They choose the portfolios when
they are young, and would sell the risk assets at the price pt+1 and then consume the wealth.
3. The return of risk assets is distributed as normal distribution and the agents’ utility function is U
(w), where w is the value of traders’ wealth. The traders maximize the expected
utility,EðU Þ ¼ w  γσ 2w , where w is expected final wealth, γ is the coefficient of absolute risk
aversion, andσ 2w is the variance of wealth.
The expected utility of rational traders is:
Emerging Markets Finance and Trade

EðUa Þ ¼ wa  γσ 2wa
 2 h
2 i (1)
¼ sðLa Þ þ λat ½r þ Et ðptþ1 Þ  pt ð1 þ rÞ  γ λat Et σ ptþ1 ;

where sðLa Þ is the function of labor income of rational traders, σ 2ptþ1 ¼½ptþ1  Et ðptþ1 Þ2 is the variance
of pt+1 of risk assets at time t, and λat represent the amounts of risky assets held by rational traders.
The expected utility of noise traders is:

EðUn Þ ¼ wn  γσ 2wn
 2 h
2 i (2)
¼ sðLn Þ þ λnt ½r þ Et ðptþ1 Þ  pt ð1 þ rÞ  γ λnt Et σ ptþ1 þ λnt ðρt Þ;

where sðLn Þ is the function of labor income of noise traders, λnt represent the amounts of risky assets
held by noise traders, and ρt is the deviation to expected price of u at time of t and distributed as a
normal random variable.
The total supply of risk assets is 1:

ð1  βÞλat þ βλnt ¼ 1 (3)

The equilibrium price of risk assets comes from the maximization of the utilities of rational and noise
traders:

 2 h
2 i
max : sðLa Þ þ λat ½r þ Et ðptþ1 Þ  pt ð1 þ rÞ  γ λat Et σ ptþ1

 2 h
2 i
max : sðLn Þ þ λnt ½r þ Et ðptþ1 Þ  pt ð1 þ rÞ  γ λnt Et σ ptþ1 þ λnt ðρt Þ

s:t: : ð1  βÞλat þ βλnt ¼ 1 (4)

Maximize the Equation (4):

 4 Y. YIN AND R. TIAN

r þ Et ðptþ1 Þ  ð1 þ rÞpt
λat ¼ h
i (5)
2γ Et σ 2ptþ1

r þ Et ðptþ1 Þ  ð1 þ rÞpt ρ
λnt ¼ h
i þ h
t i (6)
2γ Et σ 2ptþ1 2γ Et σ 2ptþ1

Put Equation (5) and (6) into Equation (4):

1 h
 i
pt ¼ r þ Et ðptþ1 Þ  2γEt σ 2ptþ1 þ βρt (7)
1þr

Equation (7) calculated recursively:

 
β ρt  ρ
t βρ
2γ h
2 i
pt ¼ 1 þ þ t  Et σ ptþ1 (8)
1þr r r

The volatility of stock price is independent on the noise traders’ expectations of the volatility of stock
Emerging Markets Finance and Trade

price:

 β2 σ 2ρt
Et σ 2ptþ1 ¼ σ 2ptþ1 ¼ (9)
ð1 þ r Þ2

We can get:
 
β ρt  ρ
t βρ
2γβ2 σ 2ρt
pt ¼ 1 þ þ t  (10)
1þr r rð1 þ rÞ2

The return of risk assets:

2 þ r
βρtþ1 þ ðβ þ rÞpt
returntþ1 ¼ r þ ptþ1  pt ð1 þ rÞ ¼  βρ þ (11)
1þr t 1þr

The expected return of risk assets:

2 þ r
ðβ þ rÞpt
Et ðreturntþ1 Þ ¼  βρ þ (12)
1þr t 1þr

Then:

2þr
@Et ðreturntþ1 Þ=@ρ
t ¼  β<0 (13)
1þr

Equation (13) imply that a higher traders’ sentiment ρ
t interpret a lower return of risk assets.
Compared to the capital markets of developed countries, the proportion of institution investors is
smaller and individual investors are crucial for the formation of stock prices. If the investors’ sentiment
is higher that means they hold more optimistic attitudes to the stock return and they act over-
confidently in the trades. There are significant contrarian effects in the Chinese stock market. Liu
and Pi (2007) find the significant contrarian effect and reject the momentum based on full sample data
from 1994–2005 in China Shanghai and Shenzhen A share market. They conclude that risk accounts
 ROLE OF SHORT-SALES CONSTRAINTS 5

partly for such a phenomenon. Li (2013) combines the model of Rodriguez and Sbuelz (2006) with
robustness and predicts that the investors with robustness partly carry out contrarian strategies. In our
model, rational investors exploit the mispricing of noise investors and carry out contrarian strategies.
Based on the annual data, we find if the investors’ sentiment is higher at present, the stock price have
the momentum effect in short time. But in a longer period, the stock tends to have the contrarian effect
because of the rational investors’ carrying out adverse strategies. In a word, a higher the sentiment now
implies a lower the stock return.
Chen and Zhang (2009) investigate the micro-mechanism of the crash of Chinese stock market with
short-sale constraint based on the Hong and Stein (2003) heterogeneous beliefs model. They find the
absence of short-sale mechanism raises the crash risk because of differences of investors’ opinions and
predict that the security financing would reduce the potential of stock crash. According to Hong and
Stein (2003), stock price only reveals the expectations of optimistic investors rather than pessimistic
investors. Then the bearish information gets hidden and accumulates, which raises the crash risk when
the market goes down. With the short-sale constraint, investors would make security financing trades
to release the bearish information, and then different types of investors learn to know the bearish
information and stock price goes down slowly, reducing the crash risk. Therefore we develop these
hypotheses:
H1: With the short-sale constrain, a higher investors’ sentiment indicates a larger crash risk. If security
Emerging Markets Finance and Trade

financing is available, there is no significant relationship between sentiment and crash risk.
The listed companies produce information, and have the tendency to release information selectively
to meet their own interests. To maintain the stock price, the managers tend to hide bad news. The bad
news accumulates until the board cannot be burdened. Then large amounts of bad news pour into the
market and the company stock crashes (Jin and Myers 2006). Hutton, Marcus, and Tehranian (2009)
uses earning management to represent the transparency of financial statements, and they find the
opaque of financial reports has a significantly negative relationship with crash risk. We can get that the
higher the investors’ sentiment, the bigger crash risk in the future. Even if the accumulated bad news
gets released at present, the crash risk still becomes higher because of the short-sale constraint.
H2: With the short-sale constraint, the transparency would strengthen the positive relationship between the
investor sentiment and the potential of stock crash.

