large foreign ownership and stock return volatility in ....doc
DESCRIPTION
TRANSCRIPT
Large Foreign Ownership and Stock Return Volatility in Emerging Markets1
Donghui Li
Quang Ngoc Nguyen
Peter Kien Pham
School of Banking and Finance
The University of New South Wales
Sydney, NSW
Australia
and
Steven X. Wei2
School of Accounting and Finance
Hong Kong Polytechnic University,
Hung Hom, Kowloon
Hong Kong
This version: May 2006
1 We appreciate the helpful comments of Mike Firth, Toan Pham, Wilson Tong, and Yong Wang. Li, Nguyen and Pham acknowledge financial support from the University of New South Wales and Wei acknowledges financial support from the Hong Kong Polytechnic University Research Grant A-PG31. We bear responsibility for any mistakes and inaccuracies.
2 The corresponding author: Tel: (853) 27667056; fax: (852) 23309845.E-mail address: [email protected] (Steven X. Wei).
Large Foreign Ownership and Stock Return Volatility in Emerging Markets
Abstract
This paper documents a robust negative relationship between large foreign ownership (LFO) and firm-level stock return volatility by using a unique data set from 32 emerging markets. We demonstrate that our large foreign ownership measure substantially differs from the investibility measure used in the literature, so that our major finding makes an incremental contribution to the literature. We interpret this finding along three directions: (1) LFO stabilizes stock return as it can be viewed as an approximation to foreign direct investment (FDI), confirming the prediction about FDI made by Stiglitz (1999, 2000 and 2004); (2) LFO plays a role on improving firms’ corporate governance. This role stabilizes firm’s performances so as to reduce stock return volatility. In addition, we find that this role strongly exists in good corporate governance environments but it disappears in bad ones; and (3) LFO implies a broadened investor base so as to lower stock return volatility as theoretically predicted from Merton’s (1987) model of capital market equilibrium with incomplete information.
JEL classification: G15, O16Keywords: large foreign ownership, stock return volatility
2
It has been well established that capital market liberalization delivers substantial
benefits to emerging economies. Various international capital asset pricing models
(Stapleton and Subrahmanyan (1977), Errunza and Losq (1985), Eun and
Janakiramanan (1986), Alexander, Eun, and Janakiramanan (1987), and Stulz (1999a
and 1999b)) and supporting empirical evidence (Chari and Henry (2004), Henry
(2000 and 2003), Bekaert and Harvey (2000), Kim and Singal (2000)) show that stock
market liberalization reduces the cost of equity capital by allowing for risk sharing
between domestic and foreign agents. The financial development literature (Boyd and
Smith (1996), Levine and Zervos (1996 and 1998), and Rajan and Zingales (1998))
demonstrates that a reduction in the cost of capital facilitates investments in the
economy. Bekaert, Harvey, and Lundblad’s (2005) find significant increase in
economic growth following equity market liberalizations. Milton (2006) finds that are
open to foreign investors experience better operating performances in terms of
growth, investment, profitability, and efficiency.
However, past financial crises in emerging economies also initiated grave concerns
about capital market liberalization. Foreign investors were widely blamed for the
severity of capital flight out of crisis countries, which further amplifies stock return
volatility and worsen the crises. On the one hand, event studies such as Kim and
Singal (2000) and Bekaert and Harvey (1997 and 2000) document a decrease in the
volatility of stock returns following capital market liberalizations. From a theoretical
point of view, Merton (1987) implies that an increase in a stock’s investor-base would
lead to a decrease in the stock return volatility. His theory is supported by empirical
evidence from the Mexican stock market (Clark and Berko (1997)). On the other
hand, