Chia 2020

Emerging Markets Review 44 (2020) 100696
Contents lists available at ScienceDirect
Emerging Markets Review

journal homepage: www.elsevier.com/locate/emr
More shareholders, higher liquidity? Evidence from an emerging

T
stock market
Yee-Ee Chiaa,b, Kian-Ping Limb, , Kim-Leng Gohc
⁎
a
Labuan Faculty of International Finance, Universiti Malaysia Sabah, F.T. Labuan, Sabah, Malaysia
b
Department of Economics, Faculty of Economics & Administration, University of Malaya, Kuala Lumpur, Malaysia
c
Department of Applied Statistics, Faculty of Economics & Administration, University of Malaya, Kuala Lumpur, Malaysia
ARTICLE INFO ABSTRACT
Keywords: Using data assembled from all non-financial firms traded on the Malaysian stock exchange, we
Shareholder base provide evidence of a nonlinear relationship between the number of shareholders and liquidity.
Stock liquidity While more shareholders are associated with higher liquidity, the negative effect of wider spreads
Ownership dispersion kicks in when shareholder base exceeds a threshold level due to higher volatility induced by noise
PIN
trading. However, the threshold level is considerably higher than the number of shareholders of
Nonlinearity
Malaysia
most Malaysian public listed firms, suggesting much room for shareholder expansion in the local
market. Our findings call for corporate managers to actively manage and expand their share-
JEL classifications: holder bases.
G10
G18
G32
G38
1. Introduction
In his seminal paper, Merton (1987) shows theoretically that a larger shareholder base is associated with lower cost of capital and
higher firm value. Existing empirical studies suggest the benefits of greater breadth of ownership also extend to liquidity. Demsetz
(1968), Benston and Hagerman (1974) and Jacoby and Zheng (2010) provide direct evidence on the positive relationship between
the number of shareholders and stock liquidity. Subsequent academic studies largely omit shareholder base when specifying their
liquidity model in favour of percentage share ownership variables, which could be due to the lack of quality data in commercial
databases. For instance, with the availability of shareholder information in Thomson Datastream, the literature witnesses the
emergence of cross-country liquidity studies using the number of free-float shares (Ding et al., 2016) and foreign blockholdings (Ng
et al., 2016; Lee and Chung, 2018), but none on the number of shareholders. Such omission should be rectified as their association has
⁎
Corresponding author at: Department of Economics, Faculty of Economics & Administration, University of Malaya, 50603 Kuala Lumpur,
Malaysia.
E-mail addresses: chiayeeee@ums.edu.my (Y.-E. Chia), kianpinglim@um.edu.my (K.-P. Lim), klgoh@um.edu.my (K.-L. Goh).
https://doi.org/10.1016/j.ememar.2020.100696
Received 5 January 2019; Received in revised form 10 January 2020; Accepted 2 April 2020
Available online 08 April 2020
1566-0141/ © 2020 Elsevier B.V. All rights reserved.
Y.-E. Chia, et al. Emerging Markets Review 44 (2020) 100696
been implied in the literature of investor recognition.1 Adding to that, shareholder-boosting measures are gaining attention in
academia and industry. Citing previous empirical work, Amihud and Mendelson (2000, 2008) repeatedly advocate expanding
shareholder base as an effective strategy to enhance stock liquidity.2 Anecdotal evidence also indicates that public listed companies in
developed markets actively strategize to expand their shareholder bases or to target specific types of investors they desire.3
To fill the void in the existing empirical literature, we provide an in-depth analysis on the relationship between shareholder base
and liquidity using firm-level data from the emerging stock market of Malaysia. The pioneering work of Demsetz (1968) explores the
determinants of bid-ask spread for stocks traded on the New York Stock Exchange, and reports the coefficient for number of
shareholders is negative and highly significant because more shareholders increase the probability of bidding by market participants
and hence greater time rate of transactions. The above negative association between number of shareholders and bid-ask spread is
reaffirmed by Benston and Hagerman (1974) who replicate the same analysis on stocks traded in the over-the-counter market.
Motivated by the limitation of blockholdings as measurement of ownership dispersion, Jacoby and Zheng (2010) advocate the
number of shareholders and find that their newly proposed variable is negatively and significantly associated with bid-ask spread
using U.S. cross-sectional data in year 1995. Notably, the above three studies do not elaborate on the underlying theories, hence it is
unclear why the relationship between shareholder base and stock liquidity has to be linear. Amihud and Mendelson (2000, 2008)
caution that maximizing the number of shareholders is not necessarily an optimal strategy. This is because some strategies require
substantial managerial time and monetary resources, which are quite costly for small and young public firms in emerging markets.
Furthermore, agency costs tend to increase when ownership becomes more dispersed, thereby weakening incentives of monitoring,
leading to higher likelihood of free-rider problem and managerial entrenchment (Jensen and Meckling, 1976; Grossman and Hart,
1980).4 Such environment of weak internal and external governance is found to be detrimental to liquidity (Chung et al., 2010; Jain
et al., 2016). Furthermore, a larger shareholder base might yield wider bid-ask spread either due to greater information asymmetry or
higher stock volatility. If more informed investors are added to the shareholder base, such expansion exacerbates information
asymmetry, increases adverse selection costs and leads to lower liquidity. In the case where more noise traders are involved, liquidity
will decline if their noise trading generates excessive volatility. Given the possibility of negative liquidity effect when ownership
becomes more dispersed, it is thus reasonable to expect an optimal level of shareholder base.
While the limited studies on U.S. find a positive relationship between shareholder base and liquidity (Demsetz, 1968; Benston and
Hagerman, 1974; Jacoby and Zheng, 2010), we argue that it is unreasonable to expect such finding to prevail in emerging stock
markets for three reasons. First, unlike their counterparts in developed markets, corporate managers of emerging market firms invest
less resources in expanding their investor clientele, which is evidenced by the persistence of ownership concentration over time
(Claessens et al., 2000; Carney and Child, 2013). Instead, it is the stock exchange operators and regulators who actively promote
stock market participation and shape liquidity through regulations and incentives. Second, it is well acknowledged that emerging
market firms suffer from lower levels of liquidity relative to firms in developed markets (see Lesmond, 2005; Griffin et al., 2010; Kang
and Zhang, 2014). It is unclear whether the expansion of shareholder base would be effective in such environment characterized by
high trading frictions. For instance, recent evidence from the emerging China stock markets by Yung and Jian (2017) reports a
negative relationship between the number of shareholders and firm value which they attribute to elevated agency conflicts between
the controlling shareholders and dispersed small investors. Third, the theoretical channels through which shareholder base improves
liquidity – the degree of investor recognition and information competition among informed investors – might not be dominant in
emerging markets. In the first instance, Kaniel et al. (2012) find the high volume return premium associated with changes in investor
recognition across 20 developed markets is pervasive, but less persistent in their sample of 21 emerging markets. Morck et al. (2000),
on the other hand, postulate that weak private property rights protection in emerging markets deters the participation of informed
investors, leaving the markets to uninformed noise traders.
Motivated by its differences in institutional features from developed markets, this paper re-examines the relationship between
1
For instance, Kadlec and McConnell (1994) find that firms switching their listing to New York Stock Exchange are associated with increases in
the number of registered shareholders and reductions in bid-ask spread, which they attribute to enhanced degree of investor recognition. Grullon
et al. (2004) show that higher advertising expenditures are associated with larger number of both individual and institutional investors, lower bid-
ask spread, smaller price impact and larger quoted depth. This is because product market advertising increases a firm's visibility and attracts stock
investors whose decisions are driven primarily by familiarity. Green and Jame (2013) find that public firms with more fluent names are associated
with more shareholders and improved liquidity. Using Google search volume index, Ding and Hou (2015) find that active attention from retail
investors broadens shareholder base and reduces bid-ask spread.
2
Corporate practices that have been proven effective in boosting shareholder base and improving liquidity in developed stock markets include the
reduction in lot size (Amihud et al., 1999; Ahn et al., 2014), stock splits (Mukherji et al., 1997; Li et al., 2017b, noncash shareholder perks (Karpoff
et al., 2018), listing on major stock exchanges (Kadlec and McConnell, 1994; King and Segal, 2009), strategic corporate disclosures (Bushee and Noe,
2000; Bushee, 2004), effective investor relations programs (Bushee and Miller, 2012; Karolyi and Liao, 2017), increases in company name fluency
(Green and Jame, 2013), addition to stock indices (Chen et al., 2004) and higher levels of advertising expenditures (Grullon et al., 2004).
3
According to Parrino et al. (2003), the presence of consulting firms who specialize in shareholder composition proves that corporate managers do
strategize to attract certain investor types. A survey by Stanford's Rock Center for Corporate Governance and the National Investor Relations
Institute reaffirms that managers of most U.S. corporations recognize the tangible stock market benefits and thus dedicate considerable time to
manage their diverse shareholder bases (see Beyer et al., 2014).
4
The motivation for underpricing initial public offerings in order to boost the size and alter the composition of shareholder base is also con-
strained by similar trade-off between higher liquidity and effective monitoring (see, for example, Booth and Chua, 1996; Brennan and Franks, 1997;
Pham et al., 2003; Zheng and Li, 2008).
2
shareholder base and liquidity using Malaysia as a representative case study. Similar to other emerging markets, the exchange
regulator (Securities Commission Malaysia) and operator (Bursa Malaysia) are the key drivers of investor participation and liquidity
in the Malaysian stock market, placing greater emphasis on boosting retail participation in small- and mid-sized listed firms.5 The
selection of Malaysia is driven by data availability on the number of shareholders.6 Our Malaysian data are of higher accuracy than
the number of shareholders provided by COMPUSTAT (see Bodnaruk and Östberg, 2013; Chang et al., 2013; Chichernea et al., 2015)
or CDA/Spectrum databases (Lehavy and Sloan, 2008; Richardson et al., 2012). The former only provides the approximate number of
shareholders derived from firms' 10-K filings with the Securities and Exchange Commission, whereas the latter covers U.S. institutions
filing Form 13F (for details, see Bodnaruk and Östberg, 2009; Anchev, 2017).
To test our hypothesis that the relationship between shareholder base and liquidity is nonlinear, we collect data for all non-
financial firms traded on Bursa Malaysia over the 16-year sample period of 2000–2015. The liquidity model regresses Closing Percent
Quoted Spread (CPQS) against the number of shareholders and a set of standard control variables. Our evidence of a U-shaped
relationship suggests at levels of shareholder base before the tipping point, a larger number of shareholders is associated with lower
spread and thus higher liquidity. However, when the shareholder base exceeds a certain threshold level and becomes too large, the
negative effect dominates and causes liquidity to drop. It is worth highlighting that the threshold level is considerably high compared
to the number of shareholders of most Malaysian public listed firms. About 95% of firm-year observations fall below the threshold
number of shareholders. This suggests shareholder expansion and liquidity enhancing goals cannot be achieved by stock exchange
alone, but requires concerted efforts of corporate managers to actively manage and expand their shareholder bases. The nonlinear
relationship passes a series of robustness checks – alternative measure of shareholder base, alternative estimation methods, dis-
aggregate investor types, endogeneity tests, exclusion of the crisis years of 2008–2009 and industry-specific regressions. The richness
of our data allows further analyses to explore why liquidity declines when the number of shareholders becomes too large, which we
find is attributed to higher volatility induced by noise trading. Our findings also provide empirical support to the policy focus of
Malaysian authorities to boost retail participation, but more efforts are needed to expand the size of local individual investors in the
Malaysian stock market given that the number and percentage of local individual account holders are far below the threshold levels.
The capital market effects of larger shareholder base have not been explored rigorously until recently, and we contribute to this
limited literature. Most of the earlier studies explore the relationship between number of shareholders and cross-section of stock
returns within the framework of investor recognition (see Lehavy and Sloan, 2008; Bodnaruk and Östberg, 2009; Richardson et al.,
2012; Chang et al., 2013; Choi et al., 2013; Anchev, 2017). The focus of recent research, however, has shifted to explore the effects of
broader shareholder base on payout policy (Bodnaruk and Östberg, 2013), stock volatility (Chichernea et al., 2015; Jankensgård and
Vilhelmsson, 2018) and firm value (Yung and Jian, 2017; Karpoff et al., 2018). Our contribution lies in establishing the relationship
between shareholder base and liquidity, in particularly showing that liquidity declines when the number of shareholders exceeds the
threshold level due to the often neglected volatility channel induced by noise trading. Such threshold analysis is useful not only for
public listed firms but also for policy purposes because exchange regulators always strive to promote greater stock market partici-
pation and higher liquidity. We note that some exchanges explicitly include the minimum number of shareholders in their listing
requirements, perhaps to ensure adequate liquidity for listed stocks.7 Our analysis thus complements the exchange-determined lower
bound by prescribing an upper bound for shareholder base that public listed firms might strive to attain.
We also contribute to the literature on ownership structure and stock liquidity, which at present is dominated by the use of
percentage share ownership.8 The main drawback is such computation of ownership uses only the number of outstanding shares, and
thus ignores the number of shareholders that better reflects ownership dispersion (Jacoby and Zheng, 2010; Choi et al., 2013). For
instance, firms with a high percentage of retail ownership might not realize the expected liquidity improvement simply because the
shares are concentrated in the hands of a few large wealthy individuals who possess information advantage (see Chen et al., 2015; Li
et al., 2017a). Given the possibility of such distortion, some authors advocate the use of number of institutional investors or the
number of blockholders.9 We establish the theoretical grounds and empirical relevance for including the number of shareholders in a
5
Among the measures undertaken over the past two decades include the reduction of lot size from 1000 to 100 units (2003), the sponsorship of
CMDF-Bursa Research Scheme (2005) and Mid and Small Cap Research Scheme (2017), the publication of investor relations manual (2007) and
corporate disclosure guide (2012) for listed companies, further liberalization of foreign ownership (2009) and margin financing rules (2018), the
launching of the community online portal of Bursa Marketplace (2014), the introduction and continuous revisions of Malaysian Code of Corporate
Governance (2000, 2007, 2012, 2017), and the removal of intraday short-selling restrictions for all investors (2018).
6
Similar type of shareholdings data is available for Sweden (see Bodnaruk and Östberg, 2009; Anchev, 2017; Jankensgård and Vilhelmsson,
2018). The dataset, assembled by Swedish central securities depository known as Euroclear Sweden (formerly Värdepapperscentralen, VPC),
contains detailed information on all investors who own shares and the number of shares they own in public listed Swedish companies.
7
Browsing through the websites of selected stock exchanges, we find that the minimum number of shareholders is included in their listing
requirements, for example, by Bursa Malaysia (1000 shareholders), Indonesia Stock Exchange (1000 shareholders for Main Board, and 500 for
Development Board), Singapore Exchange (500 shareholders), Stock Exchange of Thailand (1000 shareholders), Nasdaq Stock Market and New York
Stock Exchange (both 2200 shareholders).
8
Among the significant determinants of liquidity are institutional ownership (Rubin, 2007; Jiang et al., 2011; Dang et al., 2018), blockholder
ownership (Heflin and Shaw, 2000; Attig et al., 2006; Brockman et al., 2009), government ownership (Choi et al., 2010; Boubakri et al., 2019; Ding
and Suardi, 2019), foreign ownership (Rhee and Wang, 2009; Ng et al., 2016; Lee and Chung, 2018), insider ownership (Chiang and Venkatesh,
1988; Kini and Mian, 1995; Rubin, 2007) and ownership concentration (Byun et al., 2011; Rubin, 2007; Leaño and Pedraza, 2018).
9
For instance, Sias et al. (2001) find that contemporaneous stock returns are more strongly associated with the number of institutional investors
than institutional ownership due to the price impact of informed trading. Similarly, Blume and Keim (2012) demonstrate that the number of
3
liquidity model to formalize their association often implied in the literature of investor recognition, and lend empirical support to the
growing use of shareholder-boosting corporate strategies. Notably, the inclusion of ownership variables in our model does not reduce
the statistical and economic significance of shareholder base, suggesting the number of shareholders represents distinct dimension of
shareholdings that should not be ignored in future studies. The incremental explanatory power and large economic effect of
shareholder base complement recent discovery of new liquidity determinants, such as annual report readability (Boubaker et al.,
2019), intensity of foreign competition (Atawnah et al., 2018), better access to crowdsourced retail-oriented investment research
(Farrell et al., 2018), investor risk perception (Ma et al., 2019) and television distraction (Peress and Schmidt, 2020).
The remainder of the paper is structured as follows. Section 2 discusses the theories that motivate our hypothesis of a nonlinear
relationship between shareholder base and liquidity. The variables and model specification are presented in Section 3. Subsequently,
we describe the data collection process and provide a preliminary overview of the final data. Section 5 discusses the baseline results
for the liquidity model and their robustness checks. Further analyses on the data are conducted in Section 6. To conclude, we
summarize the key findings and prescribe recommendations.
2. Theoretical basis for hypothesis development
Previous studies on the relationship between shareholder base and liquidity is limited and confined to the U.S. stock markets
(Demsetz, 1968; Benston and Hagerman, 1974; Jacoby and Zheng, 2010). Though their evidence of a positive relationship is con-
sistent with the widely held view among corporate managers and policymakers that more shareholders are associated with higher
liquidity, these authors do not elaborate on the underlying theories so as to shed lights on the channels linking shareholder base to
liquidity.
Unlike the above-cited U.S. studies, this paper hypothesizes the existence of a nonlinear relationship between shareholder base
and stock liquidity. To motivate our research hypothesis, we draw from different strands of literature to outline the plausible the-
oretical channels. More specifically, the existence of a threshold implied in a nonlinear relationship between number of shareholders
and liquidity suggests the relationship is not determined solely by the size of the shareholder base, but also shaped by its composition
of informed versus noise traders. Such dichotomy gives rise to the countervailing effects predicted by different strands of theoretical
models, namely investor recognition, information competition, adverse selection costs and noise trading.
2.1. Investor recognition
The theoretical model of Merton (1987) focuses on the central role of investor recognition in the determination of expected stock
returns and firm value. With the assumption that investors consider only a subset of all public listed firms in their investment choices,
Merton (1987) characterizes the degree of investor recognition as the number of outside investors who know about the firm. Even
though the model is silent on liquidity, empirical studies testing Merton's (1987) theoretical prediction find that enhanced investor
recognition – due to exchange switching (Kadlec and McConnell, 1994), effective advertising (Grullon et al., 2004), company name
fluency (Green and Jame, 2013) and reduction in minimum trading unit (Amihud et al., 1999) – generally leads to larger shareholder
base and higher stock liquidity. His model thus provides the theoretical basis for subsequent empirical work to proxy investor
recognition with the number of shareholders, especially in explaining the cross-section stock returns.10
2.2. Information competition
While larger shareholder base reflects enhanced investor recognition and hence improved liquidity, the positive relationship
might also operate through information competition among informed investors. Existing strategic trader models predict that in-
creased competition among multiple privately informed traders who act strategically causes the private information they possess to
be swiftly incorporated into stock prices and the market becomes more information efficient, thereby lowering the extent of in-
formation asymmetry and increases liquidity (see Subrahmanyam, 1991; Spiegel and Subrahmanyam, 1992; Holden and
Subrahmanyam, 1992; Foster and Viswanathan, 1994). Empirically, Akins et al. (2012) show that greater competition is associated
with lower pricing of information asymmetry because there is little room for informed traders to exploit their private information.
Consistent with extant theoretical models, these authors measure the extent of information competition with the number of informed
traders.11 In the liquidity literature, the largely ignored information competition channel is first explored by Agarwal (2007) who
(footnote continued)
institutional investors has a higher explanatory power relative to institutional shareholdings for the cross-section variation of liquidity. In the
theoretical model of Edmans and Manso (2011), the number of blockholders plays a key role in driving market efficiency and corporate governance.
Empirically, there is a rich literature on the capital market effects of having multiple large shareholders (see Maury and Pajuste, 2005; Laeven and
Levin, 2008; Attig et al., 2009), whereas the use of blockholder numbers is slowly gaining acceptance (see Bharath et al., 2013; Alvarez et al., 2018;
Benamraoui et al., 2019).
10
However, there is no consensus on the investor types. For instance, Lehavy and Sloan (2008) and Richardson et al. (2012) use the number of
institutional investors and confirm the importance of investor recognition in accounting for stock return variation. There are also studies using the
number of individual investors as proxy for investor recognition (Chang et al., 2013), but the total number of shareholders is closer to the original
theoretical exposition (see Bodnaruk and Östberg, 2009; Chichernea et al., 2015; Yung and Jian, 2017; Jankensgård and Vilhelmsson, 2018).
11
In recent years, the number of shareholders has been adopted as a proxy for information competition, where a higher value indicates greater
4
uses the number of local institutions as a proxy, and finds that liquidity is an increasing function of the number of informed investors.
2.3. Adverse selection costs
Contrary to the prediction of strategic trader models, the literature has long associated informed traders with lower liquidity due
to the dominance of asymmetric information models (Glosten and Milgrom, 1985; Easley and O'Hara, 1987). The latter's key the-
oretical prediction is that with an increase in the number of privately informed investors who act non-strategically, the bid-ask spread
becomes wider because of higher adverse selection costs. Empirically, blockholders (Brockman and Yan, 2009; He et al., 2013) and
institutional investors (Piotroski and Roulstone, 2004; Bae et al., 2012) are reported to possess information advantage due to their
privileged access to private information. A parallel strand of literature finds both local (Heflin and Shaw, 2000; Attig et al., 2006;
Brockman et al., 2009) and foreign blockholders (Ng et al., 2016; Lee and Chung, 2018) exert detrimental effect on liquidity,
consistent with the theoretical prediction of asymmetric information models.12 The possibility of the two countervailing forces
driving the informed institutional investors-liquidity relationship is highlighted by Agarwal (2007), who finds an upward trend in
liquidity that reverses after institutional ownership reaches a threshold level of 35%–40%, suggesting the dominance of information
competition and adverse selection costs at each side of the threshold point.
2.4. Noise trading
Individual investors, on the other hand, have long been regarded as noise traders who trade for liquidity reasons unrelated to
fundamental (Foucault et al., 2011) and exhibit behavioral biases (Barber and Odean, 2000). The positive effect of noise trading on
liquidity is acknowledged by Black (1986) which market microstructure models rationalize through lower adverse selection costs (see
Glosten and Milgrom, 1985; Holmström and Tirole, 1993). The literature has largely subscribed to the view that noise trading
increases liquidity to the extent that the opposing effect is ignored. However, the possibility of the latter can be gleaned from the
empirical findings of Morck et al. (2000) that emerging markets are dominated by noise traders due to the weak private property
rights protection, yet Griffin et al. (2010) document the lower liquidity of these markets relative to developed countries. To ratio-
nalize the above puzzle, we turn to the noise trader models of De Long et al. (1990), Shleifer and Summers (1990) and Barberis et al.
(1998). A key assumption of these models is that noise traders are prone to sentiment not fully justified by fundamental information,
whose trading causes mispricing and generates excessive volatility (for empirical evidence, see Brown, 1999; Foucault et al., 2011).
The resulting higher volatility is associated, theoretically and empirically, with lower liquidity (Stoll, 1978a, 1978b, 2000; Chung and
Chuwonganant, 2014). Our hypothesized stock volatility channel is consistent with recent evidence by Jankensgård and Vilhelmsson
(2018) who find that larger shareholder base increases stock price volatility through noise trading.
3. Measurement of variables and the specification of liquidity model
This section provides a brief discussion on the proxies for liquidity, shareholder base and the selected set of control variables,
along with their respective data sources. We then specify the liquidity model for examining the nonlinear relationship between
shareholder base and liquidity. Since new variables are added in our subsequent robustness checks and further analyses, Appendix A
compiles a list of all the variables used in this study.
3.1. Dependent variable of stock liquidity
Despite the multifaceted nature of liquidity, U.S. firm-level studies typically use bid-ask spread from the Trades and Quotes (TAQ)
database because it measures the cost associated with immediate trading. Such high-frequency intraday bid-ask spread is the standard
benchmark in liquidity horseraces to assess the efficacy of low-frequency liquidity proxies (see Lesmond, 2005; Goyenko et al., 2009;
Marshall et al., 2013; Fong et al., 2017). For emerging markets, microsecond tick data are only available in recent years through
Thomson Reuters Tick History (TRTH) but at high subscription costs and they involve enormous computational processing time (see
Fong et al., 2017). Nevertheless, this is not a major obstacle as daily bid-ask spreads can be constructed and are reported to be highly
correlated with their intraday benchmarks (Corwin and Schultz, 2012; Chung and Zhang, 2014; Abdi and Ranaldo, 2017). Given the
abundance of choices, the liquidity horseraces conducted by Fong et al. (2017) covering 24,240 firms from 42 global stock exchanges
provide useful research guides since the performance of each liquidity proxy varies across exchanges. For Malaysian stocks, Fong
et al. (2017) consistently show that Closing Percent Quoted Spread (CPQS) proposed by Chung and Zhang (2014) is the best liquidity
measure in the percent-cost category, outperforming its nearest competitor by large margins at both the daily and monthly intervals.
(footnote continued)
level of competition among informed traders over private information, covering the pricing of information asymmetry (Armstrong et al., 2011),
stock price efficiency (Lim et al., 2016), cross-section of stock returns (Jiao, 2016), firm operating performance (Jiao, 2016) and stock price crash
risk (Vorst, 2017).
12
Contrary to the prediction of asymmetric information models, the empirical evidence leans toward a positive relationship between non-block
institutions and liquidity, applicable to both local (Rubin, 2007; Jiang et al., 2011; Dang et al., 2018) and foreign institutional investors (Ng et al.,
2016; Ding et al., 2017).
5
Thus, CPQS is selected as our liquidity measure, and can be written as:
Closing Aski, d Closing Bidi, d
CPQSi, d = x 100
(Closing Aski, d + Closing Bidi, d)/2 (1)
where Closing Aski, d and Closing Bidi, d are respectively the closing ask and bid prices of stock i on day d sourced from Thomson
Datastream, and the multiplication by 100 is for scaling purpose. We obtain the annual liquidity estimates for each Malaysian stock
by taking time-series averages of these daily CPQS estimates. The magnitude of CPQS is inversely related to liquidity because a higher
value indicates a wider spread and thus higher trading cost for investors.
3.2. Key independent variable of shareholder base
The lack of research in the academic literature on shareholder base can be largely attributed to data constraint. U.S. firm-level
studies are only able to obtain the approximate number of shareholders from COMPUSTAT (see Bodnaruk and Östberg, 2013; Chang
et al., 2013; Chichernea et al., 2015) or CDA/Spectrum databases (Lehavy and Sloan, 2008; Richardson et al., 2012). High quality
data on shareholder base are only used by a few individual country studies on Sweden (Bodnaruk and Östberg, 2009; Anchev, 2017;
Jankensgård and Vilhelmsson, 2018), China (Yung and Jian, 2017) and Japan (Amihud et al., 1999; Ahn et al., 2014; Karpoff et al.,
2018). Notably, there is no cross-country study to generalize the capital market effects of larger shareholder base.
Our key independent variable of shareholder base is proxied by the number of shareholders (NSH). In recent years, detailed firm-
level data on shareholder information for all Malaysian public listed firms are made available by Bursa Malaysia through its
Information Services Division. The dataset “End of Year Shareholdings by Type of Investor” provides the number of shareholders and
number of shares in each firm for seven types of investors: (1) individuals; (2) banks; (3) investment trusts; (4) other corporations; (5)
government agencies; (6) nominees; and (7) others. Each investor type is further disaggregated into Malaysian and foreign owner-
ship. Bursa Malaysia is able to compile detailed shareholdings data because all investors are required to open a Central Depository
System (CDS) account, fully owned and operated by its subsidiary Bursa Malaysia Depository (formerly known as Malaysian Central
Depository). The CDS is a computerized ledger system that acts as the central database recording ownership and movement of
securities, and hence collects accurate data on the number of shareholders for all Malaysian public listed firms. Fig. 1 provides the
cross-sectional annual averages for the number of shareholders, which depicts a clear downward trend over our 16-year sample
period that is not due to the reduction in number of outstanding shares available to the investing public. While expanding shareholder
Fig. 1. Shareholdings for Malaysian public listed firms.

