foreign institutional ownership and stock liquidity in...

Foreign Institutional Ownership and Stock Liquidity in China

1


Abstract

The literature widely documents the negative liquidity impact of foreign participation in firms

that permit high foreign institutional ownership. This paper employs a unique setting for the limited

participation of qualified foreign institutional investors (QFIIs) in China`s A-share market and

examines how these impacts on stock liquidity in emerging markets. Contrary to the findings in the

literature, foreign investor participation helps enhance the liquidity of affected stocks by promoting

trade activities in increasing trading volume. The improvement in liquidity is more significant in small

firms compared to large firms. Our findings are robust to endogeneity and the possible influence of the

stock market shock, industry effects and the stock exchange. Further, the liquidity improving effects

of QFII are even stronger when the analysis is performed on a subsample of QFII firms.

Key words: liquidity, foreign institutional investor, QFII

1. Introduction

From the view of globalization, the general trend of financial liberalization and capital flows have

become an inevitable tendency and China is increasingly integrated with the global economy. China

allows foreigners to invest in A-share stocks through the QFII（qualified foreign institutional investor）

system instituted on December 1,2002. Nevertheless, the approach taken by the Chinese government

to liberalize the financial market is a cautious and conservative one. More and more foreign investors

invest in China’s listed company, which have a great impact and stock market. Our study focuses on

the influence of foreign institutional participation on stock market liquidity under China’s QFII scheme.

Liquidity is an important indicator of the efficiency of the stock market. Adequate liquidity facilities

investors to complete purchase requests for stocks faster and is also an important guarantee for

corporate fund raisers to conduct listing financing and refinancing.

A well-known stream of literature holds different arguments about the relationship between the

foreign institutional investor ownership and stock liquidity. On the one hand, foreign institutional

investors usually have information advantage, rational concept and much capital. They can help

investors to form rational investment concept and then improve market efficiency (Grinblatt and

Keloharju,2000; Froot, O’Connell and Seasholes, 2001; Seasholes,2004). Alternatively, the

introduction of foreign investment can improve the quality of information disclosure of the company,

which can reduce the potential risk or transactional cost. Hence it positively influences the liquidity

(Bae et al,2006; Li et al,2011). On the other hand, the presence of large shareholders could reduce the

number of shares available to the public for trading, thus reducing stock liquidity by lowering trading

activity (Demsetz, 1968; Bolton and Thadden, 1998; Rubin, 2007; Brockman, Chung and Yan, 2009；

Deng,2018). Or these talented foreign institutional investors could be information traders, thus bring

more sever information asymmetry which negatively influences the liquidity (Bhide,1993;

Agarwal,2007).

China’s QFII scheme is restrictive: the total shares held by each (all) QFII in one listed company


2

is not permitted to exceed 10% (20%) of the total outstanding shares of the company. Thus, it`s possible

to identify the influence of foreign institutional investment on stock liquidity in a low ownership frame

work.

Our study will employ panel regression model with firm and time fixed effect to examine the

influence of foreign institutional (QFII) ownership on stock liquidity of A share listed firms. At the

same time, the study will also try to use different measures to consider market liquidity (illiquidity

ratio and trading volume). In addition, we also try to discuss the heterogeneity in different firm size

and use different measures for robustness test.

The rest of the paper is structured as follows. Section 2 is the literature review which develops

our hypotheses. Section 3 explains the measurement of liquidity for individual stocks: this section also

develops the econometric specifications of our panel data regressions. Section 4 describes the study’s

data sources and provides descriptive statistics for liquidity measures, firm characteristics, and

ownership variables. Section 5 reports the results of our main panel data regressions. Section 6

examines the endogeneity of foreign institutional ownership and causality. Section 7 presents the

results of various robustness checks. The final section summarizes the study and provides concluding

remarks.

2. Literature review and Hypothesis development

2.1. Liquidity

Liquidity, as one of the core concepts of market microstructure theory, is one of the important

variables that affect the long-term healthy operation of the market. Moderate liquidity can promote

market transactions, improve market efficiency and reduce financing costs. Keynes put forward the

concept of liquidity in 1930, saying that assets with better liquidity are assets that are easier to realize.

According to the definition of Amihud and Mendelson (1986), liquidity is the time or cost required

to complete the transaction within a certain period of time, or the time or cost required to find an ideal

price. According to Goyenkoetal (2009)[5], liquidity includes transaction cost and transaction speed.

Although the definition of liquidity includes multiple dimensions, the use of this word in the

literature is often limited to specific indicators. For example, the bid-ask spread (Amihud and

Mendelson ,1986), the turnover rate (Chordiaetal ,2001), or the Amihud measure (Amihud ,2002) and

so on.

Macroeconomic environment, policies, industry characteristics, enterprise characteristics, asset

characteristics and so on will affect liquidity. Generally speaking, there is a negative correlation

between equity concentration and liquidity. The larger shareholders, or the higher the proportion of

shares held by a certain shareholder, will lead to a much decrease in liquidity. In addition, the level of

information in the market will also significantly affect the level of liquidity, Heflin (2001)[7] found that

for companies with more adequate information disclosure and more transparent information, the

effective spread is lower. On the other hand, the companies with different sizes, different growth rate,

are different in information transmission and disclosure, which determines the difference in the mastery

of information. As a result, the impact of foreign investors on liquidity has been magnified or reduced

to a certain extent.


3

2.2. QFII and liquidity

Studies have shown that foreign investors as informed traders may improve the liquidity of stocks.

Informed trader refers to the party who has relatively more information according to the differences in

the understanding of information among traders in the stock market, as opposed to the concept of

uninformed trader.

First of all, foreign investors as informed traders, the identity of the information advantage is easy

to cause herding effect in the market.

Seasholes (2001) [4]points out that foreign investors have more resources, more experience in

international capital markets and more effective investment strategies. Based on QFII's heavy holdings

in the mainland stock market, Sun Li and Lin Li (2006) [11]found that the capital performance of QFII

reflects the strong adaptability of foreign investors to China's economic development and better grasp

the development cycle of the industry. This significant investment advantage is likely to attract

investors' attention, attract more money and increase liquidity. Tong Yuansong and Wang Guangwei

(2001)[12] pointed out that qualified foreign investors, especially mature overseas institutions, are likely

to adopt the strategy of rational investment, which is reflected in the preference of performance stocks

in the stock market. The higher the proportion of foreign ownership, the more attention will be paid to

the stock of the listed company by other investors. With the increase of the amount of investment and

the number of transactions, the bid-ask spread of the corresponding stock decreases, which not only

increases liquidity, but also weakens the volatility of the stock price.

Second, as informed traders, foreign investors compete with other information traders in the

market.