Research Design
Dependent Variable: Stock Price Crash Risk NCSKEW and DUVOL
In literature, there are two kinds of measures to describe crash risk. One is negative coefficient of
skewness (Chen, Hong, and Stein 2001; Kim, Li, and Zhang 2011a, 2011b). The other is binary
variable CRASH, which represents whether the crash happens or not. Given the stock return follows
normal distribution, r～(u, δ), if r ≤ u-2δ, CRASH equals 1. Otherwise, CRASH equals 0 (Marin and
Olivier 2008; Hutton, Marcus, and Tehranian 2009). We follow the first variable measure (Chen,
Hong, and Stein 2001; Kim et al. 2011a, 2011b) to interpret the potential of crash risk and develop the
CRASH measure in robustness test.
Regress the Equation (14) to get residual series εi;t and specific weekly return of stock i at time t is
Wit ¼ lnð1 þ εit Þ .

ri;t ¼ αi þ β1;i  rm;t2 þ β2;i  rm;t1 þ β3;i  rm;t þ β4;i  rmþ1;t þ β5;i  rmþ2;t þ εi;t (14)

where ri;t and rm;t are correspondingly the weekly return of company i and market m. The lead and lag
items of the market are employed in Equation (14) to reduce the deviation because of no synchronous
trading (Dimson 1979).
 6 Y. YIN AND R. TIAN

Then we give the calculation of NCSKE and DUVOL.

NCSKEW comes from Equation (15):

X  
X 3=2 
NCSKEWi;t ¼  nðn  1Þ3=2 Wi;t3 = ðn  1Þðn  2Þ Wi;t2 (15)

where n is the number of stock i’s trading weeks. If NCSKEW is bigger, the stock has a more negative
skewness and a larger crash risk.
The way we get DUVOL is as follows. We divide the companies into two categories: one contains
the companies whose returns are below the period mean, “down weeks” and the other one contains the
companies whose returns are above the period mean, “up weeks.”
Then we calculate the standard deviation respectively, and DUVOL is a specific return standard
deviation of down weeks divided by the one of up weeks:
(" # " #)
X X
DUVOLi;t ¼ log ð n u  1Þ Wi;t2 = ðnd  1Þ Wi;t2 (16)
DOWN UP

where nu (nd ) is the number of up(down) weeks of stock i. That DUVOL is bigger means the crash risk
Emerging Markets Finance and Trade

become bigger.

Independent Variables
Investor Sentiment
Brown and Cliff (2004) show that the definition of sentiment can be interpreted as the degree of
optimism or pessimism. But it is difficult to give an accurate measure and different literature chooses
the specific proxy variable to represent the investor sentiment, such as turnovers (Jones and Owen
2002), IPO return (Ritter and Welch 2002), and discount rate of closed-end funds (Lee, Shleifer, and
Thaler 1991). Baker and Wurgler (2006) verify the prediction accuracy of turnovers, IPO initial return,
and discount rates of closed-end funds (CEFD) to revenue. Wu and Han (2007) uses the growth rate of
investors’ opening accounts, Open, as the proxy variable. The number of accounts in the exchanges
reveal the investors’ demand of securities. When the sentiment of outside investors becomes higher,
they are more likely to step into the stock market, and vice versa. Because of scarcity of the IPO
opportunities, there is always a high IPO initial return even if the market is not bullish. So IPO initial
return is not a proper measure of investor sentiment. Another proxy variable, Ratio, means the number
of listed company’ going up divided by the one of listed company’ melting down in correspondingly
statistics interval. This article employs turnover, CEFD, Open and Ratio as the proxy variables to
represent the sentiment.
We employ the variables to create a comprehensive sentiment index using the factor analysis
method:

Sentimentt ¼ 0:796
Ratiot þ 0:6735
CEFDt þ 0:440
Turnovert þ 0:1842
Opent : (17)

Transparency of Financial Statements: ACCM

Companies are prone to adjust the accruals to control the profits, which strengthen the asymmetry
between the investors and companies. We follow Hutton (2009).
Regress Equation (18) as the adjusted Jones model (Dechow, Sloan, and Sweeney 1995) implies.
 ROLE OF SHORT-SALES CONSTRAINTS 7

TAi;t 1 ΔSalesi;t PPEi;t

¼ α0 þ β1 þ β2 þ εi;t (18)
Asseti;t1 Asseti;t1 Asseti;t1 Asseti;t1

where, Asseti,t-1 is the total asset of company at time t–1, TAi,t is the total Accruals at time t, equals
(changes in liquid assets – changes in cash and equivalent assets) – changes in total liquid debt –
changes in short-term borrowing – changes in income tax payable) – depreciation and amortization
expense in the corresponding interval, ΔSalesi,t is the change in sales at time t and PPEi,t is the
investment in fixed assets at time t.
Put the coefficient of the regression of Equation (18) in Equation (19) to compute the transparency
of financial statements ACCM:
 