Notes: The figure provides the cross-sectional annual averages for the number of shareholders across all Malaysian public listed firms over the 16-
year sample period of 2000–2015. The average number of outstanding shares available to the investing public across all firms is computed for each
calendar year.
6
base remains a challenge for the Malaysian stock exchange, this study addresses a more fundamental question on the benefit of
having a large number of shareholders.
3.3. Control variables
Even though the incorporation of Malaysian stock market dates back to 1976, Lim et al. (2017) is perhaps the first published study
to explore the underlying liquidity determinants for Malaysian stocks. Their liquidity model regresses the Amihud (2002) illiquidity
ratio on the ownership of various investor types while controlling for the number of security analysts and standard firm char-
acteristics, covering 600 Malaysian firms that survived over the 8-year sample period of 2002–2009. These authors find that not all
ownership variables are significantly associated with liquidity. The coefficient for local institutional ownership is positive and sig-
nificant only in the linear model, whereas significant nonlinear relationships are documented for local individual ownership and
foreign nominee ownership.
Nevertheless, the recent findings of Lee and Chung (2018) demonstrate that the relationship between ownership and liquidity
differs for price impact and bid-ask spread. We thus replicate the liquidity model of Lim et al. (2017) on our larger sample firms and
longer time period, using Amihud (2002) illiquidity ratio to ensure consistency with their liquidity proxy. Similar significant results
are obtained for all three ownership variables in our data – local institutional ownership, local individual ownership and its squared
term, and foreign nominee ownership and its squared term. However, when the Amihud illiquidity ratio is replaced by Closing
Percent Quoted Spread as the dependent variable, foreign nominee ownership and its squared term lose their explanatory power.
Since CPQS is the main liquidity measure, our model thus omits foreign nominee ownership and controls for only two investor
groups, namely local institutional ownership and local individual ownership.
Apart from the above percentage share ownership, our model further controls for a number of variables correlated with liquidity.
First, analyst coverage has been found to be a significant determinant of liquidity but there is no consensus on its expected sign.
Higher number of security analysts is associated with lower bid-ask spread if adverse selection dominates (Brennan and
Subrahmanyam, 1995), but Chung et al. (1995) report contradicting results because market markers set wider spreads for stocks
covered extensively by security analysts. The data for number of security analysts are extracted from Institutional Brokers Estimate
System (I/B/E/S) database. Second, the literature finds that both local (Heflin and Shaw, 2000; Attig et al., 2006; Brockman et al.,
2009) and foreign blockholders (Ng et al., 2016; Lee and Chung, 2018) reduce liquidity due to their privileged access to private
information that gives rise to higher adverse selection costs. The shareholdings of blockholders are collected from Thomson Data-
stream. Third, corporate governance emerges as an important liquidity determinant because it reduces information asymmetry
through enhanced financial and operational transparency (Chung et al., 2010; Al-Jaifi et al., 2017). We include four board char-
acteristics – board size, board independence, CEO duality and independent chairman – with the data hand-collected from annual
reports of public listed firms. Last but not least, data for four standard firm characteristics are sourced from Thomson Datastream,
namely firm size, stock returns, turnover and return volatility. There is a consensus in the literature that higher liquidity is associated
with larger firm size, better return performance, higher turnover and lower level of volatility (see, for example, Rhee and Wang,
2009; Lim et al., 2017).13
3.4. Specification of the liquidity model
By extending the liquidity model of Lim et al. (2017) for Malaysian stocks to include a new variable of shareholder base, we
specify the following model:
CPQSit = + 2 2
0 1 ln NSHit + 2 ln NSH it + 3 LINDit + 4 LIND it + 5 LINSTit
+ 6 ln (1 + ANALYST )it + 7 RETURNit + 8 VOLit + 9 TURNOVERit
+ 10 ln SIZEit + 11 BLOCK it + 12 ln BSIZEit + 13 BINDEPit
J 1 T 1
+ 14 DUALit + 15 CHAIRit + j = 1 16j
INDj + t = 1 17t
YRt + it (2)
where ln refers to natural logarithm, liquidity is proxied by the Closing Percent Quoted Spread (CPQS) and the number of share-
holders (NSH) reflects the size of shareholder base. The control variables include: (1) local individual ownership (LIND), computed as
the total shares held by local individual investors scaled by total shares outstanding; (2) local institutional ownership (LINST),
computed as the total shares held by local institutions scaled by total shares outstanding; (3) number of unique analysts issuing
earnings forecasts (ANALYST); (4) annual stock returns (RETURN) computed by taking the time series average of daily returns; (5)
stock return volatility (VOL) measured as the standard deviation of daily stock returns over the year; (6) stock turnover (TURNOVER)
represented by the annual average of daily turnover ratios, defined as the number of shares traded scaled by the number of shares
outstanding; (7) firm size (SIZE) measured by book value of total assets at the end of the calendar year; (8) blockholder ownership
(BLOCK), defined as the shareholdings of investors with at least 5% of outstanding shares; (9) total number of directors on the board
13
Larger firms exhibit greater visibility and investor recognition, as wider dissemination of their information alleviates the extent of information
asymmetry. Equally, well performing stocks tend to attract the attention of investors. Stocks with higher turnover provide opportunities for market
makers to manage their inventory and recoup losses from informed investors. Higher volatility, however, is associated with lower liquidity because
the former reflects greater uncertainty and higher inventory costs.
7
(BSIZE); (10) proportion of independent non-executive directors (BINDEP); (11) CEO duality (DUAL), which takes a value of one if the
positions of board chairman and chief executive officer are held by the same individual, and zero otherwise; (12) CHAIR that takes a
value of one if the board chairman is an independent non-executive director, and zero otherwise. INDj is a set of industry-specific
dummy variables to control for time-invariant industry effects, where INDj = 1if firm i is in industry j and 0 otherwise, and J is the
number of industries. Year dummies YRt are included to control for common shocks, where YRt = 1 for year t and 0 otherwise, and T
is the number of years. The predicted nonlinear relationship between the number of shareholders (NSH) and liquidity (CPQS) requires
statistically significant coefficients with opposite signs for γ1 and γ2. The liquidity model is estimated with pooled Ordinary Least
Squares (OLS), where possible bias due to within-cluster correlation is accounted by double-clustered standard errors (Petersen,
2009; Gow et al., 2010).
4. The data
This section describes how we construct the list of all Malaysian public listed firms for data collection. Summary statistics and
correlation matrix are provided to give a preliminary view of the assembled panel data.
4.1. Data collection
The main challenge when extracting data from Thomson Datastream is that the database only covers stocks that are still active at
the point of retrieval. For stocks that have been delisted, their historical data have to be downloaded from the list of “Dead Stocks”.
To obtain the full list of dead and active stocks on Bursa Malaysia, we rely on the “End of Year Shareholdings by Type of Investor”
dataset which compiles shareholder information for listed firms at the end of each calendar year. Thus, the final list includes all
Malaysian stocks that have been traded on the local bourse to avoid survivorship bias, but removes financial firms that fall under a
different regulatory framework. Our data thus cover 1250 Malaysian public firms over the 16-year period of 2000–2015, but the
number of observations varies for each variable. All continuous variables, with the exception of dummies, are winsorized at the 1st
and 99th percentiles to reduce the influence of outliers.
Table 1
Summary statistics.
N Mean S.D. P5 P25 Median P75 P95
CPQS 13,827 5.3370 7.2572 0.7200 1.3219 2.7051 6.0681 19.3767