Grossman (1980)[6] pointed out that in a market with asymmetric information, the competition

and monopoly behavior of informed traders will have an impact on the integration of private

information into the stock price. Subrahmanyam (1991) [10]believes that not only foreign investors are

informed traders, but also relatives of foreign company managers and some institutional investors are

also informed traders, and their competitive behavior will make this information included in the stock

price. Thus, the information efficiency of the stock is increased, and then the liquidity of the stock is

increased. By analyzing a non-competitive market model, Subrahmanyam (1991) [10]concludes that

liquidity is non-monotonous in terms of the number of informed traders and the accuracy of

information, and that the competition between them in the market will lead to more information into

trading. The improvement of information efficiency brings the increase of stock liquidity.

In addition, qualified foreign institutional investors lead to a significant improvement in the

information disclosure of the company, and the improvement of the quality of information disclosure

can reduce the potential risks and transaction costs, and improve the liquidity of stocks (Bae, 2006 and

Lietal, 2011[2] ).

Stulz (1999)[9] believes that when foreign investors invest in domestic stocks to become

shareholders of listed companies, they have a stronger and more professional ability to supervise the

company than small and medium-sized retail investors. Therefore, it will effectively improve the level

of corporate governance of domestic listed companies, improve the level and quality of information

disclosure to a greater extent, and then reduce the degree of information asymmetry of stocks.

Therefore, the stock price can better reflect the effective information related to the company, then

reduce the information advantage of the private information holder, so that for investors who do not

have the information advantage, the risk premium required to invest in the stock is reduced and finally


4

enhance the liquidity of stocks.

Ding (2017)[3] believes that the information advantage of foreign institutional investors will be

particularly critical in developing countries, especially in China, because the problems of information

transparency and information asymmetry in China have always been very serious.

Many scholars have also discussed it from other angles. Amihud and Mendelson (2002) [1]believes

that after the introduction of foreign investors, domestic listed companies can enrich the equity

structure and allocate certain risks to foreign investors, and the diversification of shareholders increases

the liquidity of stocks. Subrahamanyam (1991)[10] in the study of stock information feedback

mechanism found that the better the stock liquidity, the more informed traders participate, and the cost

of communication and coordination between stakeholders and management is reduced. Then through

the feedback mechanism the ability of management of the company to make efficient decisions

promote and the information contents of the stock improve. Mendelson and Tunca (2004)[8] looking at

trading activities in an incomplete efficient market found that long-term information asymmetry among

traders leads to when more information is contained in a stock, a reduction in trading risk is

accompanied by a decrease in liquidity costs. That is, the liquidity of stocks increases.

Based on these, we propose the following two assumptions:

H1: The higher the QFII shareholding ratio is, the greater the stock trading volume is, and the

better the stock liquidity of the company is.

H2: Other conditions remain unchanged, the smaller the size of the company, the more obvious

the role of QFII in promoting stock liquidity

3. Measurement of variables and model specification

Due to the latent nature of liquidity and its multiple dimensions, a single measure cannot capture

all its features. We apply measures commonly used in the liquidity literature to reflect liquidity. The

control variables included in the panel data regressions are chosen mainly to conform to the literature.

3.1. Measurement of dependent variables

3.1.1. Illiquidity

To measure stock liquidity, we start by defining stock liquidity from the angle of price impact. We

use the illiquidity indicator Illiq proposed by Amihud (2002) to proxy for a stock's liquidity (Karolyi

et al., 2012). Illiq is the (Amihud) ratio of the absolute return to dollar trading volume, which can

effectively measure the price impacts of the unit trading volume (in dollars) in international stock

markets (Kang and Zhang, 2010; Fong et al., 2017).

𝐼𝑙𝑙𝑖𝑞𝑖,𝑡 = |𝑅𝑒𝑡𝑢𝑟𝑛𝑖,𝑡|

𝑉𝑜𝑙𝑢𝑚𝑒𝑖,𝑡

Where 𝑅𝑒𝑡𝑢𝑟𝑛𝑖,𝑡 is stock i`s average quarterly return in time t, 𝑉𝑜𝑙𝑢𝑚𝑒𝑖,𝑡 is quarterly trading

volume. The higher the Amihud ratio Illiq is, the higher the stock`s illiquidity, the lower the stock`s

liquidity.


5

3.1.2. Trading Volume

TV represents trading volume that we had used in the calculation of Illiquidity ratio. Intuitively

the higher the trading volume is, the higher the stock`s liquidity.

3.2. Measurement of control variables

Previous studies indicate that firm size, stock price volatility, share price and turnover are

associated with liquidity. (Benston and Hagerman, 1974; Stoll and Whaley, 1983; Agarwal, 2007;

Brockman, Chung and Yan, 2009). For firm size, Stoll and Whaley (1983) argue that trading smaller

stocks is more expensive because less relevant information about these firms is available. According

to Chordia, Roll and Subarhmanyam (2001), volatility increases the market makers’ inventory risk and

the risk of unintentionally engaging in short-term speculative trades. Previous studies show that the

spread could be correlated with price nonlinearly; hence, it is standard practice to take the natural

logarithm of share price (Brockman, Chung and Yan, 2009; Chung, Elder and Kim, 2010). Agarwal

(2007) argues that high turnover may reflect belief dispersion induced by information differences

among investors.

In our panel data regressions, we control for these effects by including stock return volatility

(VOL) estimated by the standard deviation of daily stock returns, firm size measured by the book value

of a firm (SIZE), the natural logarithm of the share price (LNP), and the turnover rate (TO) as

explanatory variables. In addition, the degree of leverage (LEV) is included because the security design

literature has recognized that a firm’s capital structure can affect the degree of information disclosure

(Diamond and Verrecchia, 1991). We also control for ownership by domestic institutional investor (DI)

by including the percentage ownership of the five largest domestic institutions (i.e., open-end funds,

security, insurance, trust companies, and pension funds).

Finally, we include two control variables unique to China’s stock market: a state-owned enterprise

(SOE) dummy (STATE) and the nontradable share ratio (NT). China’s stock market is characterized

by the dominance of SOEs, and previous studies find that whether a firm is state=owned does matter

for stock liquidity due to its links to the government (Ding, 2015). Therefore, we control for SOE

dummy as a proxy for political connections, by defining a dummy variable taking the value of 1 if a

firm is state-owned. To improve the structure of corporate governance and market liquidity, a split-

share structure reform was introduced by the Chinese government in 2005 to dismantle the dual-share

structure by converting nontradable shares into tradable shares. Before the reform, the split-share

structure and the associated overhang of the nontradable shares presumably impaired liquidity (Jiang,


6


7


8

Laurenceson and Tang, 2008; Beltratti, Bortolotti and Caccavaio, 2012). We use a firm’s NT (i.e., the

number of nontradable shares divided by the total number of shares) to control for the split-share

structure. Many explanatory variables are available only on a quarterly basis; we mearsure these

variables in the beginning of each quarter (indicated by subscript t-1 in the regression). The remaining

variables are measured as the average over each quarter (indicated by subscript t in the regression).