TAi;t 1 ΔSalesi;t  ΔReceivablesi;t ^ PPEi;t
ACCMi;t ¼  α^0 þ β^1 þ β2 (19)
Asseti;t1 Asseti;t1 Asseti;t1 Asseti;t1

^0 ; β^1 ; and β^2 are the coefficients of regression of Equation (19), ΔReceivablesi,t is changes in
where α
accounts receivable at time t. A bigger ACCM implies the worse of the transparency of financial
statements.
Emerging Markets Finance and Trade

State of MARKET: MARKET

Following Lindahl-Stevens (1980) and Xu et al. (2012), we define the period where the market return
exceeds the riskless return as bullish market, otherwise as bearish market. We use the fixed-deposit
rate as the proxy variable to describe the riskless return. If the fixed-deposit rate changes in a year, we
define the mean of different rates as the annual rate. In this way, we compute the risk premium, 2003
(3.74 percent), 2004 (−17.65 percent), 2005 (−10.58 percent), 2006 (127.91 percent), 2007 (93.45),
2008 (−69.32 percent), 2009 (77.73 percent), 2010 (−16.62 percent), 2011 (−21.2087 percent), 2012
(3.2108 percent), and 2013 (−10 percent). As the rule above indicates, we define 2003, 2006, 2007,
2009, and 2012 as a bullish market, and the rest as a bearish market. If the market is bullish,
MARKET = 1. Otherwise, MARKET = 0.

Model
We use Equation (20) to test hypothesis 1:

X
m
CRASHt ¼ α0 þ α1 SENTIMENTt1 þ αq ðControlVariablest1 Þ þ εt (20)
q¼2

where CRASHt including NCSKEWt and DUVOLt represents the crash risk, and SENTIMENTt-1
describes the investor sentiment.
Hutton, Marcus, and Tehranian (2009) predicts the leverage and company operation significantly
affect the stock returns. Kim et al. (2011b) finds that over the life, the growth-type enterprise tends to
be vulnerable to crash risk and the size of a company is positively associated with the potential of
crash.
Based on their findings, this article employs leverage LEVt-1, book to market ratio MBt-1, rate of
return on net assets ROAt-1, scale of companies SIZEt-1, nature of companies NATUREt-1, the
proportion of institutional investors’ holding Fund_holdt-1, the proportion of senior managers’ holding
Executive_holdt-1, ownership concentration Ownershipt-1, Tobin’s q Qt-1, and industry labels Industry
Dummy as control variables.
We use the Equation (21) to verify how the quality of financial statements affects the relation
between sentiment and crash risk.
 8 Y. YIN AND R. TIAN

CRASHt ¼ β0 þ β1 ACCMt1 þβ2 ACCMt1
SENTIMENTt1 þβ3 ACCMt1

X m
(21)
þ αq ðControlVariablet1 Þ þ εt
q¼3

where ACCM t-1 is the quality of financial statements and other variables are set as those in Equation (20).

Empirical Results
Sample and Descriptive Statistics
The data of listed companies from 2003–13 in China Shanghai and Shenzhen A share market are
obtained from the databases CASMAR and WIND, the quality data of company information dis-
closure in robustness tests comes from the regulatory disclosure offered by Shenzhen Stock Exchange;
and the data of newly increased openings in stock accounts every year comes from China Securities
Depository and Clearing Corporation Limited. We follow the procedure developed by Jin and Myers
(2006) to set the variable to capture the crash risk in the absence of financial companies and the
companies whose trade weeks is less than thirty. The remaining sample includes 9,179 firms.
Table 1 presents descriptive statistics of the variables included in this article. Because the corre-
Emerging Markets Finance and Trade

sponding standard deviations are 0.837 and 0.410, the two measures of crash risk have great
differences, indicating that we can measure the crash risk in two different views. The mean of variable
SYNCH is 0.454, implying that the stock prices go up and down simultaneously across the stock
markets in China and the private information is not quite revealed in stock prices. In the trade
institution, the number of companies can be involved in security financing is 713, accounting for
7.77 percent of the available sample. The number of bull years is five and the bear years is six.

Investor Sentiment and Crash Risk

The regression results of crash risk on investor sentiment are shown in Table 2, where NCSKEWt and
DUVOLt are two distinct measures. The columns (1) and (3) show that the coefficients of the investor
sentiment SENTIMENTt-1 are correspondingly 0.0012 and 0.0004, which are statistically significant at
the 1 percent level. The results above indicate that a higher investor sentiment at time t–1 is associated
with a higher crash risk at time t. To eliminate the potential influence caused by the nature of the
company, we employ a new variable to interpret the nature of the company in the regression, which
improves the goodness of fit significantly. The coefficients of the investor sentiment SENTIMENTt-1
are 0.0027 and 0.0008 respectively, which are statistically significant at the 1 percent level. This result
shows that the positive relation between investor sentiment and crash risk is not affected by the nature
of the company. The aforementioned results demonstrate hypothesis 1.

Investor Sentiment and Crash Risk: Financial Report Quality and Short-Sale Constraints
The investor expectations toward stock price and investment decision are affected by the external
environment. This article goes on further studies in the presence of the exotic conditions including the
financial report quality, the private information involved in the stock price, the trade institution, and the
market state. We investigate how the exotic conditions affect the relation between the investor sentiment
and the crash risk. Table 3 presents the following empirical results. (1) The cross terms ACCMt-
1*SENTIMENTt-1 of the financial report quality and the investor sentiment are 0.0008 and 0.0002,
respectively. The coefficients are statistically significant at the 5 percent level, which indicates the more
asymmetric the information would strengthen the positive relation between investor sentiment and crash
risk. The mechanism is that the information raises the investors’ blindness to stock price, and then blind
optimism leads to a higher crash risk. (2) The cross terms of the private information about the company
 ROLE OF SHORT-SALES CONSTRAINTS 9