NSH 13,933 6233.7390 8575.8360 963 1750 3136 6719 23,446
LIND 13,933 32.9853 21.7994 4.2293 15.5346 29.3464 46.8621 75.8972
LINST 13,933 25.3909 22.3509 0.5922 5.7344 18.8527 42.3337 68.3589
ANALYST 19,569 0.9950 3.2044 0.0000 0.0000 0.0000 0.0000 7.0000
RETURN 16,428 −0.0194 0.1882 −0.3670 −0.0942 0.0000 0.0700 0.2732
VOL 14,034 3.5105 2.2864 1.1487 1.9848 2.9221 4.3165 7.956
TURNOVER 13,894 0.3712 0.6902 0.0146 0.0482 0.1223 0.3514 1.6489
SIZE 14,462 12.5549 1.5142 10.2883 11.523 12.4157 13.4748 15.2691
BLOCK 13,634 0.3278 0.2820 0.0000 0.0000 0.3400 0.5800 0.7500
BSIZE 13,941 7.4992 2.0055 5.0000 6.0000 7.0000 9.0000 11.0000
BINDEP 13,941 0.4258 0.1258 0.2500 0.3333 0.4000 0.5000 0.6667
DUALa 626 0.0449 0.2071 0.0000 0.0000 0.0000 0.0000 0.0000
CHAIRa 4727 0.3391 0.4734 0.0000 0.0000 0.0000 1.0000 1.0000
Notes: This table presents the summary statistics for all the variables in the liquidity model (2), with detailed definitions provided in the Appendix.
Closing Percent Quoted Spread (CPQS) is computed as the ratio of the difference of closing ask and closing bid prices over the mid-point of these
prices. NSH is the number of shareholders measured at year-end. LIND (LINST) is defined as the proportion of shares held by local individuals (local
institutions) relative to total shares outstanding. ANALYST is the number of security analysts issuing earnings forecasts for a firm during a calendar
year. Stock return (RETURN) is computed as the time series averages of daily returns for each year and each firm, whereas return volatility (VOL) is
the standard deviation of daily stock returns over the year. TURNOVER is computed as the annual average of daily turnover ratios, defined as the
number of shares traded scaled by the number of shares outstanding. Firm size (SIZE) is the book value of total assets at year-end. BLOCK denotes
blockholdings, defined as the total percentage share ownership held by shareholders with at least 5% of the outstanding shares. Board size (BSIZE) is
the total number of directors on a firm's board at year-end. Board independence (BINDEP) is proxied by the ratio of independent non-executive
directors over board size at year-end. CEO duality (DUAL) is a dummy variable that takes a value of one if the chief executive officer is also the board
chairman at year-end, zero otherwise. CHAIR is a dummy variable that takes a value of one if the board chairman is an independent non-executive
director at year-end, zero otherwise. Instead of taking natural logarithm as in the liquidity model (2), this table reports the original measurement
unit in numbers for NSH, ANALYST and BSIZE for ease of interpretation. All the continuous variables, with the exception of two dummies (DUAL and
CHAIR), are winsorized at the 1st and 99th percentiles to reduce the influence of outliers.
P denotes percentiles, where statistics are provided at the representative 5th, 25th, 75th and 95th percentiles. S.D. is the standard deviation. N
denotes the number of firm-year observations.
a
N indicates the number of firm-year observations that the dummy variable takes the value of one.
8
Table 2
Correlation matrix.
Variable CPQS lnNSH LIND LINST ln(1 + ANALYST)
CPQS 1.0000
lnNSH −0.3501⁎⁎⁎ 1.0000
LIND 0.2311⁎⁎⁎ −0.2115⁎⁎⁎ 1.0000
LINST −0.0034 −0.0394⁎⁎⁎ −0.4031⁎⁎⁎ 1.0000
ln(1 + ANALYST) −0.2691⁎⁎⁎ 0.2791⁎⁎⁎ −0.4057⁎⁎⁎ 0.0917⁎⁎⁎ 1.0000
RETURN −0.1272⁎⁎⁎ −0.0676⁎⁎⁎ −0.0685⁎⁎⁎ 0.0669⁎⁎⁎ 0.0802⁎⁎⁎
VOL 0.6726⁎⁎⁎ −0.1210⁎⁎⁎ 0.3248⁎⁎⁎ −0.1242⁎⁎⁎ −0.3168⁎⁎⁎
TURNOVER −0.0918⁎⁎⁎ −0.0441⁎⁎⁎ 0.1730⁎⁎⁎ −0.1454⁎⁎⁎ −0.0701⁎⁎⁎
lnSIZE −0.4313⁎⁎⁎ 0.5853⁎⁎⁎ −0.5440⁎⁎⁎ 0.1259⁎⁎⁎ 0.5647⁎⁎⁎
BLOCK 0.0525⁎⁎⁎ −0.1130⁎⁎⁎ −0.1049⁎⁎⁎ 0.1261⁎⁎⁎ 0.0529⁎⁎⁎
lnBSIZE −0.1621⁎⁎⁎ 0.1017⁎⁎⁎ −0.2414⁎⁎⁎ 0.1042⁎⁎⁎ 0.2657⁎⁎⁎
BINDEP 0.0565⁎⁎⁎ 0.0529⁎⁎⁎ 0.0952⁎⁎⁎ −0.0795⁎⁎⁎ −0.0558⁎⁎⁎
DUAL −0.0222⁎⁎⁎ 0.0397⁎⁎⁎ −0.0289⁎⁎⁎ 0.0024 0.0435⁎⁎⁎
CHAIR 0.0012 −0.0198⁎⁎ 0.0106 −0.0443⁎⁎⁎ 0.0088
RETURN VOL TURNOVER lnSIZE BLOCK
RETURN 1.0000
VOL −0.1802⁎⁎⁎ 1.0000
TURNOVER 0.0805⁎⁎⁎ 0.2035⁎⁎⁎ 1.0000
lnSIZE 0.0851⁎⁎⁎ −0.4646⁎⁎⁎ −0.2143⁎⁎⁎ 1.0000
BLOCK 0.1150⁎⁎⁎ −0.0890⁎⁎⁎ −0.1220⁎⁎⁎ 0.0568⁎⁎⁎ 1.0000
lnBSIZE 0.0599⁎⁎⁎ −0.2516⁎⁎⁎ −0.0878⁎⁎⁎ 0.3343⁎⁎⁎ 0.0099
BINDEP 0.0292⁎⁎⁎ 0.0966⁎⁎⁎ 0.0466⁎⁎⁎ −0.0443⁎⁎⁎ 0.0540⁎⁎⁎
DUAL −0.0177⁎⁎ 0.0034 −0.0020 0.0395⁎⁎⁎ 0.0089
CHAIR 0.0260⁎⁎⁎ 0.0004 0.0254⁎⁎⁎ −0.0227⁎⁎⁎ 0.0235⁎⁎
lnBSIZE BINDEP DUAL CHAIR
lnBSIZE 1.0000
BINDEP −0.3667⁎⁎⁎ 1.0000
DUAL −0.0702⁎⁎⁎ 0.0308⁎⁎⁎ 1.0000
CHAIR −0.0181⁎⁎ 0.2158⁎⁎⁎ −0.1553⁎⁎⁎ 1.0000
Notes: This table presents the Pearson correlation coefficients between pairs of variables in the liquidity model (2), with detailed definitions
provided in the Appendix. Closing Percent Quoted Spread (CPQS) is computed as the ratio of the difference of closing ask and closing bid prices over
the mid-point of these prices. NSH is the number of shareholders measured at year-end. LIND (LINST) is defined as the proportion of shares held by
local individuals (local institutions) relative to total shares outstanding. ANALYST is the number of security analysts issuing earnings forecasts for a
firm during a calendar year. Stock return (RETURN) is computed as the time series averages of daily returns for each year and each firm, whereas
return volatility (VOL) is the standard deviation of daily stock returns over the year. TURNOVER is computed as the annual average of daily turnover
ratios, defined as the number of shares traded scaled by the number of shares outstanding. Firm size (SIZE) is the book value of total assets at year-
end. BLOCK denotes blockholdings, defined as the total percentage share ownership held by shareholders with at least 5% of the outstanding shares.
Board size (BSIZE) is the total number of directors on a firm's board at year-end. Board independence (BINDEP) is proxied by the ratio of in-
dependent non-executive directors over board size at year-end. CEO duality (DUAL) is a dummy variable that takes a value of one if the chief
executive officer is also the board chairman at year-end, zero otherwise. CHAIR is a dummy variable that takes a value of one if the board chairman
is an independent non-executive director at year-end, zero otherwise.
, and ⁎ denote statistical significance at the 1%, 5% and 10% levels, respectively.
⁎⁎⁎ ⁎⁎
4.2. Descriptive statistics
Table 1 presents the summary statistics for variables in the liquidity model (2). Our focus is on the two key variables of Closing
Percent Quoted Spread (CPQS) and the number of shareholders (NSH). First, the average for CPQS is 5.3370, which is two times
higher than the mean reported by Fong et al. (2017) for 960 Malaysian stocks over the sample period of 1996–2007. Since CPQS is an
inverse measure of liquidity, the higher value implies our sample stocks are relatively illiquid, which can be attributed to the
inclusion of additional 290 firms (mostly delisted) and the drop in liquidity after the global financial crisis. Second, the average
number of shareholders for Malaysian firms is 6234, which is slightly lower than the mean of 6436 for NASDAQ firms, but 2.7 times
lower than the average for firms traded on NYSE and AMEX (see Jacoby and Zheng, 2010).
Table 2 presents the Pearson correlation matrix for all the variables in the liquidity model (2). The correlation between the
explanatory variables and CPQS provides a preliminary view of their univariate relationship. All the control variables have the
expected signs, with the exception of board independence (BINDEP), CEO duality (DUAL) and independent board chairman (CHAIR).
However, the significance of these relationships remains to be tested when all variables are included in the multivariate regression.
Turning to the key variable of interest, NSH is negatively and significantly correlated with CPQS, consistent with the conventional
wisdom that a larger shareholder base is associated with higher liquidity. The pooled regression in the next section addresses the
functional form of their relationship. In terms of the correlations between explanatory variables, all are within the plausible ranges
with the highest coefficient being 0.5853.
9
5. Nonlinear relationship between shareholder base and stock liquidity
While the existing empirical studies suggest a monotonic positive linear relationship between the size of shareholder base and
stock liquidity, we accommodate the possibility of nonlinearity by specifying a liquidity model with a quadratic term for the number
of shareholders. The liquidity model (2) is first estimated using pooled OLS, and then subjected to a battery of robustness checks to
ensure reliable statistical inferences.
5.1. Baseline results
Before interpreting the estimation results for the liquidity model (2), Table 3 first provides evidence that the inclusion of lnNSH
and lnNSH2 brings additional information to the model.14 To achieve the latter objective, we estimate Column (1) without share-
holder base and its squared term, followed by the addition of lnNSH in Column (2), and finally our liquidity model in Column (3).
From Table 3, the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) have the lowest values for Column
(3). This means the additional variables – lnNSH in Column (2), as well as lnNSH and lnNSH2 in Column (3) – reduce information loss
from Column (1) where these variables are not included, in spite of the penalty on both information criteria due to the addition of
more variables.
We compute the weights for the two information criteria as suggested by Burnham and Anderson (2004) and Wagenmakers and
Farrell (2004) as follows:
wi (IC ) =
exp { 1
2 i (IC ) }
4
k=1
exp { 1
2 k (IC ) } (3)
where IC is the information criterion, and ∆i(IC) is the difference between the information criterion of model i with the minimum
information criterion of all models. The weight represents the probability of model i being the model with least information loss. As is
obvious from Table 3, the probability of Columns (1) and (2) being the best model for explaining stock liquidity is close to zero.
Although the inclusion of lnNSH in Column (2) increases the weight, it is still far from the weight for our liquidity model in Column
(3). This analysis suggests that the quadratic term of lnNSH2 brings additional information by increasing the explanatory power to the
2
(Rnew 2 ) / df
Rold
model significantly, and cannot be ignored in modelling stock liquidity. In addition, the F-statistic – F = 2 ) / df where df
(1 Rnew
denotes the degrees of freedom – is computed to support the inclusion of lnNSH and lnNSH2 by comparing to the model without these
variables. The F-statistics for testing the significance of increase in explanatory power due to the inclusion of lnNSH alone, and
inclusion of both lnNSH and lnNSH2 in the model are 521.47 and 659.15, respectively. The corresponding F-statistic due to the
additional lnNSH2 in the model that has already included lnNSH is 131.23. The tests (p-values are close to 0.0000) provide evidence
that lnNSH and lnNSH2 have highly significant explanatory power of liquidity.
After establishing the additional explanatory power brought by lnNSH and lnNSH2 to our liquidity model, we now interpret the
estimation results in Table 3. Column (2) specifies a linear relationship between shareholder base and liquidity. The results show that
the number of shareholders is negatively and significantly associated with Closing Percent Quoted Spread (CPQS). Because the latter
is an inverse measure of liquidity, the negative coefficient suggests a larger shareholder base correlates with higher liquidity, con-
sistent with existing empirical evidence from the developed U.S. markets (Demsetz, 1968; Benston and Hagerman, 1974; Jacoby and
Zheng, 2010). The finding also lends support to the numerous policy measures undertaken by the Malaysian stock exchange to boost
investor participation. Theoretically, the increase in liquidity due to a larger shareholder base can be attributed to greater investor
recognition, information competition among informed traders or the liquidity trading of noise traders.
For the overall significance of our liquidity model in Column (3), the F-statistic rejects the joint hypothesis that all the regression
coefficients are equal to zero. The adjusted R-squared of 64.27% is considerably high relative to recent liquidity studies exploring
new determinants (see, for example, Farrell et al., 2018; Lee and Chung, 2018; Boubaker et al., 2019; Boubakri et al., 2019). The first-
order lnNSH retains its negative and significant coefficient whereas its squared term is positively and significantly associated with
CPQS, suggesting a nonlinear relationship between shareholder base and liquidity. Our finding of a U-shaped curve indicates that at
levels of shareholder base before the tipping point, a larger number of shareholders is associated with lower spread and thus higher
liquidity. However, when the shareholder base exceeds a certain threshold level and becomes too large, the negative effect dominates
and causes liquidity to drop. While our results support shareholder-boosting strategies by Malaysian authorities to improve liquidity,
the existence of a threshold point implies the potential costs of maintaining a very large shareholder base can outweigh its associated
benefits. The nonlinear relationship not only challenges the popular view that “more is better”, but prescribes a threshold maximum
level for the number of shareholders beyond which the liquidity of firms will decline.15 In our sample, only 627 firm-year ob-
servations exceed the threshold level. This represents 4.5% of the total observations, suggesting there is still much room for
shareholder expansion in the local bourse, and greater prospect for liquidity improvement.
14
We thank the anonymous referees of the journal for suggesting this analysis to assess the additional information brought by lnNSH and lnNSH2
to the liquidity model.
15
The threshold level can be computed using the estimated coefficients of lnNSH and lnNSH2 from our liquidity model in Column (3) of Table 3,
that is −γ1/2γ2 which gives lnNSH = 10.1367. Taking e10.1367, the threshold maximum level is 25,253 shareholders.
10
Table 3
Number of shareholders and stock liquidity.
(1) (2) (3) Economic Impact
lnNSH −1.1853⁎⁎⁎ −7.4241⁎⁎⁎ −7.2311

(0.1313) (1.0684)
lnNSH2 0.3662⁎⁎⁎ 6.0649
(0.0619)
LIND −0.0865 ⁎⁎⁎
−0.0489 ⁎⁎⁎
−0.0473⁎⁎⁎ −1.0311
(0.0171) (0.0129) (0.0129)
2
LIND 0.0009⁎⁎⁎ 0.0006⁎⁎⁎ 0.0006⁎⁎⁎ 1.0965
(0.0002) (0.0002) (0.0002)
LINST 0.0110⁎⁎⁎ 0.0094⁎⁎⁎ 0.0090⁎⁎⁎ 0.2012
(0.0036) (0.0033) (0.0033)
ln(1 + ANALYST) −0.0739 −0.0397 −0.0634 −0.0428
(0.0833) (0.0864) (0.0798)
RETURN 0.3515 −0.1135 −0.1964 −0.0370
(1.0557) (1.0479) (1.0544)
VOL 1.8479⁎⁎⁎ 1.9296⁎⁎⁎ 1.9141⁎⁎⁎ 4.3764
(0.1652) (0.1583) (0.1609)
TURNOVER −2.1449⁎⁎⁎ −2.0589⁎⁎⁎ −2.0325⁎⁎⁎ −1.4028
(0.2762) (0.2596) (0.2628)
lnSIZE −1.0708⁎⁎⁎ −0.4443⁎⁎⁎ −0.4262⁎⁎⁎ −0.6454
(0.1213) (0.1101) (0.1069)
BLOCK 1.5339⁎⁎⁎ 1.1661⁎⁎⁎ 1.1341⁎⁎⁎ 0.3198
(0.2312) (0.2471) (0.2637)
lnBSIZE 0.5421⁎⁎ 0.4101⁎ 0.3420 0.0903
(0.2712) (0.2459) (0.2509)
BINDEP −1.0934⁎⁎ −0.4374 −0.4053 −0.0510
(0.5042) (0.4956) (0.4709)
DUAL −0.2549 −0.2598 −0.3188 −0.0660
(0.2165) (0.2087) (0.1981)
CHAIR −0.1109 −0.1153 −0.1158 −0.0548
(0.1186) (0.1101) (0.1080)
CONSTANT 13.5560⁎⁎⁎ 17.1145⁎⁎⁎ 42.1284⁎⁎⁎
(1.6579) (1.7299) (4.6769)
Year dummies Yes Yes Yes
Industry dummies Yes Yes Yes
N 10,664 10,664 10,664
F-statistic 176.36⁎⁎⁎ 176.08⁎⁎⁎ 180.95⁎⁎⁎
Adj. R2 0.6206 0.6382 0.6427
AIC 58,389 57,882 57,748
BIC 58,549 58,057 57,916
wi(AIC)a 6.44x10−140 7.98x10−30 1.0000
wi(BIC)b 3.51x10−138 2.41x10−31 1.0000
Notes: This table presents the pooled OLS estimation results for liquidity model (2) where the dependent variable is CPQS. The economic impacts of
all explanatory variables on CPQS are computed by multiplying one standard deviation of the variable with its corresponding coefficient estimate
from Column (3). Closing Percent Quoted Spread (CPQS) is computed as the ratio of the difference of closing ask and closing bid prices over the mid-
point of these prices. NSH is the number of shareholders measured at year-end. LIND (LINST) is defined as the proportion of shares held by local
individuals (local institutions) relative to total shares outstanding. ANALYST is the number of security analysts issuing earnings forecasts for a firm
during a calendar year. Stock return (RETURN) is computed as the time series averages of daily returns for each year and each firm, whereas return
volatility (VOL) is the standard deviation of daily stock returns over the year. TURNOVER is computed as the annual average of daily turnover ratios,
defined as the number of shares traded scaled by the number of shares outstanding. Firm size (SIZE) is the book value of total assets at year-end.
BLOCK denotes blockholdings, defined as the total percentage share ownership held by shareholders with at least 5% of the outstanding shares.
For brevity, year and industry dummies are suppressed. Double-clustered standard errors are reported in parentheses. N denotes the number of firm-
year observations.
a,b
denote weights computed for the information criterion that represents probability of the estimated equation being the model with least in-
formation loss.
⁎⁎⁎ ⁎⁎
Turning to the control variables in Column (3), the local individual ownership and local institutional ownership are statistically
significant at the 1% level. Their signs reinforce previous findings by Lim et al. (2017) for a smaller sample of Malaysian public listed
firms over shorter time period. In the first case, liquidity increases with larger individual ownership but declines after their share-
holdings exceed the threshold level. An increase in local institutional ownership, on the other hand, is associated with higher spread
and hence lower liquidity. Our extension of their model to include the ownership of large shareholders (BLOCK) provides evidence
11
consistent with the interpretation that blockholders exacerbate information asymmetry and reduce liquidity due to their privileged
access to private information (Heflin and Shaw, 2000; Ng et al., 2016). Notably, the inclusion of these three ownership variables does
not suppress the explanatory power of shareholder base, suggesting the number of shareholders represents a distinct dimension of
shareholdings that should not be ignored in future studies. Among the control variables, analyst coverage and stock return are
insignificant whereas the remaining three variables yield the predicted signs. Liquid firms are associated with larger size, higher
turnover and lower volatility. While corporate governance has been found to be an important determinant of liquidity (see Chung
et al., 2010; Al-Jaifi et al., 2017), our evidence suggests the contrary for Malaysian firms as none of the four proxies – board size
(BSIZE), board independence (BINDEP), CEO duality (DUAL) and independent chairman (CHAIR) – are statistically significant.
Following Boubaker et al. (2019), we report the economic impacts of all explanatory variables on CPQS in the last column of
Table 3, computed by multiplying one standard deviation of the variable with its corresponding coefficient estimate from Column (3).
For example, a one standard deviation increase in lnNSH is associated with a reduction of 7.2311 in CPQS, which is economically
large. After the threshold level, CPQS rises by 6.0649 in response to a one standard deviation increase of lnNSH. Putting the economic
magnitudes of all variables in the liquidity model into perspective, the largest effect comes from our newly included variable of
shareholder base, which is at least four times larger than the combined effects of all three ownership variables – local individual
ownership, local institutional ownership and blockholdings.16
5.2. Robustness checks
We conduct a series of robustness checks on our key finding of a nonlinear relationship between shareholder base and stock
liquidity.
5.2.1. Alternative measure of shareholder base

Throughout this paper, our measure of shareholder base is the natural logarithm of the number of shareholders. Citing previous
empirical evidence, Bodnaruk and Östberg (2013) caution the strong correlation between number of shareholders and some control
variables such as firm size and firm age. Following their orthogonalization method, we construct excess shareholder base (ESHB) by
collecting the residuals from the following model17:
ln NSHit = 0 + 1 ln AGEit + 2 ROEit + 3 ln SIZEit + 4 BMit + 5 RETURNit
J 1 T 1
+ 6 TURNOVERit + 7 VOLit + j = 1 8j
INDj + t = 1 9t YRt + it (4)
where ln refers to natural logarithm, the dependent variable is the number of shareholders (NSH), while the regressors consist of firm
age in years since incorporation (AGE), return on equity (ROE), firm size measured by book value of total assets (SIZE), ratio of book
value to market value of equity (BM), annual stock returns (RETURN), stock turnover (TURNOVER) and stock return volatility (VOL).
INDj and YRt are industry and year dummies, respectively.
Panel A of Table 4 presents the estimation results for the liquidity model (2) but using the orthogonalized excess shareholder base
(ESHB) in place of the number of shareholders (NSH). The signs and statistical significance for all the regressors are consistent with
the baseline findings reported in Table 3. In the linear model, ESHB has a significant negative coefficient, reinforcing that a larger
shareholder base is associated with lower spread and hence higher liquidity. However, when a quadratic term for ESHB is added to
the model, the relationship becomes U-shaped which imposes an upper bound on liquidity benefit for firms pursuing shareholder-
boosting strategies. Since the results are qualitatively similar, the number of shareholders is a robust measure not driven by firm
characteristics.
5.2.2. Alternative liquidity measures