3.3. Specification of panel regression models

Equation (1) specifies the panel data model for examining the relationship between the QFII

participation and stock market liquidity. The dependent variable measuring liquidity is the illiquidity

ratio and trading volume.

We follow Rubin (2007), Brockman, Chung and Yan (2009) and Chung, Elder and Kim (2010)

and estimate the following main panel regression for firm i and time t:

𝐿𝐼𝑄𝑖,𝑡 = 𝛼0 + 𝑎1𝑄𝐹𝐼𝐼𝑖,𝑡−1 + 𝛼2𝐷𝐼𝑖,𝑡−1 + 𝛼3𝑆𝐼𝑍𝐸𝑖,𝑡−1 + 𝛼4𝑆𝑇𝐴𝑇𝐸𝑖,𝑡−1 + 𝛼5𝐿𝐸𝑉𝑖,𝑡−1 +

𝛼6𝑉𝑂𝐿𝑖,𝑡 + 𝛼7𝐿𝑁𝑃𝑖,𝑡 + 𝛼8𝑇𝑂𝑖,𝑡 + 𝛼9𝑁𝑇𝑖,𝑡−1 + ∑ 𝛽𝑞𝐷𝑞 + 𝜀𝑖,𝑡𝑞 (1)

where LIQ = Illiq or TV, and QFII is the percentage ownership by qualified foreign institutional

investors. Due to the high skewness and kurtosis, we transform all dependent variables by taking the

natural logarithm. The control variables, namely domestic institutional ownership (DI), firm size

(SIZE), SOE dummy (STATE), leverage ratio (LEV), stock return volatility (VOL), the natural

logarithm of the share price (LNP), turnover rate (TO), and the nontradable share ratio (NT). The

quarterly time dummies (D) capture common shocks and potential time trends. For each measure of

liquidity, we run two panel regressions. The first regression has QFII ownership as an explanatory

variable, and the second has both QFII ownership and domestic institutional ownership as explanatory

variables (all regressions include the remaining control variables). If QFII ownership is positively

related to liquidity, the coefficient on QFII should have a negative sign for the illiquidity and a positive

sign for TV.

4. Data

4.1. Data sources and data filtering

The data for the QFII ownership which is our independent variable is from CSMAR. The

dependent variables which measure the liquidity of stocks are from REESET. The Illiq and TV are

quarterly averaged across all trading days for each stock in each year.

There are several control variables in our paper. Agarwal (2007) argues that high turnover may

reflect belief dispersion induced by information differences among investors. The turnover rate is

calculated as the total trading volume in a year divided by shares outstanding. Volatility is calculated

as the standard deviation of daily stock returns over quarters and according to Chordia, Roll and

Subrahmanyam (2001), volatility increases the market makers’ inventory risk and the risk of

unintentionally engaging in short-term speculative trades. The security design literature has recognized

that a firm’s capital structure can affect the degree of information disclosure (Diamond and Verrecchia,


9

1991) so consequently, the capital structure can be associated with market liquidity through the

informational channel. The leverage ratio is measured as the company’s total debt divided by its total

assets. The firm age is measured by current year deducting the established year. We also control for

ownership by domestic institutional investors (DI) by including the percentage ownership of the five

largest domestic institutions. Return on assets is measured by net profit divided by total assets. Finally,

because China’s stock market is characterized by the dominance of SOEs, and previous studies find

that whether a firm is state-owned does matter for stock liquidity due to its links to the government

(Ding, 2015). Therefore, we control for SOE dummy as a proxy for political connections, by defining

a dummy variable taking the value of 1 if a firm is state-owned.

4.2. Descriptive statistics and preliminary analysis

In this section, we present the descriptive statistics of the dependent variables, foreign and

domestic institutional ownership variables, and control variables. We proceed to a preliminary

univariate analysis comparing firms with participating foreign institutional investors and firms without

participating foreign institutional investors.

4.2.1. Summary statistics

Table 1 presents the summary statistics of the variables used in this study. We see that, on average,

QFIIs hold approximately 1.6% of a firm’s outstanding shares, while domestic institutions, in aggregate,

hold approximately 3.7% (open-end funds, security, insurance, trust companies, and pension funds).

The average Illiquidity ratio is approximately -21.919. For trading volume, the mean is 0.137 million

shares per day.

4.2.2. Correlation coefficient

Table 2(a) is the correlation coefficient. We can have a preliminary idea that Illiq is negatively

correlated to QFII ownership, and TV is positively correlated to QFII ownership.

4.2.3. Preliminary univariate analysis

To provide a visual impression of the dynamic relationship between liquidity and ownership, we

plot the total value of QFIIs’ shareholdings each quarter and the two average quarterly liquidity

measures, the Illiq and TV, as seen in Figure 1. The right vertical axis shows QFII participation, and

the left vertical axis shows liquidity over the sample period (2004Q3–2019Q1). The visual evidence in

Figure 1 suggests a negative relationship between QFII holdings and market illiquidity over time.

Figure 2 presents a visual evident of positive relationship between QFII holdings and trading volume.

We perform preliminary univariate tests on the two liquidity measures, Illiq, TV and the two

trading activity measures for two subsamples of firms: those with QFIIs and those without (non-QFII

firms).

We assume that, if QFII participation increase liquidity, we expect to find different (average)

values of the various measures for the two groups of firms. The evidence in Table 2(b) indeed suggests


10

that this is the case, and all differences are statistically significant at the 5% level.

Figure 1

Figure 2


11

Table 2 (b)

Preliminary analysis of the relationship between liquidity and QFII

variables Mean for QFII firms

Observations=10642

Mean for non-QFII firms

Observations=119701

p-Value

Illiquidity

-22.475 -21.870 0.000

Trading Volume

(in Billions, per day)

18.356 17.856 0.000

5. Empirical results

We begin our empirical analysis by investigating whether foreign institutional investors (QFIIs)

are associated with liquidity on the Chinese stock market using the panel regression in Equation (1).

Next, we analyze whether there is time effect. In the following section, we include large ownership by

strategic foreign institutional investors as an additional explanatory variable in addition to QFII

ownership to differentiate the link to liquidity between institutions with smaller and larger investments.