Table 1. Descriptive statistics

Variable N Mean Std 5% 25% Median 75% 95%

Dependent Variable
Crash Risk
NCSKEWt 9179 −0.181 0.837 −1.842 −0.458 −0.044 0.088 1.288
DUVOLt 9179 −0.106 0.410 −0.770 −0.369 −0.108 0.155 0.568
CRASHt 9179 0.366 0.482 0 0 0 1 1
Independent Variable
Investor Sentiment
SENTIMENTt-1 9179 25.127 29.891 −7.732 2.334 17.011 39.586 85.110
Report Quality
ACCMt-1 9179 0.091 1.122 −0.575 −0.105 0.074 0.281 0.921
INFOt-1 3071 2.186 0.654 1 2 2 3 3
Synchronicity
SYNCH t-1 9179 0.454 0.680 −0.537 0.028 0.403 0.821 0.924
IDIOSYNCRATIC t-1 9179 1.430 0.428 0.796 1.125 1.395 1.702 2.178
Trade Institution
CONSTRAINT = 1 713
Emerging Markets Finance and Trade

CONSTRAINT = 0 8466
Market State
BULL MARKET 5
BEAR MARKET 6
Control Variable
LEVt-1 9179 0.078 0.111 0 0.001 0.030 0.114 0.310
MBt-1 9179 0.623 0.268 0.205 0.418 0.610 0.813 1.079
RETt-1 9179 0.002 0.016 −0.014 −0.006 −0.001 0.007 0.025
ROAt-1 9179 0.048 0.251 −0.047 0.030 0.053 0.082 0.154
SIGMAt-1 9179 0.061 0.055 0.033 0.044 0.055 0.070 0.100
SIZEt-1 9179 14.310 1.2515 12.451 13.391 14.229 15.087 16.523
NATUREt-1 9179 0.443 0.497 0 0 0 1 1
HOLDt-1 9179 3.834 7.190 0 0 0.429 4.276 19.698

and the investor sentiment have coefficients of 0.0007 and 0.0002, which are positive significantly on the
1 percent and 5 percent level. Such results can be interpreted as the less the amount of private
information revealed in stock price, the more significant the relationship between investor sentiment
and crash risk. This finding is in accord with the aforementioned finding, which indicates that the more
the amount of private information revealed in stock price, the less financial report quality is involved in
stock prices. (3) The coefficients of the cross terms of short-sale constraints and investor sentiment are
−0.0026 and −0.0007 respectively, which are both significant at the 5 percent level. Such results reveal
that the security financing reduces the positive relationship between the investor sentiment and the crash
risk with a high significance, which is consistent with the study carried out by Hong and Stein (2003).
For example, in the absence of short-sale constraints, the bear information from some investors comes
into the stock price readily, and other investors recognize such information and then adopt
their expectations about the stock returns, which reduces the crash risk. (4) The cross terms
MARKET t-1*SENTIMENTt-1 of the market state and the investor sentiment have coefficients of
−0.0007 and −0.0014, which are significantly negative on the 5 percent level. We conclude from the
external risk that the market is a bull market, which would strengthen the positive relationship between the
investor sentiment and the crash risk. In bull markets, the investors share more optimistic sentiment (Xu,
Yu, and Yi 2013), which induces a higher crash risk. Such conclusion is in keeping with the significant
contrarian effect in medium- and long-term in China stock markets found by (Liu and Pi 2007).
 10 Y. YIN AND R. TIAN

Table 2. Investor sentiment and crash risk

NCSKEWt DUVOLt

(1) (2) (3) (4)

SENTIMENTt-1 0.0012*** 0.0027*** 0.0004*** 0.0008***

(4.41) (7.19) (2.93) (4.72)
LEVt-1 −0.1214 −0.0281
(−1.35) (−0.68)
MBt-1 0.2520*** 0.1994***
(6.28) (10.87)
RETt-1 −8.7244*** −3.5766***
(−10.41) (−9.33)
ROAt-1 −0.0483 0.0234
(−1.31) (1.39)
SIGMAt-1 2.0729*** 0.7853***
(9.62) (7.97)
SIZEt-1 0.0005 0.0367***
(0.06) (9.15)
NATUREt-1 0.0006 −0.0336***
Emerging Markets Finance and Trade

(0.03) (−3.89)
HOLDt-1 −0.0007 −0.0040***
(−0.49) (−6.38)
_CONS 0.2854*** 0.0158 0.0743** −0.5778***
(4.31) (0.11) (2.29) (−8.58)
IND_DUM YES YES YES YES
OBS 9179 9179 9179 9179
ADJ_R2 0.0043 0.0249 0.0063 0.0448

Notes: Here *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels respectively.

Robustness Tests
Based on four aspects of the information disclosure of the companies, authenticity, completeness,
legitimacy and equality, Shenzhen stock exchange ranks and divides the companies into four camps,
A, B, C, and D, in order from 2001. A represents the highest quality of information disclosure. The
higher quality of a company’s disclosure indicates the lesser financial report quality faced by
investors. This article converts the quality of disclosure to numerical scores as follows: A takes
on a value of 4; B 3; C 2; and D 1. The bigger number is associated with a higher financial report
quality. We use variable INFO as financial report quality. The idiosyncratic volatility is stock
volatility minus market volatility, which is attributed to private information in stock price. Ang
et al. (2006) finds the idiosyncratic volatility has a negative relationship with the stock return. This
article employs the idiosyncratic volatility as proxy variable to capture the private information
contained in stock prices.
We calculate the idiosyncratic volatility as follows:
Regress the Fama-French three factors model:

Ri;t  RM ;t ¼ αi;t þ βi;t ðRM ;t  Rf ;t Þ þ si;t SMBi;t þ hi;t HMLi;t þ εi;t (24)

where, Ri,t is the weekly return of stocks, Rm,t is the weekly return of stocks, Rf,t is the riskless return,
SMBi,t is the sale, and HMLi,t is the book-to-market ratio.
 Emerging Markets Finance and Trade