Lee and Chung (2018) find that foreign ownership correlates positively with the price impact of trades but negatively with bid-ask
spread, which highlight the importance of distinguishing the different dimensions of liquidity – adverse selection risks versus trading
costs. To address their concern, we replace the percent-cost proxy of CPQS with price impact measures. For Malaysian stocks, the
horseraces conducted by Fong et al. (2017) find that the price impact version of CPQS (hereafter referred to as CPQSIM) and Amihud
(2002) illiquidity ratio (hereafter referred to as ILLIQ) are the top two performing proxies in the cost-per-volume category. We
compute their annual liquidity estimates for each stock by averaging the computed daily ratios across all trading days for each
calendar year, where the daily data are all sourced from Thomson Datastream. Since the Amihud illiquidity ratio is highly skewed, we
follow the literature in taking natural logarithm of one plus ILLIQ x 106. The same approach is applied for CPQS Impact by taking the
natural logarithm of one plus CPQSIM x 104. These two price impact proxies are inverse measures of liquidity, where higher values
indicate greater degree of illiquidity.
16
We also follow Chan and Covrig (2012) to standardize all variables, with the exception of dummies, to have means of zero and standard
deviations of one. Our unreported estimation results for liquidity model (2) using the standardized variables reaffirm that the largest economic effect
comes from the number of shareholders (NSH).
17
The orthogonalization method is commonly used in empirical finance research to extract the component of the dependent variable that has been
purged from other strongly correlated variables through the regression residuals (see, for example, Hong et al., 2000, 2008; Kale and Loon, 2011). In
the present context, Bodnaruk and Östberg (2013) utilize the orthogonalization method to remove the effect of selected variables on shareholder
base, so that their results are not driven by firm characteristics unrelated to investor recognition.
12
Table 4
Robustness check with alternative measures of shareholder base and liquidity.
Excess shareholder base Price impact measures
Panel A: CPQS Panel B: CPQSIM Panel C: ILLIQ
ESHB −1.3638⁎⁎⁎ −1.4501⁎⁎⁎

(0.1376) (0.1394)
ESHB2 0.3296⁎⁎⁎
(0.0754)
lnNSH −0.5991⁎⁎⁎ −1.2488⁎⁎ −0.5601⁎⁎⁎ −0.0457
(0.0462) (0.5204) (0.0425) (0.5090)
lnNSH2 0.0380 −0.0302
(0.0312) (0.0300)
LIND −0.0437⁎⁎⁎ −0.0410⁎⁎⁎ −0.0160⁎⁎⁎ −0.0159⁎⁎⁎ −0.0043 −0.0044
(0.0127) (0.0128) (0.0041) (0.0041) (0.0042) (0.0042)
2
LIND 0.0005⁎⁎⁎ 0.0005⁎⁎⁎ 0.0002⁎⁎⁎ 0.0002⁎⁎⁎ 0.0001⁎ 0.0001⁎
(0.0002) (0.0002) (0.00004) (0.00004) (0.00004) (0.00004)
LINST 0.0100⁎⁎⁎ 0.0101⁎⁎⁎ 0.0042⁎⁎⁎ 0.0042⁎⁎⁎ 0.0031⁎⁎ 0.0031⁎⁎
(0.0033) (0.0033) (0.0011) (0.0011) (0.0012) (0.0012)
ln(1 + ANALYST) 0.0266 0.0449 −0.6043⁎⁎⁎ −0.6070⁎⁎⁎ −0.7743⁎⁎⁎ −0.7722⁎⁎⁎
(0.0874) (0.0887) (0.0417) (0.0405) (0.0358) (0.0356)
RETURN 0.5867 0.6285 −0.3757 −0.3858 −0.6249⁎⁎⁎ −0.6198⁎⁎⁎
(1.0577) (1.0570) (0.2859) (0.2893) (0.2105) (0.2118)
VOL 1.8372⁎⁎⁎ 1.8234⁎⁎⁎ 0.4884⁎⁎⁎ 0.4869⁎⁎⁎ 0.4521⁎⁎⁎ 0.4535⁎⁎⁎
(0.1495) (0.1505) (0.0300) (0.0292) (0.0303) (0.0296)
TURNOVER −2.1265⁎⁎⁎ −2.1248⁎⁎⁎ −1.0467⁎⁎⁎ −1.0446⁎⁎⁎ −1.0556⁎⁎⁎ −1.0577⁎⁎⁎
(0.2521) (0.2506) (0.0771) (0.0771) (0.0671) (0.0668)
lnSIZE −0.9553⁎⁎⁎ −0.9693⁎⁎⁎ −0.3761⁎⁎⁎ −0.3743⁎⁎⁎ −0.3995⁎⁎⁎ −0.4008⁎⁎⁎
(0.1146) (0.1144) (0.0402) (0.0398) (0.0361) (0.0361)
BLOCK 1.0460⁎⁎⁎ 1.0736⁎⁎⁎ 0.5818⁎⁎⁎ 0.5789⁎⁎⁎ 0.4014⁎⁎⁎ 0.4040⁎⁎⁎
(0.2447) (0.2449) (0.1517) (0.1539) (0.1285) (0.1267)
lnBSIZE 0.3772 0.3577 0.0971 0.0897 0.1085 0.1150
(0.2466) (0.2491) (0.0885) (0.0895) (0.0839) (0.0849)
BINDEP −0.5044 −0.4646 −0.2673 −0.2648 −0.0708 −0.0729
(0.5012) (0.4962) (0.1974) (0.1973) (0.2007) (0.2003)
DUAL −0.2777 −0.2959 0.0415 0.0349 0.0531 0.0587
(0.2057) (0.2059) (0.0980) (0.0990) (0.0872) (0.0876)
CHAIR −0.1220 −0.1199 −0.1141⁎⁎ −0.1144⁎⁎ −0.1137⁎⁎ −0.1137⁎⁎
(0.1095) (0.1085) (0.0515) (0.0518) (0.0501) (0.0499)
CONSTANT 11.2937⁎⁎⁎ 11.4285⁎⁎⁎ 11.1694⁎⁎⁎ 13.7951⁎⁎⁎ 13.7849⁎⁎⁎ 11.7206⁎⁎⁎
(1.5715) (1.5741) (0.5017) (2.0556) (0.3832) (2.0594)
Year dummies Yes Yes Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes Yes Yes
N 10,629 10,629 10,266 10,266 10,216 10,216
F-statistic 177.09⁎⁎⁎ 173.05⁎⁎⁎ 626.50⁎⁎⁎ 631.44⁎⁎⁎ 790.49⁎⁎⁎ 757.92⁎⁎⁎
Adj. R2 0.6435 0.6448 0.6801 0.6804 0.7170 0.7171
Notes: This table presents the pooled OLS estimation results for liquidity model (2) but using alternative measures of shareholder base and liquidity.
In Panel A, the dependent variable is CPQS, but shareholder base is replaced with excess shareholder base (ESHB) generated as the residuals from
model (4). In Panel B and Panel C, the dependent variable is replaced with CPQS Impact (CPQSIM) and Amihud illiquidity ratio (ILLIQ), respectively.
director at year-end, zero otherwise.
year observations.
⁎⁎⁎ ⁎⁎
Table 4 presents the estimation results for liquidity model (2) with the dependent variable replaced by CPQS Impact (Panel B) and
Amihud illiquidity ratio (Panel C). While our baseline result reveals a nonlinear relationship between number of shareholders and
bid-ask spread, the last four columns of Table 4 suggest a different functional form, that is the relationship between shareholder base
and price impact is at best linear. Previous Table 3 indicates that the bid-ask spread becomes wider when the number of shareholders
13
exceeds the threshold level, but there is no such upper limit imposed when liquidity is proxied by price impact measures, since a
larger number of shareholders is associated monotonically with lower price impact. Such linear relationship arises because the
positive effects dominate the negative channels at all levels of shareholder base, which implies the shareholder-boosting strategies do
not pose greater adverse selection risks to liquidity providers. Liew et al. (2018) reach similar conclusion on the differing liquidity
effects for Malaysian stocks in that foreign trading broadens bid-ask spread but does not affect the price impact. Thus, corporate
managers and stock exchange regulators should pay exclusive attention to the negative effect of a very large shareholder base on the
trading costs incurred by liquidity demanders.
5.2.3. Alternative estimation methods

In our baseline Table 3, the liquidity model (2) is estimated using pooled OLS with the standard errors adjusted for within-cluster
correlations. We further determine whether the nonlinear relationship between shareholder base and liquidity is robust to alternative
estimation methods – Fama-MacBeth two-step regression and quantile regression. The former involves estimating cross-sectional
regression for each year separately, and inferences are drawn from the time-series averages of the estimated coefficients, thus picking
up cross-sectional effects. The latter goes beyond the conditional mean to examine the effects of shareholder base along the entire
range of the liquidity conditional distribution, especially at the extreme upper and lower tails.
Table 5 presents the estimation results for liquidity model (2) using both methods. In the first case of Fama-MacBeth two-step
regression, the signs and statistical significance for all regressors mirror those obtained through pooled OLS, with the exception that
ANALYST and DUAL now become statistically significant and correlated with higher liquidity. The U-shaped curve for shareholder
base-CPQS remains intact. Turning to the quantile regression, the table reports estimates at the 10th, 25th, 50th, 75th and 90th
percentiles of the liquidity conditional distribution. Across the five representative quantiles, the coefficients for lnNSH and lnNSH2
are highly significant with the signs consistent with a U-shaped function, suggesting the nonlinear relationship is not confined to
firms with liquidity around the mean but widespread across all firms with liquidity at different quantiles of its distribution.18 To
reiterate, both estimation methods reinforce our key finding of a threshold level in shareholder-boosting strategies for improving
liquidity.
5.2.4. Disaggregate investor types

The proxy for shareholder base in our baseline result is the total number of shareholders for Malaysian public listed firms.
Motivated by the growing research on investor heterogeneity, we further determine whether the nonlinear relationship prevails
across all investor types – total local investors, local individuals, local institutions, local government, local nominees, total foreign
investors, foreign individuals, foreign institutions and foreign nominees. We re-estimate the liquidity model (2) but replace the total
number of shareholders with the breakdown according to different investor types, each entering the model separately. Panel A of
Table 6 presents the key summary statistics for these disaggregate investor types. The regression results in Panel B confirm the
robustness of our key finding since both lnNSH and lnNSH2 are highly significant across all nine investor types, with signs that are
consistent with a U-shaped curve. This adds further credence to the inclusion of shareholder base in a liquidity model, as its ex-
planatory power is not suppressed by percentage ownership variables and not driven by specific investor types.
5.2.5. Endogeneity
Endogeneity, in particularly reverse causality, is a valid concern given the close-knit relationship between shareholder base and
liquidity. Within the scarce literature, Bodnaruk and Östberg (2013) establish the causal relationship from shareholder base to
dividend payout and cash holdings using the 2001 decimalization of tick size as exogenous shock to shareholder base. However, this
natural experiment has also been widely used as a standard exogenous liquidity shock for U.S. studies (see Fang et al., 2009; Bharath
et al., 2013; Fang et al., 2014; Dou et al., 2018), illustrating the difficulty to disentangle shareholder base from liquidity.19
Jankensgård and Vilhelmsson (2018) employ the number of shares as an exogenous instrument for shareholder base in a two-stage
least squares (2SLS) approach, arguing that there is no reason for the former to be correlated with volatility. However, this is
inapplicable in our context given recent evidence that liquidity increases when more shares are available either by altering trading
activities or alleviating information asymmetries (Ding et al., 2016; El-Nader, 2018).
Ding et al. (2016) encounter similar challenges in establishing the causal relationship from free-float shares to liquidity,
prompting them to resort to some mechanical approaches. Given the lack of natural experiments and strictly exogenous instruments,
we follow this route and employ three statistical techniques to tackle endogeneity. First, we use fixed effects approach as the
extensive simulations in Gormley and Matsa (2014) show that it yields consistent estimates in the presence of time-invariant un-
observed heterogeneity. The firm fixed effects estimation results in Table 7 reaffirm the nonlinear relationship between shareholder
base and liquidity, hence ruling out the role of time-invariant unobservable firm factors that might correlate with both liquidity and
18
This is confirmed by our unreported graphical plots for the coefficient estimates of lnNSH and lnNSH2 against the entire conditional distribution
of CPQS. Across all quantiles, the coefficient of lnNSH is negative and trending downward, whereas its squared term is positive and sloping upward.
All the quantile-varying estimates for both variables are statistically significant as their corresponding 95% confidence intervals do not overlap with
zero. The figures are available upon request from the authors.
19
We also rule out another potential candidate of exogenous shock – the lot size reduction used by Chia et al. (2020) – as evidence from Japan
shows that the reduction in minimum trading unit expands substantially the number of small individual investors and increases liquidity (see
Amihud et al., 1999; Ahn et al., 2014). Thus, it is difficult to identify shock to shareholder base that is unrelated to liquidity.
14
Table 5
Robustness check with alternative estimation methods.
Fama-MacBeth Quantile regression
10th 25th 50th 75th 90th
lnNSH −8.4520⁎⁎⁎ −1.2066⁎⁎⁎ −2.8287⁎⁎⁎ −4.7616⁎⁎⁎ −9.5784⁎⁎⁎ −14.5201⁎⁎⁎

(0.7730) (0.1953) (0.3284) (0.3940) (0.8889) (1.7475)
lnNSH2 0.4329⁎⁎⁎ 0.0524⁎⁎⁎ 0.1394⁎⁎⁎ 0.2380⁎⁎⁎ 0.4938⁎⁎⁎ 0.7502⁎⁎⁎
(0.0448) (0.0113) (0.0187) (0.0217) (0.0491) (0.0974)
LIND −0.0440⁎⁎⁎ −0.0103⁎⁎⁎ −0.0191⁎⁎⁎ −0.0291⁎⁎⁎ −0.0463⁎⁎⁎ −0.0817⁎⁎⁎
(0.0103) (0.0033) (0.0030) (0.0034) (0.0060) (0.0103)
LIND2 0.0005⁎⁎⁎ 0.0002⁎⁎⁎ 0.0003⁎⁎⁎ 0.0003⁎⁎⁎ 0.0005⁎⁎⁎ 0.0009⁎⁎⁎
(0.0001) (0.00004) (0.00004) (0.00005) (0.0001) (0.0002)
LINST 0.0089⁎⁎⁎ 0.0006 0.0010 0.0029⁎⁎⁎ 0.0048⁎⁎⁎ 0.0084⁎⁎⁎
(0.0023) (0.0007) (0.0007) (0.0008) (0.0011) (0.0022)
ln(1 + ANALYST) −0.1261⁎⁎ −0.1822⁎⁎⁎ −0.1322⁎⁎⁎ −0.0686⁎⁎ −0.0808⁎⁎ −0.0612
(0.0507) (0.0150) (0.0219) (0.0321) (0.0391) (0.0486)
RETURN −0.5020 −0.5056⁎⁎⁎ −0.6098⁎⁎⁎ −0.9165⁎⁎⁎ −0.8173⁎⁎ −1.8825⁎⁎
(0.7923) (0.1090) (0.1171) (0.2305) (0.3206) (0.7363)
VOL 1.6841⁎⁎⁎ 0.7098⁎⁎⁎ 1.0537⁎⁎⁎ 1.5199⁎⁎⁎ 1.9874⁎⁎⁎ 2.6667⁎⁎⁎
(0.1171) (0.0326) (0.0272) (0.0285) (0.0472) (0.0732)
TURNOVER −2.1256⁎⁎⁎ −1.4691⁎⁎⁎ −1.5980⁎⁎⁎ −1.6826⁎⁎⁎ −1.6209⁎⁎⁎ −1.6346⁎⁎⁎
(0.3042) (0.0892) (0.0863) (0.0521) (0.0729) (0.1025)
lnSIZE −0.4353⁎⁎⁎ −0.1993⁎⁎⁎ −0.2584⁎⁎⁎ −0.2964⁎⁎⁎ −0.2789⁎⁎⁎ −0.3117⁎⁎⁎
(0.1000) (0.0215) (0.0106) (0.0198) (0.0310) (0.0367)
BLOCK 1.0099⁎⁎⁎ −0.0621 0.0935 0.4943⁎⁎⁎ 0.8801⁎⁎⁎ 1.0789⁎⁎⁎
(0.2795) (0.0942) (0.1013) (0.1301) (0.1676) (0.1729)
lnBSIZE 0.2255 0.1320⁎ 0.2458⁎⁎⁎ 0.4080⁎⁎⁎ 0.3095⁎⁎⁎ 0.3111⁎⁎
(0.1831) (0.0676) (0.0559) (0.0581) (0.0767) (0.1509)
BINDEP −0.4991 −0.0934 −0.0540 −0.3413⁎⁎ −0.8192⁎⁎⁎ 0.0594
(0.2916) (0.1156) (0.1336) (0.1643) (0.2678) (0.3832)
DUAL −0.2958⁎⁎⁎ 0.0623 −0.0209 0.0436 −0.3045⁎⁎⁎ −0.4902⁎⁎⁎
(0.0943) (0.0551) (0.0580) (0.0824) (0.0969) (0.1121)
CHAIR −0.0782 −0.0697⁎⁎⁎ −0.0907⁎⁎⁎ −0.0804⁎⁎ −0.1646⁎⁎⁎ −0.1764⁎⁎
(0.0558) (0.0262) (0.0243) (0.0346) (0.0588) (0.0776)
CONSTANT 46.0279⁎⁎⁎ 9.7444⁎⁎⁎ 17.3469⁎⁎⁎ 27.0259⁎⁎⁎ 49.4703⁎⁎⁎ 72.6522⁎⁎⁎
(3.9990) (1.0903) (1.4379) (1.8911) (3.8482) (7.7260)
Year dummies No Yes Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes Yes Yes
N 10,664 10,664 10,664 10,664 10,664 10,664
R2/Pseudo R2 0.6110 0.1981 0.2905 0.3999 0.4842 0.5477
Notes: This table presents the results for liquidity model (2) where the dependent variable is CPQS but replaces pooled OLS with Fama-MacBeth two-
step regression and quantile regression.
director at year-end, zero otherwise.
For brevity, year and industry dummies are suppressed. Standard errors are reported in parentheses. N denotes the number of firm-year ob-
servations.
⁎⁎⁎ ⁎⁎
number of shareholders.
Second, we estimate a two-stage least squares (2SLS) instrument variable regression to further address endogeneity, especially
when the unobservable factors are not constant over time. Our approach follows Boubaker et al. (2019) in using two-year and three-
year lagged values of the endogenous variable as instruments, in our case the second and third lags of the number of shareholders.
Lagged variables are used as exogenous instruments because the unobservable factors that affect current liquidity are not expected to
correlate with the number of shareholders in previous years (for similar argument, see Fang et al., 2009; Jayaraman and Milbourn,
2012; Wang and Zhang, 2015). The p-value of the Sargan test in Table 7 shows that the instruments are not correlated with the errors
of the second-stage regression. Both the predicted values for number of shareholders and its squared terms remain highly significant
with the signs consistent with a U-shaped relationship, reaffirming that the baseline results are robust to the endogeneity of
15
Table 6
Robustness check with disaggregate investor types.
Total local Local Local Local Local nominees Total foreign Foreign Foreign Foreign
investors individuals institutions government investors individuals institutions nominees
Panel A: Descriptive statistics