Table 3

QFII relationship with liquidity

The table shows the panel regressions using the two liquidity measures regressed on lagged ownership of

QFIIs. The two measures of liquidity are the Illiquidity ratio in Amihud(2002) and the trading volume

(TV).DI denotes the lagged domestic institutional ownership in the top 10 outstanding shareholders of a

firm. Remaining control variables refer to firm and share characteristic: firm size measured by the log of

book value(SIZE),degree of leverage(LEV), the log volatility of stock returns(VOL),the log of share

price(LNP) and the log of turnover rate of shares traded(TO). We also control for the nontradable ratio(NT)

and the state dummy(State),with the dummy variable equal to 1 if a firm is state-owned. The period of study

if from 2004Q1 to 2019Q1. Driscoll-Kraay standard errors reported in parentheses are robust to correlation

across residuals within a firm over time and across firms in the same year and different firms.

Dependent variables

Independent

variables

(1)

Illiq

(2)

Illiq

(3)

TV

(4)

TV

QFIIi,t -0.0111*

(0.0063)

-0.0113*

(0.0063)

0.0105*

(0.0055)

0.0103*

(0.0054)

DIi,t

0.0030***

(0.0009)

0.0027***

(0.0009)

TOi,t -0.1402***

(0.0044)

-0.1391***

(0.0044)

0.2186***

(0.0057)

0.2196***

(0.0058)

SIZEi,t-1 -0.6729***

(0.0176)

-0.6710***

(0.0176)

0.5760***

(0.0157)

0.5777***

(0.0158)

LEVi,t-1 -1.0718***

(0.0670)

-1.0714***

(0.0670)

1.0483***

(0.0632)

1.0487***

(0.0632)

VOLi,t 1.5607***

(0.4898)

1.5538***

(0.4877)

0.8061**

(0.3694)

0.7998**

(0.3678)


12

(Continued)

NTi,t-1 0.0220***

(0.0004)

0.0219***

(0.0004)

-0.0220***

(0.0004)

-0.0221***

(0.0004)

LNPi,t-1 -0.7480***

(0.0137)

-0.7597***

(0.0141)

0.7270***

(0.0135)

0.7164***

(0.0141)

State Yes Yes Yes Yes

N 57865 57865 57865 57865

adj. R2 0.695 0.695 0.730 0.730

F 3703.9643 3224.0997 4289.7284 3768.3899

Standard errors in parentheses (* p<0.1, ** p<0.05, *** p<0.01)

5.1. Foreign institutional ownership and stock market liquidity

Table 3 shows the results of the first panel data regressions. Our focus is on the link between

foreign institutional ownership and market liquidity. The first two regressions (columns) show the

results for the illiquidity ratio. We find clear evidence that increased foreign institutional ownership is

associated with a lower illiquidity ratio, which means that increased foreign institutional ownership is

associated with higher stock market liquidity. The inclusion of domestic institutions in the regressions

does not alter the relationship between foreign institutional ownership and the illiquidity ratio. The

coefficient estimate on foreign institutional ownership (QFII) is statistically highly significant and

stable at –0.0111 and –0.0113 in the two regressions. The numerical magnitudes of the coefficient

estimates indicate that a 10% higher foreign institutional ownership is associated with an

approximately -0.0111 lower Illiquidity ratio.

Table 4

Fixed effect model

The table shows the fixed effects panel regressions of the Illiquidity (Illiq) and the trading volume (TV) on

the QFII ownership. The period of study is from 2004Q3 to 2019Q1. Driscoll-Kraay standard errors reported

in parentheses are robust to correlation across residuals within a firm over time and across firms in the same

year and different quarters.

Dependent variables

Independent

variables

(1)

Illiq

(2)

Illiq

(3)

TV

(4)

TV

QFIIi,t-1 -0.0091***

(0.0032)

-0.0077**

(0.0032)

0.0145***

(0.0053)

0.0126**

(0.0052)

DIi,t

-0.8629***

(0.0396)

1.1495***

(0.0914)

SIZEi,t-1 -0.3233***

(0.0045)

-0.3211***

(0.0045)

0.2910***

(0.0115)

0.2881***

(0.0114)

LEVi,t-1 -0.5500***

(0.0208)

-0.5377***

(0.0207)

0.6351***

(0.0459)

0.6187***

(0.0450)

VOLi,t 1.1389***

(0.0628)

1.1616***

(0.0626)

0.8658**

(0.3445)

0.8355**

(0.3353)


13

(Continued)

NTi,t-1 0.0081***

(0.0003)

0.0082***

(0.0003)

-0.0105***

(0.0005)

-0.0106***

(0.0005)

LNPi,t -0.5528***

(0.0053)

-0.5155***

(0.0056)

0.5168***

(0.0114)

0.4672***

(0.0120)

TOi,t -0.1057***

(0.0011)

-0.1091***

(0.0011)

0.1649***

(0.0041)

0.1694***

(0.0042)


Time dummy Yes Yes Yes Yes

N 64654 64654 64654 64654

adj. R2 0.780 0.781 0.800 0.803

F 11563.6160 11119.6109 4096.9190 3934.5918


The next two regressions show the results for TV (with and without domestic institutional

ownership as a control). Our evidence shows that foreign institutional ownership is significantly and

positively related to trading volume, with coefficient estimates 0.0105 and 0.0103 (i.e., foreign

institutional ownership is again linked to a higher liquidity, as with the spread). The numerical

magnitude of the coefficient estimate indicates that a 10% increase in foreign institutional ownership

is associated with an approximately 1% increase in the trading volume.

5.2. Fixed effect model

Unobservable time-invariant factors may simultaneously affect both the left-hand side and the

right-hand side of the regression in Equation (1). If so, the regression can suffer from omitted variable

bias. We estimate a time fixed effects regression in an attempt to control for possible omitted variables.

We report the results of the firm fixed effects regression in Table 4.

We find that the positive relationship between the QFII ownership and liquidity measures dose

not altered when the fixed effects model is applied. The results for the fixed effects estimation

nonetheless again strengthen our previous finding that the positive relationship between foreign

institutional ownership and liquidity is mainly due to increased trading volume.

Table 5

Different firm size

The table shows the fixed effects panel regressions of the Illiquidity (Illiq) and the trading volume (TV) on

the QFII ownership in subsample divided by the medium of firm size. The period of study is from 2004Q3

to 2019Q1. Driscoll-Kraay standard errors reported in parentheses are robust to correlation across residuals

within a firm over time and across firms in the same year and different quarters.