Table 3. Investor sentiment and crash risk

NCSKEWt DUVOLt

(1) (2) (3) (4) (5) (6) (7) (8)

SENTIMENTt-1 0.0026*** 0.0044*** 0.0025*** 0.0022*** 0.0008*** 0.0013*** 0.0007*** 0.0008***

(6.90) (7.97) (6.57) (5.59) (4.55) (5.09) (4.35) (4.48)
ACCMt-1*SENTIMENTt-1 0.0008** 0.0002**
(2.42) (2.23)
ACCMt-1 −0.0004 0.0045
(−0.03) (0.80)
SYNCHt-1*SENTIMENTt-1 0.0007*** 0.0002**
(3.31) (2.36)
SYNCHt-1 0.0585*** 0.0123**
(6.52) (2.98)
CONSTRAINTt-1*SENTIMENTt-1 −0.0026** −0.0007**
(−2.32) (−2.79)
CONSTRAINTt-1 −0.1366** −0.0490*
(−2.48) (−1.94)
MARKET t-1*SENTIMENTt-1 −0.0007** −0.0014**
(−2.73) (−2.92)
MARKET t-1 0.3095*** 0.1141***
(9.65) (7.75)
LEVt-1 −0.1224 −0.1454 −0.1249 −0.0770 −0.0282 −0.0316 −0.0279 −0.0154
(−1.36) (−1.62) (−1.39) (−0.86) (−0.69) (−0.77) (−0.68) (−0.38)
MBt-1 0.2521*** 0.2953*** 0.2619*** 0.1331*** 0.1997*** 0.2060*** 0.2044*** 0.1624***
(6.28) (7.26) (6.50) (3.23) (10.88) (11.05) (11.09) (8.59)
RETt-1 −8.7029*** −9.2570*** −8.7897*** −6.3500*** −3.5623*** −3.7033*** −3.6514*** −3.0772***
(−10.39) (−11.01) (−10.48) (−7.20) (−9.30) (−9.61) (−9.52) (−7.60)
ROAt-1 −0.0491 −0.0378 −0.0508 −0.0423 0.0230 0.0247 0.0222 0.0249
(−1.33) (−1.03) (−1.38) (−1.16) (1.36) (1.47) (1.32) (1.49)
SIGMAt-1 2.0641*** 2.1741*** 2.0798*** 1.3987*** 0.7810*** 0.8067*** 0.7956*** 0.6274***
(9.58) (10.08) (9.65) (6.17) (7.92) (8.16) (8.07) (6.03)
SIZEt-1 0.0008 0.0089 0.0113 0.0347*** 0.0368*** 0.0381*** 0.0431*** 0.0470***
(0.09) (1.00) (1.15) (3.76) (9.18) (9.40) (9.59) (11.12)

(Continued )

11
 Emerging Markets Finance and Trade

12
Table 3. Investor sentiment and crash risk (Continued)
NCSKEWt DUVOLt

(1) (2) (3) (4) (5) (6) (7) (8)

NATUREt-1 0.0009 0.0096 −0.0013 −0.0212 −0.0334*** −0.0326*** −0.0336*** −0.0397***

(0.05) (0.51) (−0.07) (−1.12) (−3.87) (−3.75) (−3.88) (−4.58)
HOLDt-1 −0.0008 −0.0013 −0.0008 −0.0047*** −0.0041*** −0.0042*** −0.0041*** −0.0052***
(−0.56) (−0.93) (−0.56) (−3.29) (−6.45) (−6.58) (−6.48) (−8.07)
_CONS 0.01846 −0.2748* −0.1302 −0.3935** −0.5777*** −0.6287*** −0.6647*** −0.7013***
(0.13) (−1.79) (−0.82) (−2.61) (−8.58) (−8.93) (−9.15) (−10.15)
IND_DUM YES YES YES YES YES YES YES YES
OBS 9179 9179 9179 9179 9179 9179 9179 9179
ADJ_R2 0.0261 0.0293 0.0254 0.0382 0.0455 0.0421 0.0459 0.0512

Notes: Here *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels respectively.
 ROLE OF SHORT-SALES CONSTRAINTS 13

Then we get the idiosyncratic volatility:

^i;t þ β^i;t ðRM ;t  Rf ;t Þ þ ^si;t SMBi;t þ ^hi;t HMLi;t

IDIOSYNCRATICi;t ¼ ðRi;t  RM ;t Þ  α (25)

where α ^i;t , β^i;t , ^si;t and ^

hi;t are the estimates of coefficients in Equation (24).
The sample included in security financing in Table 4 is regressed in the general regression.
And in China the security financing has been carried out since 2009. To unify the sample
interval, we study the sample in and since 2009 to find out how the security financing affects the
relationship between investor sentiment and crash risk. In particular, some stocks exit from the
security financing sample, which offers us an opportunity to compare the relationship between
investor sentiment and crash risk of the same stock with and in the absence of security
financing.
One of the robustness tests shown in Table 4. INFOt-1 is the quality of listed company information
disclosure in the Shenzhen Stock Exchange; IDIOSYNCRATICt-1 is the stock idiosyncratic volatility
calculated from Equation (25); CONSTRAINT1 t-1 implies whether the stock can be shortened or not,
CONSTRAINT1 t-1 = 1 means the stock can be shortened, and CONSTRAINT1 t-1 = 0 means the stock
cannot be shortened. The variable CONSTRAINT2 t-1 is calculated from sample in and since 2009,
CONSTRAINT2 t-1 = 1 means the stock can be shortened, and CONSTRAINT2 t-1 = 0 means the stock
Emerging Markets Finance and Trade