N 13,933 13,933 13,933 13,933 13,933 13,933 13,933 13,933 13,933
Mean 5845.7450 5056.9860 78.8618 2.2166 689.2706 342.9012 96.0474 3.0741 242.4332
S.D. 7735.1860 6797 93.0860 3.3310 872.8159 971.3544 163.2683 5.4825 835.8709
P5 937 790 7 0 86 10 6 0 2
P25 1708 1455 22 0 202 29 17 0 9
Median 3061 2613 44 1 380 61 34 1 23
P75 6429 5520 95 3 780 180 88 3 82
P95 21,303 18,637 287 9 2471 1591 449 16 1137
Panel B: Disaggregate investor types

lnNSH −7.3906⁎⁎⁎ −5.9430⁎⁎⁎ −1.5567⁎⁎⁎ −0.8130⁎⁎⁎ −7.7765⁎⁎⁎ −2.2010⁎⁎⁎ −1.7174⁎⁎⁎ −0.9726⁎⁎⁎ −1.2903⁎⁎⁎
(1.1217) (1.0129) (0.4720) (0.1960) (1.0642) (0.2962) (0.3065) (0.1892) (0.1642)
lnNSH2 0.3653⁎⁎⁎ 0.2913⁎⁎⁎ 0.1152⁎⁎ 0.2994⁎⁎⁎ 0.4949⁎⁎⁎ 0.1653⁎⁎⁎ 0.1416⁎⁎⁎ 0.3067⁎⁎⁎ 0.1041⁎⁎⁎
(0.0656) (0.0607) (0.0516) (0.0770) (0.0829) (0.0253) (0.0334) (0.0606) (0.0163)
N 10,664 10,664 10,664 10,664 10,664 10,664 10,664 10,664 10,664
F-statistic 180.59⁎⁎⁎ 177.26⁎⁎⁎ 181.90⁎⁎⁎ 171.27⁎⁎⁎ 196.80⁎⁎⁎ 179.40⁎⁎⁎ 176.29⁎⁎⁎ 171.48⁎⁎⁎ 179.38⁎⁎⁎
Adjusted R2 0.6427 0.6388 0.6268 0.6217 0.6662 0.6331 0.6308 0.6228 0.6324
Notes: Panel A of this table provides the summary statistics for the disaggregate investor types. P denotes percentiles, where statistics are provided at
the representative 5th, 25th, 75th and 95th percentiles. S.D. is the standard deviation. N denotes the number of firm-year observations.
Panel B of this table presents the pooled OLS estimation results for liquidity model (2) where the dependent variable is CPQS, but disaggregate the
number of shareholders according to investor types. Closing Percent Quoted Spread (CPQS) is computed as the ratio of the difference of closing ask
and closing bid prices over the mid-point of these prices. NSH is the number of shareholders measured at year-end. The control variables in the
liquidity model are LIND, LIND2 LINST, ln(1 + ANALYST), RETURN, VOL, TURNOVER, lnSIZE, BLOCK, lnBSIZE, BINDEP, DUAL and CHAIR. LIND
(LINST) is defined as the proportion of shares held by local individuals (local institutions) relative to total shares outstanding. ANALYST is the
number of security analysts issuing earnings forecasts for a firm during a calendar year. Stock return (RETURN) is computed as the time series
averages of daily returns for each year and each firm, whereas return volatility (VOL) is the standard deviation of daily stock returns over the year.
TURNOVER is computed as the annual average of daily turnover ratios, defined as the number of shares traded scaled by the number of shares
outstanding. Firm size (SIZE) is the book value of total assets at year-end. BLOCK denotes blockholdings, defined as the total percentage share
ownership held by shareholders with at least 5% of the outstanding shares. Board size (BSIZE) is the total number of directors on a firm's board at
year-end. Board independence (BINDEP) is proxied by the ratio of independent non-executive directors over board size at year-end. CEO duality
(DUAL) is a dummy variable that takes a value of one if the chief executive officer is also the board chairman at year-end, zero otherwise. CHAIR is a
dummy variable that takes a value of one if the board chairman is an independent non-executive director at year-end, zero otherwise.
For brevity, estimates for control variables, constant, industries and year dummies are suppressed but available upon request. Double-clustered
standard errors are reported in the parentheses. N denotes the number of firm-year observations.
⁎⁎⁎ ⁎⁎
shareholder base.20
Third, to address the possibility of reverse causality, we use the system generalized method-of-moments (GMM) estimator ad-
vocated by Wintoki et al. (2012) for corporate finance research, and our implementation follows Roodman (2009). This system of two
simultaneous equations in both levels and differences deals with time-invariant unobserved heterogeneity by means of first-differ-
encing. Simultaneous causality is addressed through the use of internal instruments consisting of the lagged levels and lagged first-
differences of the explanatory variables for equations in first differences and levels, respectively (for details, see Wintoki et al., 2012).
We rewrite the baseline liquidity model (2) by adding the lagged dependent variable of CPQS as a regressor to reflect the dynamic
relationship between shareholder base and liquidity. Table 7 presents the system GMM estimation results for the dynamic panel
model. Both lnNSH and lnNSH2 retain their signs and statistical significance, indicating a causal relationship that runs from share-
holder base to liquidity.
While the results from firm fixed effects, 2SLS and system GMM provide strong corroborative evidence of a causal relationship
between shareholder base and liquidity, we refrain from claiming that the endogeneity problem has been entirely resolved, and are
mindful of the bias that might be induced by omitted time-varying unobservable heterogeneity.
5.2.6. Other unreported robustness checks

Tran et al. (2018) find that the 2008 global crisis affects the relationship between ownership structure and liquidity of Vietnamese
20
Following Jankensgård and Vilhelmsson (2018), we also use the number of shares as an exogenous instrument for shareholder base. In the first
stage, we regress number of shareholders against the number of shares and the same set of control variables. The instrumental variable is positively
associated with number of shareholders and statistically significant at the 1% level. In the second stage, we re-estimate the liquidity model (2) but
replace the number of shareholders with the predicted values from the first-stage regression. The results continue to show a nonlinear relationship
between shareholder base and liquidity. However, we do not report these 2SLS results due to earlier concern that the number of shares is not strictly
exogenous (Ding et al., 2016; El-Nader, 2018). Our concern is vindicated as the specification test rejects the null that the instrument is valid.
16
Table 7
Robustness check on endogeneity.
Firm fixed effects 2SLS System GMM
First-stage (lnNSH) Second-stage (CPQS)
lnNSH −7.5051⁎⁎⁎ −3.9524⁎⁎