Dependent variables

Large firms Small firms

Independent

variables

(1)

Illiq

(2)

TV

(3)

Illiq

(4)

TV

QFIIi,t-1 -0.0022

(0.0059)

0.0068

(0.0062)

-0.0106**

(0.0052)

0.0202***

(0.0044)


14

(Continued)

DIi,t-1 -0.6926***

(0.1181)

0.9246***

(0.1197)

-0.4560***

(0.1164)

0.9676***

(0.1179)

SIZEi,t-1 -0.3514***

(0.0192)

0.3114***

(0.0189)

-0.2750***

(0.0181)

0.2677***

(0.0174)

LEVi,t -0.2907***

(0.0712)

0.3530***

(0.0731)

-0.6550***

(0.0568)

0.7860***

(0.0600)

VOLi,t 1.3563

(0.8703)

0.4012

(0.5039)

1.1775***

(0.3743)

1.0405***

(0.3212)

NTi,t-1 0.0057***

(0.0007)

-0.0090***

(0.0007)

0.0086***

(0.0007)

-0.0103***

(0.0006)

LNPi,t -0.5693***

(0.0174)

0.5203***

(0.0182)

-0.4798***

(0.0159)

0.4087***

(0.0155)

TOi,t -0.1383***

(0.0081)

0.2177***

(0.0108)

-0.0978***

(0.0037)

0.1524***

(0.0043)



N 32814 32814 31840 31840

adj. R2 0.783 0.799 0.769 0.801

F 1610.4196 1840.2160 1514.0860 1554.2132


5.3. Different firm size

We investigate whether the positive relationship between QFII participation and liquidity is

prevalent in both the large and small firm. We therefore perform a panel regression analysis of the

model in Equation (1) by splitting the total sample into two subsamples of stocks listed by different

firm size. Table 5 shows that the coefficient of QFII ownership for the Illiq (the TV) is negative

(positive) and significant only on small firms. We find that the magnitudes of the coefficient estimate

for Illiq and TV on the small firms (–0.0106 and 0.0202, respectively) are similar to the full sample

estimates; however, the influence of QFII ownership is insignificant on large firms. One possible

reason for this difference is that large firms themselves are more regulated, the influence of QFII

ownership is then less obvious than in small firms

6. Endogeneity of foreign institutional ownership

Another intrinsic explanation for the positive correlation between QFII ownership is that QFII is

more likely to choose stocks with high liquidity, which may lead to reverse causality. Therefore, in this

section, we will study endogeneity to reduce this concern about causality and endogeneity.


15

6.1. Endogeneity test

We perform an endogeneity test developed by Wu (1973) and Hausman (1978) to examine

whether the QFII ownership or the illiquidity variables are endogenous. We conduct two-stage least

squares (2SLS) regressions, where the first-stage regression includes the same control variables as in

the main regression (Table 3), together with a set of new explanatory variables. We rely on previous

work on the investment preferences of foreign funds by Kang and Stulz (1997), Heflin and Shaw (2000),

Dahlquist and Robertsson (2001), Rubin (2007), and Liu, Bredin, Wang and Yi (2014) and use the

following additional explanatory variables in the first-stage regression: the return on assets (net profit

divided by total assets, ROA), the firm age (AGE), and an ownership concentration index, the

Herfindahl 10 index (OC)1.We also include industry fixed effect dummies (IND) that are equal to 1 if

the firm operates in a given industry2. The first-stage regression is：

𝑄𝐹𝐼𝐼𝑖,𝑡 = 𝜔0 + 𝜔1𝐿𝐼𝑄𝑖,𝑡−1 + 𝜔2𝐷𝐼𝑖,𝑡−1 + 𝜔3𝑆𝐼𝑍𝐸𝑖,𝑡−1

+𝜔4𝑆𝑇𝐴𝑇𝐸𝑖.𝑡−1 + 𝜔5𝐿𝐸𝑉𝑖,𝑡−1 + 𝜔6𝑉𝑂𝐿𝑖,𝑡 + 𝜔7𝐿𝑁𝑃𝑖,𝑡 + 𝜔8𝑇𝑂𝑖,𝑡

+𝜔9𝑁𝑇𝑖,𝑡−1 + ∑ 𝛽𝑞𝐷𝑞

𝑞

+ 𝜔10𝑅𝑂𝐴𝑖,𝑡−1 + 𝜔11𝐴𝐺𝐸𝑖,𝑡−1 + 𝜔12𝑂𝐶𝑖,𝑡−1

+ ∑ 𝜋𝑚𝐼𝑁𝐷𝑚,𝑖,𝑡−1𝑚 + 𝜇𝑖,𝑡 (2)

The second stage estimates the baseline regression in the main regression (Table 3) by replacing

the actual QFII holding with its lagged residual value (R_QFII) from the first-stage regression. If QFII

is endogenous, the coefficient on R_QFII in the second-stage regression is statistically different from

zero; if QFII is not endogenous, the coefficient is not statistically different from zero. The results are

shown in Table 6.

Table 6

Endogeneity test

The table shows a 2SLS regression analysis. The QFII ownership is the dependent variable in the first stage

estimation described in (first stage), while the second stage estimates the baseline model in the main

regression (Table 3), by replacing the actual QFII ownership with lagged value of residuals R_QFII from

the first stage estimation. We use two illiquidity measures, the Illquidity from Amihud（2002）and the

trading volume (TV). In addition to the control variables in the main regression (Table 3), we add new

control variables to the first stage estimations: the return on assets (ROA), the firm age (AGE), an ownership

concentration index (Herfindahl 10 index, OC) and industry fixed effect dummies that are equal to 1 if the

firm operates in a given industry. The period of study if from 2004Q1 to 2019Q1. Driscoll-Kraay standard

errors reported in parentheses are robust to correlation across residuals within a firm over time and across

firms in the same year and different firms. There are 63592 and 63594 observations in each 2SLS analysis.

Dependent variables

Independent

variables

(1)

QFII

(2)

Illiq

(3)

QFII

(4)

TV

Illiqi.t-1

-0.0011

(0.0051)

1 The Herfindahl 10 index measures the degree of ownership dispersion in the top 10 shareholder structure. 2 The industry classification is released by the CSRC, and the data are provided in the CCER database. There are 13

different industries at the level of classification used in equation 2.


16

(Continued)

TVi,t-1

-0.0022

(0.0052)

R_QFIIi,t-1

-0.0085*

(0.0045)

0.0100**

(0.0045)

SIZEi,t 0.0003

(0.0131)

-0.3168***

(0.0122)

0.0013

(0.0131)

0.3046***

(0.0116)

LEVi,t-1 -0.0337

(0.0484)

-0.6046***

(0.0454)

-0.0324

(0.0484)

0.6229***

(0.0449)

VOLi,t 0.0252

(0.3649)

9.5309***

(0.4206)

0.0585

(0.3594)

21.4262***

(0.3942)

NTi,t-1 0.0008

(0.0009)

0.0043***

(0.0006)

0.0008

(0.0009)

-0.0052***

(0.0005)

LNPi,t 0.0589***

(0.0123)

-0.5640***

(0.0113)

0.0607***

(0.0123)

0.4784***

(0.0108)

TOi,t -0.0027

(0.0025)

-0.1396***

(0.0029)

-0.0026

(0.0025)

0.1478***

(0.0029)

ROAi,t 0.0008

(0.0008)

0.0008

(0.0008)

AGEi,t 0.0124**

(0.0060)

0.0126**

(0.0060)

COi,t -0.0404

(0.1278)

-0.0466

(0.1256)



Industry dummy Yes Yes Yes Yes

N 63592 63592 63594 63594

adj. R2 0.004 0.806 0.004 0.831

F 2.7906 2185.1637 2.8181 2653.1149


We find that the coefficient estimates on R_QFII, in column 2 for the Illiquidity (Illiq) and in

column 4 for the trading volume (TV), are not statistically different from zero. These results suggest

that our findings about the relationship between the QFII ownership and liquidity do not suffer from

endogeneity bias.