cannot be shortened. Table 4 presents the cross terms of the financial report quality and the investor
sentiment in columns (1) and (5) have coefficients, −0.0005 and −0.0002, which are significantly
negative at the 5 percent level. Such results indicate that a higher information disclosure reduces the
positive relation between investor sentiment and crash risk. In columns (2) and (6), the cross terms of
idiosyncratic volatility and investor sentiment IDIOSYNCRATICt-1*SENTIMENTt-1 have negative
coefficients, −0.0021 and −0.0010, which are significant at the 1 percent level. We conclude that a
higher idiosyncratic volatility (the stock price reveals more private information) reduces the positive
relation between investor sentiment and crash risk. The cross terms of the absence of short-sale constraint
and investor sentiment have negative coefficients, which are significant at the 5 percent or 1 percent
level. This result indicates the absence of short-sale constraint reduces the positive relationship between
investor sentiment and crash risk significantly. The robustness tests are in keeping with the conclusions
in the main body above.
We also follow Hutton, Marcus, and Tehranian (2009) to define the crash. If the weekly return of
the stock is 3.09 times standard deviation less than the expected return, the crash happens. We choose
the figure 3.09 because the figure corresponds with the probability of 0.1 percent in the normal
distribution. If the crash happens, CRASH = 1. Otherwise, CRASH = 0. Table 5 demonstrates the
robustness of the conclusion by redefining the crash risk and employing the logit regression. The
results indicate that the relationship between the investor sentiment and the crash risk is significantly
positive. Furthermore, a higher financial report quality, the less private information revealed in stocks,
the absence of short-sale constraints, and bull markets can strengthen the positive relationship above.
The robustness tests are consistent with the conclusions in the main body above.

Conclusion
This article employs the rational models with incomplete information aggregation into the frame of
behavioral finance and study how investor sentiment affects stock return and the relationship between
investor sentiment and crash risk in different conditions. By using data of listed companies in the
China Shanghai and Shenzhen A share market, we set variables to describe investor sentiment and the
potential of crash, and find that the investor sentiment is positively related to future stock price crash
risk and poorer financial report quality and short-sale constraint will strengthen this association.
What’s more, the less private information revealed in stocks, the absence of short-sale constraints
and bull market state will strengthen such a positive association.
 Emerging Markets Finance and Trade

14
Table 4. Robustness test: Investor sentiment and crash risk
NCSKEWt DUVOLt

(1) (2) (3) (4) (5) (6) (7) (8)

SENTIMENTt-1 0.0044** 0.0053*** 0.0039*** 0.0054*** 0.0016* 0.0023*** 0.0008** 0.0014***

(2.32) (4.74) (4.10) (6.78) (1.71) (4.52) (2.49) (3.73)
INFOt-1*SENTIMENTt-1 −0.0005** −0.0002**
(−2.63) (−2.39)
INFOt-1 −0.0779** −0.0395**
(−2.46) (−2.57)
IDIOSYNCRATICt-1*SENTIMENTt-1 −0.0021*** −0.0010***
(−3.00) (−3.03)
IDIOSYNCRATICt-1 0.1294*** 0.0184
(3.95) (1.23)
CONSTRAINT1 t-1*SENTIMENTt-1 −0.0048** −0.0065**
(−2.05) (−2.63)
CONSTRAINT1 t-1 0.1486* 0.0960**
(1.92) (2.29)
CONSTRAINT2 t-1*SENTIMENTt-1 −0.0082** −0.0077***
(−2.63) (−3.21)
CONSTRAINT2 t-1 0.1540* 0.0986**
(1.86) (2.49)
LEVt-1 −0.1887 −0.1301 −0.3884* −0.2026 −0.0241 −0.0274 −0.1745 −0.0755
(−1.30) (−1.45) (−1.96) (−1.20) (−0.34) (−0.67) (−1.62) (−0.93)
MBt-1 0.2321*** 0.2801*** 0.1014 0.1934** 0.1769*** 0.1997*** 0.1271** 0.1580***
(3.59) (6.85) (1.07) (2.41) (5.65) (10.67) (2.48) (4.12)
RETt-1 −5.4103*** −7.8608*** 0.4766 −7.7864*** −3.1619*** −3.4734*** −0.3770 −4.2275***
(−4.26) (−9.08) (0.21) (−4.94) (−5.15) (−8.77) (−0.30) (−5.62)
ROAt-1 −0.2257** −0.0470 −0.7074** −0.2781 −0.0283 0.0230 −0.3111* 0.0030
(−2.32) (−1.28) (−2.28) (−0.99) (−0.60) (1.36) (−1.85) (0.02)
SIGMAt-1 1.3918*** 1.8250*** −0.0936 1.6590*** 0.6926*** 0.7578*** 0.0128 0.7991***
(4.84) (8.12) (−0.20) (4.69) (4.98) (7.37) (0.05) (4.73)
SIZEt-1 −0.0212 0.0011 −0.0382* −0.0510*** 0.0318*** 0.0361*** 0.0096 0.0206**
(−1.26) (0.13) (−1.83) (−3.19) (3.89) (8.97) (0.85) (2.69)
 Emerging Markets Finance and Trade

NATUREt-1 −0.0115 0.0022 −0.0292 −0.0026 −0.0365** −0.0326*** −0.0123 −0.0113

(−0.36) (0.12) (−0.71) (−0.07) (−2.36) (−3.77) (−0.55) (−0.66)
HOLDt-1 0.0024 −0.0019 −0.0064** 0.0004 −0.0035** −0.0041*** −0.0053*** −0.0030***
(0.91) (−1.33) (−2.69) (0.17) (−2.79) (−6.35) (−4.12) (−2.96)
_CONS 0.5733* −0.1496 1.0862*** 0.9108*** −0.4187** −0.5926*** −0.0019 −0.3074**
(1.92) (−0.98) (3.08) (3.40) (−2.90) (−8.45) (−0.01) (−2.40)
IND_DUM YES YES YES YES YES YES YES YES
FIRM EFFECT YES YES YES YES YES YES YES YES
OBS 3071 9179 1537 2488 9179 3071 1537 2488
ADJ_R2 0.0263 0.0264 0.0599 0.0453 0.0512 0.0362 0.0447 0.0481