(0.5465) (1.7321)
lnNSH2 0.3712⁎⁎⁎ 0.1991⁎
(0.0319) (0.1064)
lnNSHt-2 0.9056⁎⁎⁎
(0.0280)
lnNSHt-3 −0.0479⁎⁎
(0.0220)
lnNSH (predicted) −7.2737⁎⁎⁎
(1.4452)
lnNSH2 (predicted) 0.3621⁎⁎⁎
(0.0831)
LIND −0.0492⁎⁎⁎ 0.0067⁎⁎⁎ −0.0559⁎⁎⁎ −0.0016
(0.0071) (0.0011) (0.0141) (0.0159)
LIND2 0.0006⁎⁎⁎ −0.0001⁎⁎⁎ 0.0007⁎⁎⁎ −0.0001
(0.0001) (0.0000) (0.0002) (0.0002)
LINST 0.0089⁎⁎⁎ −0.0001 0.0109⁎⁎⁎ 0.0107⁎⁎⁎
(0.0018) (0.0002) (0.0041) (0.0036)
ln(1 + ANALYST) −0.0740 0.0516⁎⁎⁎ 0.0066 −0.1113
(0.0606) (0.0107) (0.0865) (0.0863)
RETURN −0.1723 −0.3157⁎⁎⁎ −0.7314 −1.0804⁎⁎⁎
(0.2285) (0.0420) (1.1126) (0.2554)
VOL 1.9029⁎⁎⁎ 0.0037 1.9368⁎⁎⁎ 1.4073⁎⁎⁎
(0.0212) (0.0031) (0.1688) (0.0491)
TURNOVER −2.0101⁎⁎⁎ 0.0991⁎⁎⁎ −2.0715⁎⁎⁎ −1.6146⁎⁎⁎
(0.0644) (0.0153) (0.2839) (0.0958)
lnSIZE −0.4477⁎⁎⁎ 0.0759⁎⁎⁎ −0.5717⁎⁎⁎ −0.8142⁎⁎⁎
(0.0482) (0.0077) (0.1236) (0.1577)
BLOCK 1.0795⁎⁎⁎ −0.0552⁎⁎⁎ 1.2471⁎⁎⁎ 0.4095⁎
(0.1802) (0.0199) (0.2370) (0.2360)
lnBSIZE 0.3486⁎⁎ −0.0321⁎⁎ 0.4181 −0.3632
(0.1585) (0.0142) (0.3056) (0.2964)
BINDEP −0.4403 0.0017 −0.6542 −0.1834
(0.3298) (0.0315) (0.5199) (0.5650)
DUAL −0.2973⁎ 0.0008 −0.2190 −0.0804
(0.1720) (0.0179) (0.2269) (0.3113)
CHAIR −0.1205 0.0078 −0.0815 0.1708
(0.0773) (0.0077) (0.1169) (0.1848)
CPQSt-1 0.2542⁎⁎⁎
(0.0149)
CONSTANT 41.8808⁎⁎⁎ 0.3953⁎⁎⁎ 42.7258⁎⁎⁎ 28.3717⁎⁎⁎
(4.2652) (0.0869) (6.2487) (7.3079)
Year dummies Yes Yes Yes Yes
Industry dummies No Yes Yes Yes
N 10,664 9255 9222 9012
Adj. R2 0.6454 0.9258 0.6418
Sargan test (p-value) 0.0132 (0.9086) 64.9000 (0.9690)
AR(1) (p-value) −4.5119 (0.0000)
AR(2) (p-value) −1.0200 (0.3077)
Notes: This table addresses endogeneity using firm fixed effects estimator, two-stage least squares (2SLS) instrument variable regression and system
GMM. The dependent variable is Closing Percent Quoted Spread (CPQS), computed as the ratio of the difference of closing ask and closing bid prices
over the mid-point of these prices. NSH is the number of shareholders measured at year-end. LIND (LINST) is defined as the proportion of shares held
by local individuals (local institutions) relative to total shares outstanding. ANALYST is the number of security analysts issuing earnings forecasts for
a firm during a calendar year. Stock return (RETURN) is computed as the time series averages of daily returns for each year and each firm, whereas
return volatility (VOL) is the standard deviation of daily stock returns over the year. TURNOVER is computed as the annual average of daily turnover
ratios, defined as the number of shares traded scaled by the number of shares outstanding. Firm size (SIZE) is the book value of total assets at year-
end. BLOCK denotes blockholdings, defined as the total percentage share ownership held by shareholders with at least 5% of the outstanding shares.
The first column estimates the liquidity model (2) with a firm fixed effects estimator. In the 2SLS, two-year and three-year lagged values of the
number of shareholders are used as the instruments. The first stage involves regressing lnNSH against both instrumental variables and the same set
of control variables. The second stage then re-estimates the liquidity model (2) but replacing lnNSH and lnNSH2 with their predicted values from the
first stage regression. The final column specifies the liquidity model (2) as a dynamic panel model that is estimated with system GMM. AR(1) and AR
17
(2) tests are under the null of no first-order and second-order serial correlation, respectively, in the first-differenced residuals. Sargan test of over-
identifying restrictions is under the null that all instruments are valid.
To conserve space, the coefficients for year and industry dummies are not reported. Entries in parentheses are standard errors. N denotes the number
of observations.
⁎⁎⁎ ⁎⁎
firms. Even though its effects on Malaysia are less severe than the 1997 Asian financial crisis, a structural break analysis conducted by
Liew et al. (2016) reveals that the global crisis causes aggregate liquidity of the Malaysian stock market to drop sharply. The number
of shareholders, on the other hand, has been on a declining trend since 2000 as depicted in Fig. 1. There is thus a possibility our
baseline result in Table 3 is driven by the strong correlations detected during the global crisis. To address this concern, we re-estimate
the liquidity model (2) for three sub-periods: (i) 2000–2007 (before crisis); (ii) 2010–2015 (after crisis); (iii) 2000–2015 but ex-
cluding the crisis years of 2008–2009. In our unreported results, lnNSH and its squared term remain highly significant across all three
sub-periods, and the signs of their coefficients reinforce the existence of a U-shaped relationship. Thus, we establish the robustness of
our key finding across time, and this finding is not driven by the global crisis.
While our regression includes industry-specific dummies to control for time-invariant industry effects, we further address the
possibility that the nonlinear shareholder base-liquidity relationship might be driven by a few dominant industries. The liquidity
model (2) is re-estimated by industries but we exclude those with less than 100 firm-year observations. In our unreported results, both
lnNSH and lnNSH2 retain their significant coefficients in seven industries, with finance and construction the notable exceptions. Since
our sample excludes financial firms due to a different regulatory framework, it is thus reasonable to conclude the nonlinear re-
lationship is widespread across industries.
6. Further analyses
The richness of the data allows us to conduct further analyses to explore why liquidity declines when the number of shareholders
becomes too large, and to assess the rationale of Malaysian stock exchange's priority in boosting retail participation.
6.1. Is the negative liquidity effect attributable to informed trading?
Our battery of robustness checks establish the key finding of a nonlinear relationship between shareholder base and stock li-
quidity. Before the threshold level, the increase in liquidity due to a larger number of shareholders can be attributed to greater
investor recognition, information competition among informed traders or the liquidity trading of noise traders, consistent with
theoretical predictions and conventional wisdom that more shareholders are associated with higher liquidity. However, the U-shaped
curve suggests the dominance of negative liquidity effect after the threshold point which might arise from greater adverse selection
costs imposed by informed trading or higher volatility induced by noise trading. Identifying the source of this liquidity decline
requires our analysis to move beyond the size of the shareholder base to its composition of informed versus noise traders.
We explore whether greater adverse selection cost is responsible for the decreases in liquidity when the number of shareholders
expands beyond the threshold level. There is no empirical evidence to suggest who the privately informed investors are in the
Malaysian stock market since information-based trade is not directly observable. While different approaches have been used in the
literature to infer which investor groups are informed, we follow Easley et al. (1998) in employing the theoretically grounded
probability of information-based trading (PIN) which is designed to compute the proportion of trades motivated by private in-
formation (for theoretical grounds, see Easley et al., 1996). These authors explore the information role of financial analysts by
computing PIN for a sample of U.S. stocks that differ in analyst coverage. They find that a larger number of analysts is associated with
lower level of information-based trading, suggesting financial analysts generate more uninformed trade and their recommendations
are based on public instead of private information. Furthermore, several studies show that PIN is highly correlated with both the
adverse selection component of spread (Chung and Li, 2003; Brennan et al., 2016) and ex-ante firm characteristics associated with
information asymmetry (Aslan et al., 2011; Lai et al., 2014).
To identify informed investors in the Malaysian stock market, we obtain the annual PIN data, defined as the ratio of orders arise
from informed traders over total number of trades, for Malaysian stocks over a shorter sample period of 2000–2011 from Lai et al.
(2014).21 These authors compute two PIN measures using global stock transactions data provided by Thomson Reuters Tick History
(TRTH) database for 30,095 firms across 47 countries including Malaysia – the original PIN (Easley et al., 1996) and adjusted PIN
measures (Duarte and Young, 2009). We follow Lai et al. (2014) and Brennan et al. (2016) to assess the quality of these PIN estimates
for Malaysian stocks based on their correlations with three illiquidity measures (CPQS, Amihud illiquidity ratio and CPQS impact).
Our unreported results show that PIN has higher positive correlations with all three illiquidity measures (0.3011–0.3736) than
adjusted PIN (0.1094–0.1314), and hence the former is used in our regression.
Our model regresses PIN on investor types and a set of commonly used control variables with available data (see Brown and
Hillegeist, 2007; Aslan et al., 2011; Byun et al., 2011; Lai et al., 2014):
21
We thank Bohui Zhang for his generosity in sharing the PIN data for Malaysian firms from Lai et al. (2014).
18
Table 8
Number of shareholders and the probability of information-based trading (PIN).
Y.-E. Chia, et al.
Total number of Total local Local individual Local institution Local Local nominee Total foreign Foreign Foreign Foreign
shareholders investors government investors individual institution nominee
lnNSH −0.0365⁎⁎⁎ −0.0365⁎⁎⁎ −0.0352⁎⁎⁎ −0.0323⁎⁎⁎ −0.0072⁎⁎⁎ −0.0374⁎⁎⁎ −0.0198⁎⁎⁎ −0.0247⁎⁎⁎ −0.0159⁎⁎⁎ −0.0139⁎⁎⁎
(0.0030) (0.0030) (0.0029) (0.0034) (0.0023) (0.0033) (0.0018) (0.0021) (0.0023) (0.0015)
lnSIZE 0.0001 −0.0004 −0.0009 −0.0017 −0.0142⁎⁎⁎ −0.0010 −0.0035⁎ −0.0029 −0.0098⁎⁎⁎ −0.0060⁎⁎⁎
(0.0024) (0.0024) (0.0023) (0.0026) (0.0016) (0.0024) (0.0021) (0.0022) (0.0020) (0.0020)
lnAGE 0.0012 0.0007 0.0006 0.00004 −0.0053⁎⁎⁎ −0.0005 0.0044⁎⁎ 0.0054⁎⁎⁎ −0.0021 0.0026
(0.0021) (0.0021) (0.0021) (0.0021) (0.0020) (0.0019) (0.0021) (0.0020) (0.0023) (0.0022)
lnPRICE 0.0061⁎⁎⁎ 0.0061⁎⁎⁎ 0.0062⁎⁎⁎ 0.0110⁎⁎⁎ 0.0117⁎⁎⁎ 0.0054⁎⁎⁎ 0.0098⁎⁎⁎ 0.0099⁎⁎⁎ 0.0117⁎⁎⁎ 0.0104⁎⁎⁎
(0.0013) (0.0013) (0.0013) (0.0016) (0.0018) (0.0012) (0.0016) (0.0015) (0.0018) (0.0017)
ln(1 + ANALYST) −0.0085⁎⁎⁎ −0.0086⁎⁎⁎ −0.0093⁎⁎⁎ −0.0050 −0.0057 −0.0041 −0.0054 −0.0073⁎⁎ −0.0064 −0.0046
(0.0032) (0.0031) (0.0032) (0.0040) (0.0043) (0.0028) (0.0039) (0.0037) (0.0040) (0.0040)
BM −0.0007 −0.0006 −0.0005 −0.0007 0.0007 −0.0004 −0.0009 −0.0007 −0.0001 −0.0006
(0.0011) (0.0011) (0.0011) (0.0013) (0.0012) (0.0011) (0.0011) (0.0011) (0.0013) (0.0011)
LEV −0.0223⁎⁎⁎ −0.0216⁎⁎⁎ −0.0217⁎⁎⁎ −0.0293⁎⁎⁎ −0.0186⁎⁎⁎ −0.0159⁎⁎ −0.0240⁎⁎⁎ −0.0236⁎⁎⁎ −0.0225⁎⁎⁎ −0.0221⁎⁎⁎
(0.0067) (0.0067) (0.0067) (0.0069) (0.0067) (0.0065) (0.0066) (0.0064) (0.0069) (0.0067)
ROE −0.0029 −0.0030 −0.0027 0.0034 0.0056 −0.0041 0.0013 0.0003 0.0048 0.0022
(0.0042) (0.0043) (0.0042) (0.0043) (0.0042) (0.0043) (0.0041) (0.0038) (0.0041) (0.0042)
VOL 0.0056⁎⁎⁎ 0.0055⁎⁎⁎ 0.0054⁎⁎⁎ 0.0036⁎⁎⁎ 0.0029⁎⁎⁎ 0.0057⁎⁎⁎ 0.0045⁎⁎⁎ 0.0047⁎⁎⁎ 0.0035⁎⁎⁎ 0.0041⁎⁎⁎
(0.0014) (0.0014) (0.0014) (0.0012) (0.0011) (0.0015) (0.0013) (0.0013) (0.0012) (0.0012)
TURNOVER 0.0149⁎⁎⁎ 0.0149⁎⁎⁎ 0.0144⁎⁎⁎ 0.0125⁎⁎⁎ 0.0130⁎⁎⁎ 0.0192⁎⁎⁎ 0.0145⁎⁎⁎ 0.0144⁎⁎⁎ 0.0126⁎⁎⁎ 0.0149⁎⁎⁎
19
(0.0025) (0.0025) (0.0025) (0.0028) (0.0035) (0.0025) (0.0030) (0.0030) (0.0033) (0.0032)
BLOCK 0.0190⁎⁎⁎ 0.0195⁎⁎⁎ 0.0200⁎⁎⁎ 0.0212⁎⁎⁎ 0.0287⁎⁎⁎ 0.0186⁎⁎⁎ 0.0204⁎⁎⁎ 0.0206⁎⁎⁎ 0.0279⁎⁎⁎ 0.0221⁎⁎⁎
(0.0060) (0.0061) (0.0062) (0.0065) (0.0070) (0.0054) (0.0056) (0.0057) (0.0062) (0.0058)
KLCI 0.0007 −0.0003 −0.0004 −0.0022 −0.0045 −0.0013 0.0014 0.00002 −0.0021 −0.0002
(0.0050) (0.0050) (0.0051) (0.0054) (0.0061) (0.0050) (0.0057) (0.0054) (0.0061) (0.0059)
CONSTANT 0.6359⁎⁎⁎ 0.6428⁎⁎⁎ 0.6339⁎⁎⁎ 0.4969⁎⁎⁎ 0.5455⁎⁎⁎ 0.5753⁎⁎⁎ 0.4530⁎⁎⁎ 0.4486⁎⁎⁎ 0.4869⁎⁎⁎ 0.4509⁎⁎⁎
(0.0187) (0.0183) (0.0188) (0.0265) (0.0228) (0.0188) (0.0273) (0.0282) (0.0286) (0.0274)
Year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
N 4872 4872 4872 4872 4872 4872 4872 4872 4872 4872
F-statistic 57.88⁎⁎⁎ 57.51⁎⁎⁎ 56.85⁎⁎⁎ 46.76⁎⁎⁎ 34.82⁎⁎⁎ 56.26⁎⁎⁎ 48.83⁎⁎⁎ 51.28⁎⁎⁎ 40.31⁎⁎⁎ 44.37⁎⁎⁎
Adj. R2 0.2655 0.2642 0.2598 0.2156 0.1551 0.2758 0.2192 0.2358 0.1764 0.1989
Notes: This table presents the pooled OLS estimation results for model (5) where the dependent variable is the probability of information-based trading (PIN) over a shorter sample period of 2000–2011.
NSH is the number of shareholders at the aggregate and granular levels by investor types. Firm size (SIZE) is the book value of total assets at year-end whereas firm age (AGE) is the number of years since
incorporation prior to year-end. PRICE is computed as the time series averages of daily closing stock prices for each year and each firm. ANALYST is the number of security analysts issuing earnings
forecasts for a firm during a calendar year. Book-to-market (BM) is the ratio of book value of equity divided by the market value of equity, leverage (LEV) the ratio of book value of debts over the book
value of assets, and return on equity (ROE) the ratio of operating income scaled by the book value of equity. Return volatility (VOL) is the standard deviation of daily stock returns over the year.
TURNOVER is computed as the annual average of daily turnover ratios, defined as the number of shares traded scaled by the number of shares outstanding. BLOCK denotes blockholdings, defined as the
total percentage share ownership held by shareholders with at least 5% of the outstanding shares. Stock index membership (KLCI) is a dummy variable that takes a value of one if a stock is included in the
main index of the Malaysian stock market, and zero otherwise.
For brevity, year and industry dummies are suppressed. Double-clustered standard errors are reported in the parentheses. N denotes the number of firm-year observations.
⁎⁎⁎ ⁎⁎
PINit = 0 + 1 ln NSHit
+ 2 ln SIZEit + 3 ln AGEit + 4 ln PRICEit
+ 5 ln (1 + ANALYST )it + 6 BMit + 7 LEVit + 8 ROEit + 9 VOLit
J 1 T 1
+ 10 TURNOVERit + 11 BLOCK it + 12 KLCIit + j=1 13j IND j + t = 1 14t
YRt + it (5)
where ln refers to natural logarithm, the dependent variable is the probability of information-based trading (PIN) proposed by Easley
et al. (1996) and the key independent variable of NSH is proxied by the total number of shareholders and its components, with each
investor group entering the regression model separately. The control variables are firm size measured by book value of total assets
(SIZE), firm age in years since incorporation (AGE), time-series averages of daily closing prices (PRICE), number of security analysts
issuing earnings forecasts (ANALYST), ratio of book value to market value of equity (BM), leverage ratio (LEV), return on equity
(ROE), stock return volatility (VOL), stock turnover (TURNOVER), blockholder ownership (BLOCK) and index membership (KLCI)
that takes a value of one if the stock is a component of the main market index, and zero otherwise. INDj and YRt are industry and year
dummies, respectively.
The results for model (5) at the aggregate level and different investor groups are tabulated in Table 8. Across all columns, the
coefficients for the number of shareholders are negatively and significantly associated with PIN, suggesting that none of the investor
groups (total local investors, local individuals, local institutions, local governments, local nominees, total foreign investors, foreign
individuals, foreign institutions and foreign nominees) can be regarded as informed traders who act on private information. This
appears to contradict the inferences in Lim et al. (2016) and Liew et al. (2018) that foreign investors are informed traders in the
Malaysian stock market, derived from price delay measure and CPQS, respectively. Since PIN is designed to capture trading based on
private information, our findings imply that the informed trading of foreign investors documented in the above two studies is most
likely generated from their superior skills in processing public information rather than the access to firm-specific information. Only
the coefficient of BLOCK is significantly associated with higher probability of informed trading, which is consistent with the literature
that blockholders have privileged access to private firm-specific information (see Brockman and Yan, 2009; He et al., 2013). This
result, interpreted together with the negative relationship between blockholdings and liquidity in Table 3, suggests that the con-
sequence of higher PIN for blockholders is lower liquidity due to greater adverse selection costs. However, for our key variable of
shareholder base, the evidence in Table 8 rules out the possibility that the dominance of negative liquidity effect after the threshold
point for Malaysian public listed firms is attributed to informed trading.
6.2. Can volatility explain the negative relationship between shareholder base and liquidity?
We now address the second possibility of whether the drop in liquidity after the number of shareholders exceeds the threshold
point is attributable to higher volatility induced by noise trading. While market microstructure models predict a positive relationship
between noise trading and liquidity due to lower adverse selection costs (see Glosten and Milgrom, 1985; Holmström and Tirole,
1993), the negative effect can arise through the return volatility channel as predicted by noise trader models (De Long et al., 1990;
Shleifer and Summers, 1990; Barberis et al., 1998). This can be rationalized by the consensus in the literature that a higher volatility
is associated with lower liquidity (Stoll, 1978a, 1978b, 2000; Chung and Chuwonganant, 2014). In a recent paper, Jankensgård and
Vilhelmsson (2018) explore the relationship between shareholder base and return volatility. Contrary to theoretical predictions, the
authors find that volatility rises with an increase in the total number of shareholders, the number of large shareholders, the number of
small investors, and the number of institutional investors.
Since our results in Table 8 provide no evidence of trading on private information by any investor groups, it is imperative to
further explore the volatility channel since it is highly likely that they engage in noise trading. We follow Jankensgård and
Vilhelmsson (2018) to explore the relationship between shareholder base and return volatility by replicating their model subject to
data availability:
VOLit = 0 + 1 ln NSHit + 2 ln SIZEit + 3 LEVit + 4 CPQSit
+ 5 DIVIDENDit + 6 BMit + 7 INTANGIBLESit + 8 EARNINGSit
J 1 T 1
+ j = 1 9j
INDj + t = 1 10t
YRt + it (6)
where ln refers to natural logarithm, the dependent variable is return volatility computed as the standard deviation of daily stock
returns (VOL), and the key independent variable of NSH is proxied by the total number of shareholders and its components, with each
investor group entering the regression model separately. The control variables are firm size measured by book value of total assets
(SIZE), leverage ratio (LEV), Closing Percent Quoted Spread (CPQS), ratio of book value to market value of equity (BM), ratio of
intangible assets to book value of total assets (INTANGIBLES), ratio of operating income to book value of total assets (EARNINGS), as
well as the DIVIDEND dummy that takes a value of one if a firm pays dividend in a given year and zero otherwise. INDj and YRt are
industry and year dummies, respectively.
The results for model (6) at the aggregate level and different investor groups are tabulated in Table 9. Across all columns, the
coefficients for the number of shareholders are positively and significantly associated with return volatility, consistent with the
findings of Jankensgård and Vilhelmsson (2018) for the Swedish market. More specifically, volatility increases when the shareholder
base expands and this applies to the total number of shareholders and all investor groups – total local investors, local individuals,
local institutions, local governments, local nominees, total foreign investors, foreign individuals, foreign institutions and foreign
nominees. Since our earlier PIN results suggest none of the investor groups are informed traders but they are more likely to engage in
noise trading, the higher volatility is consistent with the prediction of noise trader models (De Long et al., 1990; Shleifer and
20
Y.-E. Chia, et al.
Table 9
Number of shareholders and stock volatility.
Total number of Total local Local individual Local institution Local Local nominee Total foreign Foreign Foreign Foreign
shareholders investors government investors individual institution nominee
lnNSH 0.4189⁎⁎⁎ 0.4227⁎⁎⁎ 0.3927⁎⁎⁎ 0.1328⁎⁎⁎ 0.1037⁎⁎⁎ 0.5215⁎⁎⁎ 0.1652⁎⁎⁎ 0.1764⁎⁎⁎ 0.1005⁎⁎⁎ 0.1299⁎⁎⁎
(0.0355) (0.0356) (0.0355) (0.0369) (0.0290) (0.0361) (0.0212) (0.0250) (0.0242) (0.0183)
lnSIZE −0.4097⁎⁎⁎ −0.4047⁎⁎⁎ −0.3912⁎⁎⁎ −0.3085⁎⁎⁎ −0.2738⁎⁎⁎ −0.4396⁎⁎⁎ −0.3569⁎⁎⁎ −0.3382⁎⁎⁎ −0.2820⁎⁎⁎ −0.3508⁎⁎⁎
(0.0278) (0.0279) (0.0272) (0.0305) (0.0287) (0.0314) (0.0300) (0.0274) (0.0286) (0.0329)
LEV 1.3760⁎⁎⁎ 1.3668⁎⁎⁎ 1.3635⁎⁎⁎ 1.3989⁎⁎⁎ 1.3839⁎⁎⁎ 1.3076⁎⁎⁎ 1.4300⁎⁎⁎ 1.4075⁎⁎⁎ 1.3969⁎⁎⁎ 1.4210⁎⁎⁎
(0.1503) (0.1497) (0.1498) (0.1548) (0.1530) (0.1467) (0.1576) (0.1548) (0.1578) (0.1597)
CPQS 0.1498⁎⁎⁎ 0.1501⁎⁎⁎ 0.1487⁎⁎⁎ 0.1414⁎⁎⁎ 0.1392⁎⁎⁎ 0.1597⁎⁎⁎ 0.1426⁎⁎⁎ 0.1432⁎⁎⁎ 0.1393⁎⁎⁎ 0.1424⁎⁎⁎
(0.0075) (0.0075) (0.0076) (0.0079) (0.0079) (0.0069) (0.0078) (0.0079) (0.0078) (0.0077)
DIVIDEND −0.7590⁎⁎⁎ −0.7585⁎⁎⁎ −0.7717⁎⁎⁎ −0.9202⁎⁎⁎ −0.9308⁎⁎⁎ −0.6897⁎⁎⁎ −0.8670⁎⁎⁎ −0.8675⁎⁎⁎ −0.9218⁎⁎⁎ −0.8754⁎⁎⁎
(0.0504) (0.0506) (0.0506) (0.0525) (0.0539) (0.0490) (0.0509) (0.0510) (0.0529) (0.0511)
BM −0.0002 −0.0019 −0.0033 0.0127 0.0111 0.0026 0.0199 0.0126 0.0145 0.0231
(0.0314) (0.0314) (0.0316) (0.0318) (0.0330) (0.0301) (0.0320) (0.0319) (0.0325) (0.0322)
INTANGIBLES 0.6552⁎⁎⁎ 0.6544⁎⁎⁎ 0.6680⁎⁎⁎ 0.8674⁎⁎⁎ 0.8482⁎⁎⁎ 0.5773⁎⁎ 0.7638⁎⁎⁎ 0.7704⁎⁎⁎ 0.8424⁎⁎⁎ 0.7548⁎⁎⁎
21
(0.2522) (0.2517) (0.2523) (0.2550) (0.2568) (0.2424) (0.2571) (0.2550) (0.2557) (0.2581)
EARNINGS −1.8619⁎⁎⁎ −1.8657⁎⁎⁎ −1.8931⁎⁎⁎ −2.2117⁎⁎⁎ −2.2669⁎⁎⁎ −1.7726⁎⁎⁎ −2.0921⁎⁎⁎ −2.0840⁎⁎⁎ −2.2580⁎⁎⁎ −2.1352⁎⁎⁎
(0.2769) (0.2767) (0.2783) (0.2708) (0.2659) (0.2617) (0.2689) (0.2735) (0.2627) (0.2644)
CONSTANT 2.7893⁎⁎⁎ 2.6926⁎⁎⁎ 2.8657⁎⁎⁎ 5.2654⁎⁎⁎ 5.4872⁎⁎⁎ 4.0882⁎⁎⁎ 5.7984⁎⁎⁎ 5.6250⁎⁎⁎ 5.5303⁎⁎⁎ 6.0106⁎⁎⁎
(0.4545) (0.4685) (0.4697) (0.3733) (0.3653) (0.3966) (0.3620) (0.3519) (0.3967) (0.3988)
Year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
N 12,953 12,953 12,953 12,953 12,953 12,953 12,953 12,953 12,953 12,953
F-statistic 378.91⁎⁎⁎ 379.62⁎⁎⁎ 377.49⁎⁎⁎ 364.08⁎⁎⁎ 365.49⁎⁎⁎ 397.26⁎⁎⁎ 366.48⁎⁎⁎ 364.32⁎⁎⁎ 366.27⁎⁎⁎ 367.11⁎⁎⁎
Adj. R2 0.6649 0.6649 0.6631 0.6489 0.6481 0.6764 0.6534 0.6530 0.6485 0.6529
Notes: This table presents the pooled OLS estimation results for model (6) where the dependent variable is stock return volatility (VOL). NSH is the number of shareholders at the aggregate and granular
levels by investor types. Firm size (SIZE) is the book value of total assets at year-end. Leverage (LEV) is the ratio of book value of debts over the book value of assets. Closing Percent Quoted Spread (CPQS)
is computed as the ratio of the difference of closing ask and closing bid prices over the mid-point of these prices. DIVIDEND is a dummy variable that takes a value of one if the firm pays a common
dividend in a given year. Book-to-market (BM) is the ratio of book value of equity divided by the market value of equity, INTANGIBLES the ratio of intangibles assets over the book value of total assets, and
EARNINGS the ratio of operating income scaled by the book value of total assets.
For brevity, year and industry dummies are suppressed. Double-clustered standard errors are reported in the parentheses. N denotes the number of firm-year observations.
⁎⁎⁎ ⁎⁎
Summers, 1990; Barberis et al., 1998). Hence, we attribute the decline in liquidity when shareholder base exceeds the threshold level
to higher volatility induced by noise trading.
6.3. Boosting retail participation in the Malaysian stock market
For the Malaysian stock market, Bursa Malaysia and Securities Commission Malaysia are the prime drivers of investor partici-
pation. Their policies place greater emphasis on boosting retail participation in small- and mid-sized listed firms (see Footnote 5).
Despite Malaysia having the highest level of fixed deposits in Southeast Asia, retail trading at Bursa Malaysia which stood at 22% as at
31 July 2018 is relatively low.22 The exchange operator and regulator have embarked on numerous initiatives over the years to spur
retail participation, with year 2018 witnessing a series of aggressive measures – relaxing margin financing rules, removal of intraday
short-selling restrictions, 3-year stamp duty waiver for trading of mid- and small-cap stocks, and 6-month waiver on trading and
clearing fees for new individual investors.
Motivated by the continuous efforts of Malaysian authorities to boost retail participation, this section conducts further analysis on
the liquidity role of individual investors. We re-estimate the liquidity model (2) but replace the key independent variable of number
of shareholders (NSH) with RETAIL, the latter proxied by the number of local individual account holders (in natural logarithm),
percentage of local individual account holders per total number of shareholders, and the average number of shares per local in-
dividual account holder (in natural logarithm). The first two proxies depict a U-shaped relationship with CPQS as shown in Table 10,
indicating an increase in liquidity with a higher number and percentage of local individual account holders before their respective
threshold levels. While Bursa Malaysia should be commended for their continuous efforts in boosting retail participation, there is a
caveat that liquidity declines when the firms have too many individual investors (i.e., exceeding 26,925 local individual account
holders or 74% of total shareholders) as their noise trading is expected to induce higher volatility. However, this should not be a
cause of concern because the average number of local individual investors for Malaysian public listed firms is 5190. Only 387 firm-
year observations exceed the threshold level, which represent 2.78% of the total observations, suggesting more efforts are needed to
expand the size of local individual investors in the Malaysian stock market.
Another insight from the last column of Table 10 is that the liquidity benefit will only kick in when the number of shares per local
individual account holder exceed the threshold level of 20,305 shares. Since the minimum trading unit for all stocks listed on Bursa
Malaysia is 100 shares, our finding implies that the policy focus should not be confined to just expanding the number of individual
account holders but also their shares per account as larger shareholding size exerts greater liquidity impact. This remains a challenge
because only 30.14% of total firm-year observations exceed the computed threshold level of 20,305 shares, while the average is
12,780 shares per local individual account holder.
7. Conclusion
This paper is motivated by the shareholder-boosting strategies actively pursued by corporate managers and stock exchange
regulators, partly driven by the anticipated improvement in liquidity, but not grounded on empirical studies which are constrained by
limited data on shareholder base. Using recently assembled firm-level data from the Malaysian stock market, we provide at least three
new insights into this scarce literature. First, the relationship between the number of shareholders and liquidity proxied by Closing
Percent Quoted Spread is nonlinear. The existence of a threshold point implies the potential costs of maintaining a very large
shareholder base can outweigh its associated benefits. The nonlinear relationship not only challenges the popular view that “more is
better”, but prescribes a threshold maximum level for the number of shareholders beyond which the liquidity of firms will decline.
Second, the explanatory power of shareholder base is not suppressed by percentage share ownership, and its economic effect on CPQS
is at least four times larger than the combined effects of all three ownership variables – local individual ownership, local institutional
ownership and blockholdings. The large economic effect and strong theoretical basis support the inclusion of shareholder base in
stock liquidity model. Third, the nonlinear relationship holds at aggregate and granular levels of shareholder base, suggesting the
dominance of the theoretical channels of investor recognition, information competition and liquidity trading before the threshold
level. However, when ownership becomes too dispersed, the negative effect of wider spread begins to kick in, which we find is not
due to greater adverse selection costs but higher volatility induced by noise trading.
The above findings have implications for the Malaysian stock exchange regulator and operator in the context of their continuous
efforts to expand investor participation in the local bourse, especially retail investors. The nonlinear relationship between share-
holder base and liquidity offers specific policy implications. First, the existence of a threshold level suggests liquidity will decline
when shareholder base becomes too large as the costs outweigh the benefits. However, since less than 5% of our firm-year ob-
servations exceed the computed threshold maximum level, it implies either the existing public policies by Malaysian authorities are
not adequate or there is a limit to what stock exchange can achieve. Hence, corporate managers of Malaysian public listed firms
should heed the repeated calls by Amihud and Mendelson (2000, 2008) to actively manage and expand their shareholder bases. Some
of the effective corporate practices in developed stock markets that are relevant in the Malaysian context include stock splits, noncash
shareholder perks, strategic corporate disclosures and investor relations programs. Second, for these shareholder-targeting strategies
to be effective, corporate managers should know the sizes of their shareholder base relative to the population of Malaysian public
22
https://www.nst.com.my/business/2018/09/408893/bursa-malaysia-provides-growth-opportunity-regulated-environment (retrieved on 8
September 2018).
22
Table 10
Local individual investors and stock liquidity.
Number of local individual account Percentage of local individual account Number of shares per local individual account
holders holders holder
RETAIL −5.9430⁎⁎⁎ −1.2170⁎⁎⁎ 2.2535⁎⁎