6.2. First difference model

We estimated the panel model in the main regression (Table 3) in first difference form and results

are reported in Table 7. We find that the change in foreign ownership is significantly negatively

associated with the change of the Illiquidity. However, consistent with the results for the level of the

trading volume (TV), there is no significant association between the change in foreign ownership and


17

change in the trading volume (TV). These results strengthen our previous finding that the positive

relationship between foreign institutional ownership and liquidity.

Table 7

First Difference model

The table shows the panel regressions of changes in the Illiquidity and the trading volume (TV) on changes

in the QFII ownership. The period of study if from 2004Q1 to 2019Q1. Driscoll-Kraay standard errors

reported in parentheses are robust to correlation across residuals within a firm over time and across firms in

the same year and different firms. There are 64654 observations in each regression.

Dependent variables

Independent

variables

(1)

ΔIlliq

(2)

ΔTV

ΔQFIIi,t-1 -0.0004**

(0.0002)

0.0001

(0.0002)

ΔDIi,t-1 -0.0063***

(0.0020)

-0.0107***

(0.0021)

ΔSIZEi,t-1 -0.0011

(0.0007)

0.0008

(0.0007)

ΔLEVi,t-1 -0.0006

(0.0027)

-0.0023

(0.0025)

ΔVOLi,t -0.1135**

(0.0500)

0.0243

(0.0159)

ΔNTi,t-1 -0.0005***

(0.0001)

-0.0007***

(0.0000)

ΔLNPi,t 0.0305***

(0.0006)

0.0320***

(0.0007)

ΔTOi,t 0.0067***

(0.0002)

0.0115***

(0.0003)

State Yes Yes

Time dummy Yes Yes

N 47532 47532

adj. R2 0.459 0.614

F 1323.6975 1420.5827


The magnitude of the estimated coefficients of the change in foreign ownership on the change of

the Illiquidity is smaller compared to the estimations in level for the same variable. The economic

implication is that foreign institutions’ shorter-term net purchases have a much smaller impact on

liquidity than changes in their longer-term holdings, which are presumably governed by their longer-

term strategic requirements. However, the levels of statistical significance for the Illiquidity and trading

volume in the first difference regressions are much reduced as well. Even though our first difference

regression results are consistent with our level regression results, a plausible interpretation of the

reduced magnitude and statistical significance is that QFII ownership for many firms displays only

limited variation over time but varies substantially across firms. Taking first differences, then,

substantially reduces the cross-sectional variation and leaves only shorter-term time series information


18

with which to identify the coefficients of interest in the econometric model.

7. Additional robustness checks

We estimate three alternative specifications of the panel regression in the main regression (Table

3). First, the sample is divided into three subperiods: 2004Q3–2007Q4, 2008Q1–2014Q4 and

2015Q1-2019Q1. Second, we divide the sample into firms listed on the SZSE and GEM. Third, we

investigate whether including industry effects influence our previous results. Finally, we consider the

use of QFII firms dummy and domestic institutional firms dummy to capture their effects on liquidity.

7.1. Influence of the market shock

China's financial reform process is based on its own characteristics drawing lessons from foreign

experiences and adjusting financial regulatory framework from separate mode to functional mode, thus

enhance macro prudential supervision and regulatory coordination more effectively. Modern

supervision concept is adopted, constantly complete market and product construction so as to have

more coordinated efficient financial system.

Table 8

Influence of the market shock

The table shows panel regressions of the Illiquidity (Illiq) and the trading volume (TV) on the QFII

ownership. The period of study is from 2004Q3–2009Q1. The sample 1 period is from 2004Q3 to 2007Q4.

The sample 2 period is from 2008Q1 to 2015Q1 and sample 3 is defined from 2016Q1 to 2019Q1. Driscoll-

Kraay standard errors reported in parentheses are robust to correlation across residuals within a firm over

time and across firms in the same quarter and different years.

Dependent variables

2004Q3-2008Q4 2009Q1-2015Q4 2016Q1-2019Q1

Independent

variables

(1)

Illiq

(2)

TV

(3)

Illiq

(4)

TV

(5)

Illiq

(6)

TV

QFIIi,t-1 -0.0032

(0.0063)

0.0156**

(0.0063)

0.0128*

(0.0069)

-0.0123*

(0.0066)

-0.0228***

(0.0075)

0.0248***

(0.0066)

DIi,t-1 -.4725***

(0.0948)

0.3195***

(0.0923)

-0.9199***

(0.1491)

0.7727***

(0.1247)

SIZEi,t -.1694***

(0.0209)

0.2133***

(0.0233)

-.3475***

(0.0177)

0.3145***

(0.0179)

-0.1024***

(0.0353)

0.0431**

(0.0190)

LEVi,t-1 0.1528*

(0.0838)

-0.0692

(0.0896)

-.7854***

(0.0604)

0.9211***

(0.0667)

-0.0964

(0.1093)

0.1014

(0.0810)

VOLi,t 3.4516**

(1.3420)

-1.6049*

(0.9085)

0.7778**

(0.3146)

0.8626**

(0.3787)

23.3099***

(0.6993)

12.1785***

(0.7334)

NTi,t-1 0.0031***

(0.0005)

-.0036***

(0.0005)

0.0087***

(0.0007)

-.0133***

(0.0007)

0.0248***

(0.0044)

-0.0294***

(0.0067)

LNPi,t -.7110***

(0.0150)

0.5880***

(0.0137)

-.5570***

(0.0140)

0.5700***

(0.0145)

-0.5719***

(0.0312)

0.5209***

(0.0290)

(Continued)


19

TOi,t -.2233***

(0.0150)

0.2926***

(0.0161)

-.1012***

(0.0037)

0.1586***

(0.0046)

-0.1564***

(0.0080)

0.1670***

(0.0080)

State Yes Yes Yes Yes Yes Yes

N 11078 11078 42860 42860 10716 10716

adj. R2 0.815 0.872 0.538 0.704 0.481 0.674

F 1423.6520 2073.7114 1858.7740 3670.5704 234.1961 552.7097


Like many developing countries, China has enjoyed a booming stock market. However, the

Chinese stock market was not immune to the Global Financial Crisis of 2007–2008 and the market

shock in 2015. The stock market in China essentially crashed. Furthermore, the total value of foreign

institutional holdings dropped dramatically from the beginning of 2008 until the end of that year, before

again increasing to pre-crisis levels in the second half of 2009. Based on the changing behavior of

foreign institutions at the beginning of 2008, presumably caused by the GFC, we perform a subperiod

analysis using the panel regression in the main regression to investigate whether the link between QFII

ownership and liquidity has changed over time, by dividing the sample into two subsamples: sample

1, defined as the period 2004Q3–2007Q4, sample 2, defined as the period 2008Q1–2015Q4 and

sample 3, defined as the period 2016Q1-2019Q1. The results are reported in Table 8.