Notes: INFOt-1 is the quality of listed company information disclosure in Shenzhen Stock Exchange, it equals 1 to 4, the bigger number means better financial report quality;
IDIOSYNCRATICt-1 is the stock idiosyncratic volatility calculated from Equation (25); CONSTRAINT1 t-1 and CONSTRAINT2 t-1 are both binary variables.
CONSTRAINT1 t-1 implies whether the stock can be shortened or not, CONSTRAINT1 t-1 = 1 means the stock can be shortened, and CONSTRAINT1 t-1 = 0 means the stock
cannot be shortened. The variable CONSTRAINT2 t-1 is calculated from a sample from 2009, CONSTRAINT2 t-1 = 1 means the stock can be shortened, and CONSTRAINT2 t-
1 = 0 means the stock cannot be shortened. Here *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively.

15
 16 Y. YIN AND R. TIAN

Table 5. Robustness test

CRASHt

(1) (2) (3) (4) (5)

SENTIMENTt-1 0.0073*** 0.0071*** 0.0175*** 0.0074*** 0.0091***

(6.66) (6.34) (10.75) (6.68) (8.38)
ACCMt-1*SENTIMENTt-1 0.0020**
(2.00)
ACCMt-1 0.0178
(0.49)
SYNCHt-1*SENTIMENTt-1 −0.0037***
(−6.54)
SYNCHt-1 0.3683***
(15.57)
CONSTRAINTt-1*SENTIMENTt-1 −0.0169**
(−2.22)
CONSTRAINTt-1 0.1107
(0.53)
MARKET t-1*SENTIMENTt-1 −0.0041**
Emerging Markets Finance and Trade

(−2.62)
MARKET t-1 0.9565***
(11.77)
LEVt-1 −0.1918 −0.1951 −0.3282 −0.1854 −0.0671
(−0.86) (−0.87) (−1.44) (−0.83) (−0.30)
MBt-1 0.4294*** 0.4287*** 0.7392*** 0.4296*** 0.0193
(4.30) (4.29) (7.18) (4.28) (0.18)
RETt-1 −11.3361*** −11.2235*** −15.0898*** −11.5073*** −7.2545***
(−5.13) (−5.07) (−6.61) (−5.20) (−3.28)
ROAt-1 −0.4018** −0.4191** −0.1214 −0.4028** −0.4104**
(−2.06) (−2.13) (−0.82) (−2.06) (−2.03)
SIGMAt-1 13.0003*** 13.0557*** 14.2847*** 12.9624*** 5.9765***
(10.33) (10.34) (11.01) (10.29) (4.90)
SIZEt-1 −0.1184*** −0.1178*** −0.0651*** −0.1155*** −0.0170
(−5.33) (−5.30) (−2.87) (−4.87) (−0.72)
NATUREt-1 −0.0529 −0.0513 0.0051 −0.0503 −0.0968**
(−1.13) (−1.09) (0.11) (−1.07) (−2.04)
HOLDt-1 0.0044 0.0041 0.0007 0.0045 −0.0063*
(1.27) (1.17) (0.20) (1.28) (−1.75)
_CONS 0.1498 0.1558 −1.8274*** 0.1089 −0.8484**
(0.40) (0.42) (−4.58) (0.28) (−2.22)
IND_DUM YES YES YES YES YES
OBS 9179 9179 9179 9179 9179
Pseudo R2 0.0348 0.0359 0.0582 0.0350 0.0487

Notes: This table presents the robustness test. We follow Hutton, Marcus, and Tehranian (2009) to define the crash. If the
weekly return of the stock is 3.09 times standard deviation less than the expected return, the crash happens. We choose
the figure 3.09 because the figure corresponds with the probability of 0.1 percent in the normal distribution. If the crash
happens, CRASH = 1. Otherwise, CRASH = 0. Here *, **, and *** indicate statistical significance at the 10 percent, 5
percent, and 1 percent levels, respectively.

Our study explores the micro-mechanism of stock crash, and demonstrates that stock crash is not
only affected by the investor sentiment indicated by behavioral finance, but also influenced by the
factors implied by rational models with incomplete information aggregation, especially the transaction
system like short-sale constraint. For one thing, our findings help the investors recognize the reliability
 ROLE OF SHORT-SALES CONSTRAINTS 17

of expectations about stock returns, especially in the bull market state, stock price crash may happen in
the short future. For another, our conclusions offer theoretical and empirical supports to policymakers
and market regulators to carry out policies to improve financial report quality and to develop trade
products, especially hedge transaction products.

Acknowledgments
The authors thank the three anonymous reviewers for their constructive comments.

Funding
This work is supported by the Fundamental Research Funds for the Central Universities of China,
Project JBK1307093.