(1.0129) (0.2596) (0.9580)
RETAIL2 0.2913⁎⁎⁎ 0.0082⁎⁎⁎ −0.1136⁎⁎
(0.0607) (0.0017) (0.0508)
LIND −0.0496⁎⁎⁎ −0.0866⁎⁎⁎ −0.0946⁎⁎⁎
(0.0130) (0.0171) (0.0176)
LIND2 0.0006⁎⁎⁎ 0.0009⁎⁎⁎ 0.0010⁎⁎⁎
(0.0002) (0.0002) (0.0002)
LINST 0.0097⁎⁎⁎ 0.0076⁎⁎ 0.0105⁎⁎⁎
(0.0034) (0.0035) (0.0035)
ln(1 + ANALYST) −0.0940 0.0962 −0.0605
(0.0814) (0.0916) (0.0839)
RETURN −0.1357 0.1669 0.3569
(1.0636) (1.0016) (1.0689)
VOL 1.9131⁎⁎⁎ 1.8275⁎⁎⁎ 1.8520⁎⁎⁎
(0.1613) (0.1622) (0.1643)
TURNOVER −2.0663⁎⁎⁎ −1.9572⁎⁎⁎ −2.1470⁎⁎⁎
(0.2655) (0.2575) (0.2746)
lnSIZE −0.4830⁎⁎⁎ −1.0095⁎⁎⁎ −1.0803⁎⁎⁎
(0.1081) (0.1215) (0.1250)
BLOCK 1.2022⁎⁎⁎ 1.3446⁎⁎⁎ 1.5703⁎⁎⁎
(0.2558) (0.2373) (0.2293)
lnBSIZE 0.3936 0.3687 0.5473⁎⁎
(0.2537) (0.2843) (0.2679)
BINDEP −0.4557 −1.1238⁎⁎ −1.1169⁎⁎
(0.4792) (0.5075) (0.5245)
DUAL −0.3213 −0.1495 −0.2821
(0.2015) (0.1960) (0.2162)
CHAIR −0.1189 −0.0882 −0.1041
(0.1093) (0.1211) (0.1202)
CONSTANT 35.5070⁎⁎⁎ 54.8029⁎⁎⁎ 3.0836
(4.2188) (9.4045) (4.5094)
Year dummies Yes Yes Yes
Industry dummies Yes Yes Yes
N 10,664 10,664 10,664
F-statistic 177.26⁎⁎⁎ 172.38⁎⁎⁎ 168.59⁎⁎⁎
Adj. R2 0.6388 0.6303 0.6211
Notes: This table presents the pooled OLS estimation results for liquidity model (2) where the dependent variable is CPQS, but the key independent
variable of number of shareholders (NSH) is replaced with local retail participation (RETAIL). RETAIL is proxied by the number of local individual
account holders (in natural logarithm), percentage of local individual account holders per total number of shareholders, and the average number of
shares per local individual account holder (in natural logarithm). Closing Percent Quoted Spread (CPQS) is computed as the ratio of the difference of
closing ask and closing bid prices over the mid-point of these prices. LIND (LINST) is defined as the proportion of shares held by local individuals
(local institutions) relative to total shares outstanding. ANALYST is the number of security analysts issuing earnings forecasts for a firm during a
calendar year. Stock return (RETURN) is computed as the time series averages of daily returns for each year and each firm, whereas return volatility
(VOL) is the standard deviation of daily stock returns over the year. TURNOVER is computed as the annual average of daily turnover ratios, defined
as the number of shares traded scaled by the number of shares outstanding. Firm size (SIZE) is the book value of total assets at year-end. BLOCK
denotes blockholdings, defined as the total percentage share ownership held by shareholders with at least 5% of the outstanding shares. Board size
(BSIZE) is the total number of directors on a firm's board at year-end. Board independence (BINDEP) is proxied by the ratio of independent non-
executive directors over board size at year-end. CEO duality (DUAL) is a dummy variable that takes a value of one if the chief executive officer is also
the board chairman at year-end, zero otherwise. CHAIR is a dummy variable that takes a value of one if the board chairman is an independent non-
executive director at year-end, zero otherwise.
year observations.
⁎⁎⁎ ⁎⁎
23
listed firms. Since Bursa Malaysia is privy to shareholder information through the Central Depository System (CDS) operated by its
subsidiary Bursa Malaysia Depository, the stock exchange has the capacity to develop an online tool for corporate managers to input
their number of shareholders so as to determine where their firms are situated in the shareholder base distribution each year.23 As a
concluding remark, while we do not expect our findings to be generalizable to developed mature markets, the policy lessons from our
findings are likely to be applicable to other emerging markets with similar institutional and market features.
Acknowledgements
We are indebted to two anonymous referees and the associate editor of the journal for their constructive comments that have
improved the manuscript significantly. This paper is extracted from the PhD thesis of the first author that has benefited from feedback
given by internal assessors at FEA-UM. The second author acknowledges financial funding from the Fundamental Research Grant
Scheme (Grant No: FP020-2014B) provided by the Ministry of Higher Education Malaysia and the Population Studies Unit at
University of Malaya (Grant No: IF002-2014).
Appendix A. Definitions for All Variables.
Variable Definition
CPQS Our main proxy of liquidity is the Closing Percent Quoted Spread proposed by Chung and Zhang (2014), computed as the ratio of the
difference of closing ask and closing bid prices over the mid-point of these prices. The CPQS is computed using daily data, and then
averaged to obtain the liquidity estimates for each year and each stock.
ILLIQ Amihud (2002) illiquidity ratio is computed as the daily ratio of the absolute stock returns to the local currency trading volume. The
annual ILLIQ estimates for each stock are obtained by averaging the computed daily ratios across all trading days for each year.
CPQSIM The price impact version of the CPQS is computed as the daily ratio of the CPQS scaled by the local currency trading volume. The annual
CPQSIM estimates for each stock are obtained by averaging the computed daily ratios across all trading days for each year.
ln NSH Natural logarithm of the total number of shareholders measured at year-end. At the granular level, NSH is proxied by the number of all
local investors, number of local individuals, number of local institutions, number of local governments, number of local nominees, number
of all foreign investors, number of foreign individuals, number of foreign institutions and the number of foreign nominees.
ESHB Excess shareholder base is the residuals from model (4) that regresses the number of shareholders against firm age, return on equity, firm
size, book-to-market ratio, stock returns, turnover and return volatility.
LIND Local individual ownership is computed as the proportion of shares held by local individuals relative to total shares outstanding.
LINST Local institutional ownership is computed as the proportion of shares held by local institutions relative to total shares outstanding.
ln (1 + ANALYST) Natural logarithm of one plus the number of security analysts issuing earnings forecasts for a firm during a calendar year. Analyst coverage
is set equal to zero for a firm-year observation if a firm is not listed on the I/B/E/S database or does not have earnings forecasts for any
given year.
RETURN Stock return is computed as the time series averages of daily returns for each year and each firm.
VOL Return volatility is computed as the standard deviation of daily stock returns over the year.
TURNOVER Stock turnover is computed as the annual average of daily turnover ratios, defined as the number of shares traded scaled by the number of
shares outstanding.
ln SIZE Natural logarithm of firm size, measured by the book value of total assets at year-end.
BLOCK Blockholdings are defined as the total percentage share ownership held by shareholders with at least 5% of the outstanding shares.
ln BSIZE Natural logarithm of board size, measured by the total number of directors on a firm's board at year-end.
BINDEP Board independence is proxied by the ratio of independent non-executive directors over board size at year-end.
DUAL A CEO duality dummy variable which takes a value of one if the chief executive officer is also the board chairman at year-end, zero
otherwise.
CHAIR A dummy variable which takes a value of one if the board chairman is an independent non-executive director at year-end, zero otherwise.
ln AGE Natural logarithm of firm age, measured as the number of years since incorporation prior to year-end.
ROE Return on equity is computed as the ratio of operating income divided by the book value of equity.
BM Book-to-market is computed as the ratio of book value of equity divided by the market value of equity.
ln PRICE Natural logarithm of price, computed as the time series averages of daily closing stock prices for each year and each firm.
LEV Leverage is computed as the ratio of book value of debts over the book value of assets at year-end.
KLCI A dummy variable of stock index membership which takes a value of one if a stock is included in the main index of the Malaysian stock
market (namely Kuala Lumpur Composite Index prior to 6 July 2009, and FTSE Bursa Malaysia KLCI Index thereafter), and zero otherwise.
DIVIDEND A dummy variable which takes a value of one if the firm pays a common dividend in a given year, zero otherwise.
INTANGIBLES Intangibles is computed as the ratio of intangible assets divided by the book value of total assets.
EARNINGS Earnings is computed as the ratio of operating income divided by the book value of total assets.
RETAIL Retail participation is proxied by the natural logarithm of number of local individual account holders, the percentage of local individual
account holders per total number of shareholders, and the natural logarithm of number of shares per local individual account holder.
23
Our proposal draws from the online income comparison tools for individuals or households to check their positions in the income distribution of
a country, see https://www.saldru.uct.ac.za/income-comparison-tool/, https://www.ifs.org.uk/tools_and_resources/where_do_you_fit_in and
http://www.oecd.org/statistics/compare-your-income.htm (retrieved on 1 September 2019).
24
References
Abdi, F., Ranaldo, A., 2017. A simple estimation of bid-ask spreads from daily close, high, and low prices. Rev. Financ. Stud. 30 (12), 4437–4480.
Agarwal, P., 2007. Institutional ownership and stock liquidity. In: SSRN Working Paper, https://doi.org/10.2139/ssrn.1029395. available at.
Ahn, H.J., Cai, J., Hamao, Y., Melvin, M., 2014. Little guys, liquidity, and the informational efficiency of price: Evidence from the Tokyo Stock Exchange on the effects
of small investor participation. Pac. Basin Financ. J. 29, 163–181.
Akins, B.K., Ng, J., Verdi, R.S., 2012. Investor competition over information and the pricing of information asymmetry. Account. Rev. 87 (1), 35–58.
Al-Jaifi, H.A., Al-rassas, A.H., Al-Qadasi, A.A., 2017. Corporate governance strength and stock market liquidity in Malaysia. Int. J. Manag. Financ. 13 (5), 592–610.
Alvarez, R., Jara, M., Pombo, C., 2018. Do institutional blockholders influence corporate investment? Evidence from emerging markets. J. Corp. Finan. 53, 38–64.
Amihud, Y., 2002. Illiquidity and stock returns: Cross-section and time-series effects. J. Financ. Mark. 5 (1), 31–56.
Amihud, Y., Mendelson, H., 2000. The liquidity route to a lower cost of capital. J. Appl. Corp. Finan. 12 (4), 8–26.
Amihud, Y., Mendelson, H., 2008. Liquidity, the value of the firm, and corporate finance. J. Appl. Corp. Finan. 20 (2), 32–45.
Amihud, Y., Mendelson, H., Uno, J., 1999. Number of shareholders and stock prices: Evidence from Japan. J. Financ. 54 (3), 1169–1184.
Anchev, S., 2017. Investor base and stock return anomalies. In: SSRN Working Paper, https://doi.org/10.2139/ssrn.3048508. available at.
Armstrong, C.S., Core, J.E., Taylor, D.J., Verrecchia, R.E., 2011. When does information asymmetry affect the cost of capital? J. Account. Res. 49 (1), 1–40.
Aslan, H., Easley, D., Hvidkjaer, S., O’Hara, M., 2011. The characteristics of informed trading: Implications for asset pricing. J. Empir. Financ. 18 (5), 782–801.
Atawnah, N., Balachandran, B., Duong, H.N., Podolski, E.J., 2018. Does exposure to foreign competition affect stock liquidity? Evidence from industry-level import
data. J. Financ. Mark. 39, 44–67.
Attig, N., Fong, W.M., Gadhoum, Y., Lang, L.H.P., 2006. Effects of large shareholding on information asymmetry and stock liquidity. J. Bank. Financ. 30 (10),
2875–2892.
Attig, N., El Ghoul, S., Guedhami, O., 2009. Do multiple large shareholders play a corporate governance role? Evidence from East Asia. J. Financ. Res. 32 (4), 395–422.
Bae, K.H., Ozoguz, A., Tan, H., Wirjanto, T.S., 2012. Do foreigners facilitate information transmission in emerging markets? J. Financ. Econ. 105 (1), 209–227.
Barber, B.M., Odean, T., 2000. Trading is hazardous to your wealth: The common stock investment performance of individual investors. J. Financ. 55 (2), 773–806.
Barberis, N., Shleifer, A., Vishny, R., 1998. A model of investor sentiment. J. Financ. Econ. 49 (3), 307–343.
Benamraoui, A., Jory, S.R., Mazouz, K., Shah, N., Gough, O., 2019. The effect of block ownership on future firm value and performance. N. Am. J. Econ. Finan. 50
article no. 100982.
Benston, G.J., Hagerman, R.L., 1974. Determinants of bid-asked spreads in the over-the-counter market. J. Financ. Econ. 1 (4), 353–364.
Beyer, A., Larcker, D.F., Tayan, B., 2014. Does the composition of a company’s shareholder base really matter? In: SSRN Working Paper, available at. https://ssrn.
com/abstract=2474792.
Bharath, S.T., Jayaraman, S., Nagar, V., 2013. Exit as governance: An empirical analysis. J. Financ. 68 (6), 2515–2547.
Black, F., 1986. Noise. J. Finan. 41 (3), 528–543.
Blume, M.E., Keim, D.B., 2012. Institutional investors and stock market liquidity: Trends and relationships. In: SSRN Working Paper, https://doi.org/10.2139/ssrn.
2147757. available at.
Bodnaruk, A., Östberg, P., 2009. Does investor recognition predict returns? J. Financ. Econ. 91 (2), 208–226.
Bodnaruk, A., Östberg, P., 2013. The shareholder base and payout policy. J. Financ. Quant. Anal. 48 (3), 729–760.
Booth, J.R., Chua, L., 1996. Ownership dispersion, costly information, and IPO underpricing. J. Financ. Econ. 41 (2), 291–310.
Boubaker, S., Gounopoulos, D., Rjiba, H., 2019. Annual report readability and stock liquidity. Finan. Market. Inst. Instrum. 28 (2), 159–186.
Boubakri, N., Chen, R., El Ghoul, S., Guedhami, O., Nash, R.C., 2019. State ownership and stock liquidity: Evidence from privatization. In: SSRN Working Paper,
https://doi.org/10.2139/ssrn.2984587. available at.
Brennan, M.J., Franks, J., 1997. Underpricing, ownership and control in initial public offerings of equity securities in the UK. J. Financ. Econ. 45 (3), 391–413.
Brennan, M.J., Subrahmanyam, A., 1995. Investment analysis and price formation in securities markets. J. Financ. Econ. 38 (3), 361–381.
Brennan, M.J., Huh, S.W., Subrahmanyam, A., 2016. Asymmetric effects of informed trading on the cost of equity capital. Manag. Sci. 62 (9), 2460–2480.
Brockman, P., Yan, S.X., 2009. Block ownership and firm-specific information. J. Bank. Financ. 33 (2), 308–316.
Brockman, P., Chung, D.Y., Yan, S.X., 2009. Block ownership, trading activity, and market liquidity. J. Financ. Quant. Anal. 44 (6), 1403–1426.
Brown, G.W., 1999. Volatility, sentiment, and noise traders. Financ. Anal. J. 55 (2), 82–90.
Brown, S., Hillegeist, S.A., 2007. How disclosure quality affects the level of information asymmetry. Rev. Acc. Stud. 12 (2–3), 443–477.
Burnham, K.P., Anderson, D.R., 2004. Multimodel inference: Understanding AIC and BIC in model selection. Sociol. Methods Res. 33 (2), 261–304.
Bushee, B., 2004. Identifying and attracting the “right” investors: Evidence on the behavior of institutional investors. J. Appl. Corp. Finan. 16 (4), 28–35.
Bushee, B.J., Miller, G.S., 2012. Investor relations, firm visibility, and investor following. Account. Rev. 87 (3), 867–897.
Bushee, B.J., Noe, C.F., 2000. Corporate disclosure practices, institutional investors, and stock return volatility. J. Account. Res. 38, 171–202.
Byun, H.Y., Hwang, L.S., Lee, W.J., 2011. How does ownership concentration exacerbate information asymmetry among equity investors? Pac. Basin Financ. J. 19 (5),
511–534.
Carney, R.W., Child, T.B., 2013. Changes to the ownership and control of East Asian corporations between 1996 and 2008: The primacy of politics. J. Financ. Econ. 107
(2), 494–513.
Chan, K., Covrig, V., 2012. What determines mutual fund trading in foreign stocks? J. Int. Money Financ. 31 (4), 793–817.
Chang, E.C., Guo, C., Ren, J., 2013. Retail investor recognition and the cross-section of stock returns. In: SSRN Working Paper, https://doi.org/10.2139/ssrn.1998697.
available at.
Chen, H., Noronha, G., Singal, V., 2004. The price response to S&P 500 index additions and deletions: Evidence of asymmetry and a new explanation. J. Financ. 59 (4),
1901–1930.
Chen, H.L., Chow, E.H., Shiu, C.Y., 2015. The informational role of individual investors in stock pricing: Evidence from large individual and small retail investors. Pac.
Basin Financ. J. 31, 36–56.
Chiang, R., Venkatesh, P.C., 1988. Insider holdings and perceptions of information asymmetry: A note. J. Financ. 43 (4), 1041–1048.
Chia, Y.E., Lim, K.P., Goh, K.L., 2020. Liquidity and firm value in an emerging market: Nonlinearity, political connections and corporate ownership. N. Am. J. Econ.
Finan. 52, 101169.
Chichernea, D.C., Ferguson, M.F., Kassa, H., 2015. Idiosyncratic risk, investor base and returns. Financ. Manag. 44 (2), 267–293.
Choi, J.J., Sami, H., Zhou, H., 2010. The impacts of state ownership on information asymmetry: Evidence from an emerging market. China J. Account. Res. 3, 13–50.
Choi, J.J., Jin, L., Yan, H., 2013. What does stock ownership breadth measure? Rev. Finan. 17 (4), 1239–1278.
Chung, K.H., Chuwonganant, C., 2014. Uncertainty, market structure, and liquidity. J. Financ. Econ. 113 (3), 476–499.
Chung, K.H., Li, M., 2003. Adverse-selection costs and the probability of information-based trading. Financ. Rev. 38 (2), 257–272.
Chung, K.H., Zhang, H., 2014. A simple approximation of intraday spreads using daily data. J. Financ. Mark. 17 (1), 94–120.
Chung, K.H., McInish, T.H., Wood, R.A., Wyhowski, D.J., 1995. Production of information, information asymmetry, and the bid-ask spread: Empirical evidence from
analysts’ forecasts. J. Bank. Financ. 19 (6), 1025–1046.
Chung, K.H., Elder, J., Kim, J.C., 2010. Corporate governance and liquidity. J. Financ. Quant. Anal. 45 (2), 265–291.
Claessens, S., Djankov, S., Lang, L.H., 2000. The separation of ownership and control in East Asian Corporations. J. Financ. Econ. 58 (1–2), 81–112.
Corwin, S.A., Schultz, P., 2012. A simple way to estimate bid-ask spreads from daily high and low prices. J. Financ. 67 (2), 719–760.
Dang, T.L., Nguyen, T.H., Tran, N.T.A., Vo, T.T.A., 2018. Institutional ownership and stock liquidity: International evidence. Asia Pac. J. Financ. Stud. 47 (1), 21–53.
De Long, J.B., Shleifer, A., Summers, L.H., Waldmann, R.J., 1990. Noise trader risk in financial markets. J. Polit. Econ. 98 (4), 703–738.
Demsetz, H., 1968. The cost of transacting. Q. J. Econ. 82 (1), 33–53.
Ding, R., Hou, W., 2015. Retail investor attention and stock liquidity. J. Int. Finan. Market. Inst. Money 37, 12–26.
Ding, M., Suardi, S., 2019. Government ownership and stock liquidity: Evidence from China. Emerg. Markets Rev. 40 article no. 100625.
Ding, X.S., Ni, Y., Zhong, L., 2016. Free float and market liquidity around the world. J. Empir. Financ. 38, 236–257.
Ding, M., Nilsson, B., Suardi, S., 2017. Foreign institutional investment, ownership, and liquidity: Real and informational frictions. Financ. Rev. 52 (1), 101–144.
Dou, Y., Hope, O.K., Thomas, W.B., Zou, Y., 2018. Blockholder exit threats and financial reporting quality. Contemp. Account. Res. 35 (2), 1004–1028.
25
Duarte, J., Young, L., 2009. Why is PIN priced? J. Financ. Econ. 91 (2), 119–138.
Easley, D., O’Hara, M., 1987. Price, trade size, and information in securities markets. J. Financ. Econ. 19 (1), 69–90.
Easley, D., Kiefer, N.M., O’Hara, M., Paperman, J.B., 1996. Liquidity, information, and infrequently traded stocks. J. Financ. 51 (4), 1405–1436.
Easley, D., O’Hara, M., Paperman, J., 1998. Financial analysts and information-based trade. J. Financ. Mark. 1 (2), 175–201.
Edmans, A., Manso, G., 2011. Governance through trading and intervention: A theory of multiple blockholders. Rev. Financ. Stud. 24 (7), 2395–2428.
El-Nader, G., 2018. Stock liquidity and free float: Evidence from the UK. Manag. Financ. 44 (10), 1227–1236.
Fang, V.W., Noe, T.H., Tice, S., 2009. Stock market liquidity and firm value. J. Financ. Econ. 94 (1), 150–169.
Fang, V.W., Tian, X., Tice, S., 2014. Does stock liquidity enhance or impede firm innovation? J. Financ. 69 (5), 2085–2125.
Farrell, M., Green, T.C., Jame, R., Markov, S., 2018. The democratization of investment research: Implications for retail investor profitability and firm liquidity. In:
SSRN Working Paper, https://doi.org/10.2139/ssrn.3222841. available at.
Fong, K.Y.L., Holden, C.W., Trzcinka, C., 2017. What are the best liquidity proxies for global research? Rev. Finan. 21 (4), 1355–1401.
Foster, F.D., Viswanathan, S., 1994. Strategic trading with asymmetrically informed traders and long-lived information. J. Financ. Quant. Anal. 29 (4), 499–518.
Foucault, T., Sraer, D., Thesmar, D.J., 2011. Individual investors and volatility. J. Financ. 66 (4), 1369–1406.
Glosten, L.R., Milgrom, P.R., 1985. Bid, ask and transaction prices in a specialist market with heterogeneously informed traders. J. Financ. Econ. 14 (1), 71–100.
Gormley, T.A., Matsa, D.A., 2014. Common errors: How to (and not to) control for unobserved heterogeneity. Rev. Financ. Stud. 27 (2), 617–661.
Gow, I.D., Ormazabal, G., Taylor, D.J., 2010. Correcting for cross-sectional and time-series dependence in accounting research. Account. Rev. 85 (2), 483–512.
Goyenko, R.Y., Holden, C.W., Trzcinka, C.A., 2009. Do liquidity measures measure liquidity? J. Financ. Econ. 92 (2), 153–181.
Green, T.C., Jame, R., 2013. Company name fluency, investor recognition, and firm value. J. Financ. Econ. 109 (3), 813–834.
Griffin, J.M., Kelly, P.J., Nardari, F., 2010. Do market efficiency measures yield correct inferences? A comparison of developed and emerging markets. Rev. Financ.
Stud. 23 (8), 3225–3277.
Grossman, S.J., Hart, O.D., 1980. Takeover bids, the free-rider problem, and the theory of the corporation. Bell J. Econ. 11 (1), 42–64.
Grullon, G., Kanatas, G., Weston, J.P., 2004. Advertising, breadth of ownership, and liquidity. Rev. Financ. Stud. 17 (2), 439–461.
He, W., Li, D., Shen, J., Zhang, B., 2013. Large foreign ownership and stock price informativeness around the world. J. Int. Money Financ. 36, 211–230.
Heflin, F., Shaw, K.W., 2000. Blockholder ownership and market liquidity. J. Financ. Quant. Anal. 35 (4), 621–633.
Holden, C.W., Subrahmanyam, A., 1992. Long-lived private information and imperfect competition. J. Financ. 47 (1), 247–270.
Holmström, B., Tirole, J., 1993. Market liquidity and performance monitoring. J. Polit. Econ. 101 (4), 678–709.
Hong, H., Lim, T., Stein, J.C., 2000. Bad news travels slowly: Size, analyst coverage, and the profitability of momentum strategies. J. Financ. 55 (1), 265–295.
Hong, H., Kubik, J.D., Stein, J.C., 2008. The only game in town: Stock-price consequences of local bias. J. Financ. Econ. 90 (1), 20–37.
Jacoby, G., Zheng, S.X., 2010. Ownership dispersion and market liquidity. Int. Rev. Financ. Anal. 19 (2), 81–88.
Jain, P., Jiang, C., Mekhaimer, M., 2016. Executives’ horizon, internal governance and stock market liquidity. J. Corp. Finan. 40, 1–23.
Jankensgård, H., Vilhelmsson, A., 2018. The shareholder base hypothesis of stock return volatility: Empirical evidence. Financ. Manag. 47 (1), 55–79.
Jayaraman, S., Milbourn, T.T., 2012. The role of stock liquidity in executive compensation. Account. Rev. 87 (2), 537–563.
Jensen, M.C., Meckling, W.H., 1976. Theory of the firm: Managerial behavior, agency costs and ownership structure. J. Financ. Econ. 3 (4), 305–360.
Jiang, C.X., Kim, J.C., Zhou, D., 2011. Liquidity, analysts, and institutional ownership. Int. Rev. Financ. Anal. 20 (5), 335–344.
Jiao, Y., 2016. Investor competition over information and stock returns. In: SSRN Working Paper, https://doi.org/10.2139/ssrn.2448536. available at.
Kadlec, G.B., McConnell, J.J., 1994. The effect of market segmentation and illiquidity on asset prices: Evidence from exchange listings. J. Financ. 49 (2), 611–636.
Kale, J.R., Loon, Y.C., 2011. Product market power and stock market liquidity. J. Financ. Mark. 14 (2), 376–410.
Kang, W., Zhang, H., 2014. Measuring liquidity in emerging markets. Pac. Basin Financ. J. 27, 49–71.
Kaniel, R., Ozoguz, A., Starks, L., 2012. The high volume return premium: Cross-country evidence. J. Financ. Econ. 103 (2), 255–279.
Karolyi, G.A., Liao, R.C., 2017. The economic consequences of investor relations: A global perspective. In: SSRN Working Paper, available at. https://ssrn.com/
abstract=2591079.
Karpoff, J.M., Schonlau, R., Suzuki, K., 2018. Shareholder perks, ownership structure, and firm value. In: SSRN Working Paper, available at. http://ssrn.com/
abstract=2615777.
King, M.R., Segal, D., 2009. The long-term effects of cross-listing, investor recognition, and ownership structure on valuation. Rev. Financ. Stud. 22 (6), 2393–2421.
Kini, O., Mian, S., 1995. Bid-ask spread and ownership structure. J. Financ. Res. 18 (4), 401–414.
Laeven, L., Levine, R., 2008. Complex ownership structures and corporate valuations. Rev. Financ. Stud. 21 (2), 579–604.
Lai, S., Ng, L., Zhang, B., 2014. Does PIN affect equity prices around the world? J. Financ. Econ. 114 (1), 178–195.
Leaño, M., Pedraza, A., 2018. Ownership concentration and market liquidity: Evidence from a natural experiment. Econ. Lett. 167, 56–59.
Lee, J., Chung, K.H., 2018. Foreign ownership and stock market liquidity. Int. Rev. Econ. Financ. 54, 311–325.
Lehavy, R., Sloan, R.G., 2008. Investor recognition and stock returns. Rev. Acc. Stud. 13 (2–3), 327–361.
Lesmond, D.A., 2005. Liquidity of emerging markets. J. Financ. Econ. 77 (2), 411–452.
Li, X., Geng, Z., Subrahmanyam, A., Yu, H., 2017a. Do wealthy investors have an informational advantage? Evidence based on account classifications of individual
investors. J. Empir. Financ. 44, 1–18.
Li, F., Liu, M.H., Shi, E.Y., 2017b. Institutional ownership around stock splits. Pac. Basin Financ. J. 46, 14–40.
Liew, P.X., Lim, K.P., Goh, K.L., 2016. Aggregate liquidity for Malaysian stock market: New indicators and time series properties. Int. J. Econ. Manag. 10 (2), 297–319.
Liew, P.X., Lim, K.P., Goh, K.L., 2018. Foreign equity flows: Boon or bane to the liquidity of Malaysian stock market? N. Am. J. Econ. Financ. 45, 161–181.
Lim, K.P., Hooy, C.W., Chang, K.B., Brooks, R., 2016. Foreign investors and stock price efficiency: Thresholds, underlying channels and investor heterogeneity. N. Am.
J. Econ. Financ. 36, 1–28.
Lim, K.P., Thian, T.C., Hooy, C.W., 2017. Investor heterogeneity, trading account types and competing liquidity channels for Malaysian stocks. Res. Int. Bus. Financ.
41, 220–234.
Ma, R., Anderson, H.D., Marshall, B.R., 2019. Risk perceptions and international stock market liquidity. J. Int. Financ. Mark. Inst. Money 62, 94–116.
Marshall, B.R., Nguyen, N.H., Visaltanachoti, N., 2013. Liquidity measurement in frontier markets. J. Int. Financ. Mark. Inst. Money 27, 1–12.
Maury, B., Pajuste, A., 2005. Multiple large shareholders and firm value. J. Bank. Financ. 29 (7), 1813–1834.
Merton, R.C., 1987. A simple model of capital market equilibrium with incomplete information. J. Financ. 42 (3), 483–510.
Morck, R., Yeung, B., Yu, W., 2000. The information content of stock markets: Why do emerging markets have synchronous stock price movements? J. Financ. Econ. 58
(1–2), 215–260.
Mukherji, S., Kim, Y.H., Walker, M.C., 1997. The effect of stock splits on the ownership structure of firms. J. Corp. Finan. 3 (2), 167–188.
Ng, L., Wu, F., Yu, J., Zhang, B., 2016. Foreign investor heterogeneity and stock liquidity around the world. Rev. Finan. 20 (5), 1867–1910.
Parrino, R., Sias, R.W., Starks, L.T., 2003. Voting with their feet: Institutional ownership changes around forced CEO turnover. J. Financ. Econ. 68 (1), 3–46.
Peress, J., Schmidt, D., 2020. Glued to the TV: Distracted noise traders and stock market liquidity. J. Finan. 75 (2), 1083–1133.
Petersen, M.A., 2009. Estimating standard errors in finance panel data sets: Comparing approaches. Rev. Financ. Stud. 22 (1), 435–480.
Pham, P.K., Kalev, P.S., Steen, A.B., 2003. Underpricing, stock allocation, ownership structure and post-listing liquidity of newly listed firms. J. Bank. Financ. 27 (5),
919–947.
Piotroski, J.D., Roulstone, D.T., 2004. The influence of analysts, institutional investors, and insiders on the incorporation of market, industry, and firm-specific
information into stock prices. Account. Rev. 79 (4), 1119–1151.
Rhee, S.G., Wang, J., 2009. Foreign institutional ownership and stock market liquidity: Evidence from Indonesia. J. Bank. Financ. 33 (7), 1312–1324.
Richardson, S., Sloan, R., You, H., 2012. What makes stock prices move? Fundamentals vs. investor recognition. Finan. Anal. J. 68 (2), 30–50.
Roodman, D., 2009. How to do xtabond2: An introduction to difference and system GMM in Stata. Stata J. 9 (1), 86–136.
Rubin, A., 2007. Ownership level, ownership concentration and liquidity. J. Financ. Mark. 10 (3), 219–248.
26
Shleifer, A., Summers, L.H., 1990. The noise trader approach to finance. J. Econ. Perspect. 4 (2), 19–33.
Sias, R.W., Starks, L.T., Titman, S., 2001. The price impact of institutional trading. In: SSRN Working Paper, https://doi.org/10.2139/ssrn.283779. available at.
Spiegel, M., Subrahmanyam, A., 1992. Informed speculation and hedging in a noncompetitive securities market. Rev. Financ. Stud. 5 (2), 307–329.
Stoll, H.R., 1978a. The pricing of securiry dealer services: An empirical study of Nasdaq stocks. J. Financ. 33 (4), 1153–1172.
Stoll, H.R., 1978b. The supply of dealer services in securities markets. J. Financ. 33 (4), 1133–1151.
Stoll, H.R., 2000. Friction. J. Financ. 55 (4), 1479–1514.
Subrahmanyam, A., 1991. Risk adversion, market liquidity, and price efficiency. Rev. Financ. Stud. 4 (3), 417–441.
Tran, L.T.H., Hoang, T.T.P., Tran, H.X., 2018. Stock liquidity and ownership structure during and after the 2008 global financial crisis: Empirical evidence from an
emerging market. Emerg. Mark. Rev. 37, 114–133.
Vorst, P., 2017. Equity market competition and stock price crash risk. In: SSRN Working Paper, https://doi.org/10.2139/ssrn.2661580. available at.
Wagenmakers, E.J., Farrell, S., 2004. AIC model selection using Akaike weights. Psychon. Bull. Rev. 11 (1), 192–196.
Wang, Q., Zhang, J., 2015. Individual investor trading and stock liquidity. Rev. Quant. Finan. Acc. 45 (3), 485–508.
Wintoki, M.B., Linck, J.S., Netter, J.M., 2012. Endogeneity and the dynamics of internal corporate governance. J. Financ. Econ. 105 (3), 581–606.
Yung, K., Jian, Y., 2017. Effects of the shareholder base on firm behavior and firm value in China. Int. Rev. Econ. Financ. 49, 370–385.
Zheng, S.X., Li, M., 2008. Underpricing, ownership dispersion, and aftermarket liquidity of IPO stocks. J. Empir. Financ. 15 (3), 436–454.
27