We find that the association between the participation of QFIIs and Illiquidity remains strong

before 2008 and after 2015, yielding a negative link between ownership and Illiquidity and a positive

link between ownership and trading volume. The coefficients for the Illiquidity and trading volume are

much more statistically significant and greater than them before 2008, which implies an more and more

important role QFII played in recent market-oriented reform. For domestic institutions, the subperiod

results are again consistent with the full sample results.

7.2. Industry effects

We estimate the model in the main regression (Table 3) controlling for industry effects by

including industry fixed effect dummies that are equal to 1 if the firm operates in a given industry. The

results show that controlling for industry effects does not alter the association between QFII ownership

and liquidity. The results are shown in Table 9.

Table 9

Industry effects

The table shows the results from regressions including industry fixed effects. Based on the CSRC

classification, there are in total 13 industries. The period of study is from 2004Q3 to 2019Q1. Driscoll-

Kraay standard errors reported in parentheses are robust to correlation across residuals within a firm over

time and across firms in the different quarters. There are 65654 observations in each regression.

Dependent variables

Independent

variables

(1)

Illiq

(2)

Illiq

(3)

TV

(4)

TV

QFIIi,t-1 -0.0086*

(0.0049)

-0.0113***

(0.0031)

0.0140***

(0.0053)

0.0121**

(0.0051)

(Continued)


20

SIZEi,t-1 -0.3205***

(0.0121)

-0.3938***

(0.0039)

0.2914***

(0.0115)

0.2892***

(0.0113)

LEVi,t-1 -0.5435***

(0.0447)

-0.6973***

(0.0186)

0.6322***

(0.0462)

0.6174***

(0.0452)

VOLi,t 1.1395***

(0.3936)

1.0670***

(0.0637)

0.8680**

(0.3438)

0.8382**

(0.3346)

NTi,t-1 0.0081***

(0.0005)

0.0079***

(0.0003)

-0.0105***

(0.0005)

-0.0106***

(0.0005)

LNPi,t -0.5538***

(0.0112)

-0.4451***

(0.0054)

0.5193***

(0.0113)

0.4699***

(0.0119)

TOi,t -0.1058***

(0.0032)

-0.1033***

(0.0011)

0.1650***

(0.0041)

0.1695***

(0.0042)

DIi,t-1

-1.0657***

(0.0395)

1.1507***

(0.0910)



Industry dummy Yes Yes Yes Yes

N 64654 64654 64654 64654

adj. R2 0.789 0.801 0.803

F 2006.7448 2199.0699 2184.8097


7.3. The Shanghai and the Shenzhen stock exchanges

We investigate whether the positive relationship between QFII participation and liquidity is

prevalent in both the SZSE and the GEM. We therefore perform a panel regression analysis of the

model in the main regression (Table 3) by splitting the total sample into two subsamples of stocks listed

on the respective exchanges. Table 10 (a) shows that the coefficient of QFII ownership for the

Illiquidity（trading volume）is negative(positive) and consistent on both stock exchanges. We find that

the estimates are significant only in SZSE. One possible reason for this difference is induced by distinct

inherent factors in the two markets that govern this investment behavior. For example, SHSE attracts

more large firms and state-owned firms, SZSE attracts more small and medium-sized firms, while

GEM attracts more small-sized and growing-up firms.

Table 10 (a)

Different Stock Exchange

The table shows panel regressions of the Illiquidity (Illiq) and the trading volume (TV) on the QFII

ownership estimated for the Shenzhen Stock Exchange (SZSE) and Growth Enterprises Market (GEM)

separately. The period of study is from 2004Q3 to 2019Q1. Driscoll-Kraay standard errors reported in

parentheses are robust to correlation across residuals within a firm over time and across firms in the same

quarter and different quarters. There are 30199 and 6789 observations in each regression for the SHSE and

SZSE, respectively.

Dependent variables


21

SZSE GEM

Independent

variables

(1)

Illiq

(2)

TV

(3)

Illiq

(4)

TV

QFIIi,t-1 -0.0175**

(0.0072)

0.0196***

(0.0073)

-0.0049

(0.0077)

0.0143

(0.0089)

DIi,t-1 -0.9952***

(0.1112)

1.3600***

(0.1200)

-0.3498*

(0.1913)

0.3252*

(0.1785)

SIZEi,t-1 -0.3174***

(0.0185)

0.2732***

(0.0172)

-0.3805***

(0.0361)

0.3402***

(0.0359)

LEVi,t-1 -0.5515***

(0.0601)

0.5972***

(0.0612)

-0.8397***

(0.1305)

0.7340***

(0.1263)

OLi,t 0.6800**

(0.2649)

0.4562**

(0.1991)

16.1430***

(1.4879)

6.0802***

(1.8683)

NTi,t-1 0.0074***

(0.0007)

-0.0098***

(0.0006)

0.0021

(0.0015)

-0.0064***

(0.0013)

LNPi,t -0.4838***

(0.0163)

0.4404***

(0.0168)

-0.3645***

(0.0239)

0.3533***

(0.0217)

TOi,t -0.1001***

(0.0047)

0.1584***

(0.0059)

-0.1175***

(0.0067)

0.1169***

(0.0063)



N 30199 30199 6789 6789

adj. R2 0.775 0.792 0.666 0.787

F 1661.4011 1850.5718 419.9410 640.4883


7.4. QFII dummy variable effects

Table 2(b) indicates that about 10% of the whole sample comprises QFII firms while the other

90% are non-QFII firms. Our analysis thus far has been based on the whole sample. There is the

possibility that the small percentage of QFII firms in the sample might bias the slope coefficient

estimation. For robustness check, we reestimate the models in Tables 3 and 4 using a dummy variable

for QFIIs that takes the value 1 if a firm has QFIIs’ participation. The results reported in Table 11 are

related to the Illiquidity and trading volume. The results significant and consistent with the findings

reported in Table 3 and 4. QFIIs’ participation improves stock liquidity.