References
Ang, A., R. Hodrick, Y. Xing, and X. Zhang. 2006. The cross-section of volatility and expected returns. Journal of Finance 61
(1):259–99. doi:10.1111/jofi.2006.61.issue-1.
Atilgan, Y., and K. O. Demirtas. 2013. Downside risk in emerging markets. Emerging Markets Finance & Trade 49 (3):65–83.
Emerging Markets Finance and Trade

doi:10.2753/REE1540-496X490306.
Baker, M., and J. Wurgler. 2006. Investor sentiment and the cross-section of stock returns. Journal of Finance 61 (4):1645–80.
doi:10.1111/jofi.2006.61.issue-4.
Barlevy, G., and P. Veronesi. 2003. Rational panics and stock market crashes. Journal of Economic Theory 110 (2):234–63.
doi:10.1016/S0022-0531(03)00039-5.
Brooks, C., and A. Katsaris. 2005. A three-regime model of speculative behaviour: Modeling the evolution of the S&P 500
composite index. Economical Journal 115 (505):767–97.
Brown, G. W., and M. T. Cliff. 2004. Investor sentiment and the near-term stock market. Journal of Empirical Finance 11
(1):1–27. doi:10.1016/j.jempfin.2002.12.001.
Campbell, J. Y., and L. Hentschel. 1992. No news is good news: An asymmetric model of changing volatility in stock returns.
Journal of Financial Economics 31 (3):281–318. doi:10.1016/0304-405X(92)90037-X.
Cao, H., J. Coval, and D. Hirshleifer. 2002. Sidelined investors, trading-generated news and security returns. Review of Financial
Studies 15 (2):615–48. doi:10.1093/rfs/15.2.615.
Case, K., and R. Shiller. 1989. The efficiency of market for single-family homes. American Economic Review 79(1):125–37.
Chen, G. J., and Y. J. Zhang. 2009. Short sale restriction, heterogeneous beliefs and stock price crash in China. Journal of
Financial Research 346 (4):80–91.
Chen, J., H. Hong, and J. C. Stein. 2001. Forecasting crashes: Trading volume, past returns, and conditional skewness in stock
prices. Journal of Financial Economics 61 (3):345–81. doi:10.1016/S0304-405X(01)00066-6.
De Long, J. B., A. Shleifer, L. H. Summers, and R. J. Waldmann. 1990. Noise trader risk in financial markets. Journal of
Political Economy 98 (4):703–38. doi:10.1086/jpe.1990.98.issue-4.
Dechow, P., R. Sloan, and A. Sweeney. 1995. Detecting earnings management. Accounting Review 70 (2):193–225.
Dimson, E. 1979. Risk measurement when shares are subject to infrequent trading. Journal of Financial Economics 7
(2):197–226. doi:10.1016/0304-405X(79)90013-8.
Grech, D., and Z. Mazur. 2004. Can one make any crash prediction in finance using the local hurst exponent idea? Physica A:
Statistical Mechanics and its Applications 336 (1–2):133–45. doi:10.1016/j.physa.2004.01.018.
Griffin, J. M., F. Nardari, and R. M. Stulz. 2007. Do investors trade more when stocks have performed well? Evidence from 46
Countries. Review of Financial Studies 20 (3):905–51.
Hong, H., and J. C. Stein. 2003. Differences of opinion, short-sales constraints, and market crashes. Review of Financial Studies
16 (2):487–525. doi:10.1093/rfs/hhg006.
Hutton, A. P., A. J. Marcus, and H. Tehranian. 2009. Opaque financial reports, R2, and crash risk. Journal of Financial
Economics 94 (1):67–86. doi:10.1016/j.jfineco.2008.10.003.
Jin, L., and S. C. Myers. 2006. R2 around the world: New theory and new tests. Journal of Financial Economics 79 (2):257–92.
doi:10.1016/j.jfineco.2004.11.003.
Jones, C. M., and A. L. Owen. 2002. Short-sale constraints and stock returns. Journal of Financial Economics 66:207–39.
doi:10.1016/S0304-405X(02)00224-6.
Kim, J.-B., Y. Li, and L. Zhang. 2011a. CFOs versus CEOs: Equity incentives and crashes. Journal of Financial Economics 101
(3):713–30. doi:10.1016/j.jfineco.2011.03.013.
Kim, J.-B., Y. Li, and L. Zhang. 2011b. Corporate tax avoidance and stock price crash risk: Firm-level analysis. Journal of
Financial Economics 100 (3):639–62. doi:10.1016/j.jfineco.2010.07.007.
Lee, C. M., A. Shleifer, and R. Thaler. 1991. Investor sentiment and the closed-end fund puzzle. Journal of Finance 46
(1):75–109. doi:10.1111/j.1540-6261.1991.tb03746.x.
Li, S. Y. 2013. Robustness, momentum and investment strategies on Chinese stock market. China Economic Quarterly 12
(2):453–74.
 18 Y. YIN AND R. TIAN

Lindahl-Stevens, M. 1980. Redefining bull and bear markets. Financial Analysts Journal 36 (6):76–77. doi:10.2469/faj.
v36.n6.76.
Liu, B., and L. Pi. 2007. Momentum strategy and constrain strategy: New evidence from China A-share market. Journal of
Financial Research 326 (8):154–66.
Marin, J. M., and J. Olivier. 2008. The dog that did not bark: Insider trading and crashes. Journal of Finance 63 (5):2429–76.
doi:10.1111/jofi.2008.63.issue-5.
Ritter, J. R., and I. Welch. 2002. A review of IPO activity, pricing, and allocations. Journal of Finance 57 (4):1795–828.
doi:10.1111/jofi.2002.57.issue-4.
Rodriguez, J., and A. Sbuelz. 2006. Understanding and exploiting momentum in stock returns. In Advances in corporate finance
and asset pricing, ed. L. Rennegoog, 485–504. Amsterdam, The Netherlands: Elsevier Science.
Romer, D. 1993. Rational asset-price movements without news. American Economic Review 83 (5):1112–30.
Wu, Y. R., and L. Y. Han. 2007. Imperfect rationality, sentiment and closed-fund puzzle. Economic Research Journal
(3):117–29.
Xu, N., Y. Jiang, H. Yi, and Z. Xu. 2012. Conflicts of interest, analyst optimism and stock price crash risk. Economic Research
Journal 47 (7):127–40.
Xu, N., Y. Yu, and H. Yi. 2013. Herd behavior of institutional investors and stock price crash risk. Management World 238
(7):31–43.
Emerging Markets Finance and Trade