Chia 2020

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chia 2020

Uploaded by

Copyright:

Available Formats

Emerging Markets Review 44 (2020) 100696

Contents lists available at ScienceDirect

Emerging Markets Review

More shareholders, higher liquidity? Evidence from an emerging

ARTICLE INFO ABSTRACT

2. Theoretical basis for hypothesis development

2.1. Investor recognition

2.2. Information competition

2.3. Adverse selection costs

2.4. Noise trading

3. Measurement of variables and the specification of liquidity model

3.1. Dependent variable of stock liquidity

3.2. Key independent variable of shareholder base

Fig. 1. Shareholdings for Malaysian public listed firms.

3.3. Control variables

3.4. Specification of the liquidity model

4.1. Data collection

CPQS 13,827 5.3370 7.2572 0.7200 1.3219 2.7051 6.0681 19.3767

4.2. Descriptive statistics

5. Nonlinear relationship between shareholder base and stock liquidity

5.1. Baseline results

lnNSH −1.1853⁎⁎⁎ −7.4241⁎⁎⁎ −7.2311

5.2. Robustness checks

5.2.1. Alternative measure of shareholder base

5.2.2. Alternative liquidity measures

Panel A: CPQS Panel B: CPQSIM Panel C: ILLIQ

ESHB −1.3638⁎⁎⁎ −1.4501⁎⁎⁎

5.2.3. Alternative estimation methods

5.2.4. Disaggregate investor types

10th 25th 50th 75th 90th

lnNSH −8.4520⁎⁎⁎ −1.2066⁎⁎⁎ −2.8287⁎⁎⁎ −4.7616⁎⁎⁎ −9.5784⁎⁎⁎ −14.5201⁎⁎⁎

Panel A: Descriptive statistics

Panel B: Disaggregate investor types

5.2.6. Other unreported robustness checks

First-stage (lnNSH) Second-stage (CPQS)

lnNSH −7.5051⁎⁎⁎ −3.9524⁎⁎

6.1. Is the negative liquidity effect attributable to informed trading?

6.3. Boosting retail participation in the Malaysian stock market

RETAIL −5.9430⁎⁎⁎ −1.2170⁎⁎⁎ 2.2535⁎⁎

Appendix A. Definitions for All Variables.

You might also like