These results further support our findings and, if at all, strengthen the evidence that greater foreign

institutional participation is positively associated with greater stock market liquidity. For a necessary

condition, it requires further analysis of the difference in liquidity before and after a firm received QFII

investment. However, this is outside the scope of our study and is left for future research.

Table 10 (b)

Different firm size in SZSE

The table shows the panel regressions of the Illiquidity (Illiq) and the trading volume (TV) on the QFII


22

ownership in SZSE divided by the medium of firm size. The period of study is from 2004Q3 to 2019Q1.

Driscoll-Kraay standard errors reported in parentheses are robust to correlation across residuals within a

firm over time and across firms in the same year and different quarters.

Dependent variables

Large firms Small firms

Independent

variables

(1)

Illiq

(2)

TV

(3)

Illiq

(4)

TV

QFIIi,t-1 -0.0086

(0.0080)

0.0138*

(0.0078)

-0.0340**

(0.0169)

0.0409**

(0.0162)

DIi,t-1 -0.9029***

(0.1376)

1.2037***

(0.1542)

-0.4404**

(0.1724)

0.8790***

(0.1769)

SIZE -0.3928***

(0.0302)

0.3452***

(0.0307)

-0.2787***

(0.0306)

0.2584***

(0.0286)

LEVi,t-1 -0.4369***

(0.0950)

0.4970***

(0.0998)

-0.5917***

(0.0832)

0.6607***

(0.0860)

VOLi,t 0.8088

(0.5269)

0.0047

(0.2193)

0.7767***

(0.2923)

0.6882***

(0.2245)

NTi,t-1 0.0057***

(0.0010)

-0.0103***

(0.0010)

0.0081***

(0.0010)

-0.0098***

(0.0008)

LNPi,t -0.5147***

(0.0238)

0.4674***

(0.0259)

-0.4857***

(0.0232)

0.4157***

(0.0222)

TOi,t -0.1247***

(0.0112)

0.2050***

(0.0155)

-0.0947***

(0.0055)

0.1467***

(0.0065)



N 14349 14349 15850 15850

adj. R2 0.761 0.790 0.749 0.783

F . . 840.9017 898.1890


Table 11

QFII relationship with liquidity using interaction with dummy variables

Panel regression results using the three measures of liquidity regressed on lagged value of QFIIDUM, with

a dummy variable equal to 1 if a firm has QFII participation. The two measures of liquidity are the Illiquidity

(Illiq) and the trading volume (TV). The remaining control variables are the same as the ones used in Table

3 and 4. The period of study is from 2004Q3 to 2019Q1. Driscoll-Kraay standard errors reported in

parentheses are robust to correlation across residuals within a firm over time and across firms in the same

year and different firms. There are 64654 observations in each regression.

Dependent variables

Independent

variables

(1)

Illiq

(2)

TV


23

QFIIDUMi,t-1 -0.0175**

(0.0086)

0.0218**

(0.0085)

DIi,t-1 -0.8707***

(0.0399)

1.2136***

(0.0394)

SIZEi,t-1 -0.3122***

(0.0044)

0.2786***

(0.0044)

LEVi,t -0.5211***

(0.0205)

0.6060***

(0.0202)

VOLi,t 1.1185***

(0.0620)

0.8370***

(0.0613)

NTi,t-1 0.0078***

(0.0003)

-0.0101***

(0.0002)

LNPi,t -0.5227***

(0.0056)

0.4738***

(0.0055)

TOi,t -0.1073***

(0.0011)

0.1677***

(0.0011)

State Yes Yes

Time dummy Yes Yes

N 64654 64654

adj. R2 0.786 0.798

F 10002.0131 10776.3372


8. Summary and concluding remarks

With the continuous opening of the capital market and the entry of foreign capital into the stock

market, there are many controversies about its influence, which has become a hot topic in the academia.

This paper employs a unique setting for the limited participation of qualified foreign institutional

investors (QFIIs) in China`s A-share market and examines how these impacts on stock liquidity in

emerging markets.

Contrary to the findings in the literature, our results reveal that greater foreign institutional

participation is positively associated with stock market liquidity. This positive relationship operates

mainly through promoting trade activities by increasing trading volume. We also find that there exist

heterogeneity in the effect of the QFII ownership on different size of firms. The improvement in

liquidity is more significant in small firms compared to large firms. Our findings are robust to

endogeneity and the possible influence of the stock market shock, industry effects and the stock

exchange. Further, the liquidity improving effects of QFII are even stronger when the analysis is

performed on a subsample of QFII firms.

References

[1] Amihud Y , Mendelson H , Pedersen L H . Market Liquidity: Illiquidity and Stock

Returns:Cross-Section and Time-Series Effects[J]. Journal of Financial Markets, 2002,

5(1):31-56.


24

[2] Bae K H , Bailey W , Mao C X . Stock market liberalization and the information

environment[J]. Journal of International Money and Finance, 2006, 25(3):0-428.

[3] Ding M , Nilsson B , Suardi S . Foreign Institutional Investment, Ownership, and Liquidity:

Real and Informational Frictions[J]. Financial Review, 2017, 52(1):101-144.

[4] Froot K A , O'Connell P G J , Seasholes M S . The Portfolio Flows of International Investors,

I[J]. Journal of Financial Economics, 2001, 59(2):151-193.\

[5] Goyenko R , Holden C W , Trzcinka C . Do Liquidity Measures Measure Liquidity? [J].

Social Science Electronic Publishing, 2008, 92(2):153-181.

[6] Grossman S J . On the Impossibility of Informationally Efficient Markets[J]. American

Economic Review, 1980, 70(3).

[7] Heflin F , Shaw K W . Adverse Selection, Inventory-Holding Costs, and Depth[J]. Journal

of Financial Research, 2001, 24(1):65-82.

[8] Mendelson H , Tunca T I . Strategic Trading, Liquidity, and Information Acquisition[J].

Review of Financial Studies, 2004, 17(2):295-337.

[9] Opler T, Pinkowitz L, Stulz R, et al. The determinants and implications of corporate cash

holdings[J]. Nber Working Papers, 1999, 52(1):3-46.

[10] Subrahmanyam, A. Risk Aversion, Market Liquidity, and Price Efficiency[J]. Review of

Financial Studies, 1991, 4(3):417-441.

[11] 孙立, 林丽. QFII 投资中国内地证券市场的实证分析[J]. 金融研究, 2006(7):123-133.

[12] 童元松, 王光伟. 境外机构投资者持股、公司业绩与股市质量[J]. 会计与经济研究,

2015(6):101-110.

foreign institutional ownership and stock liquidity in...

Documents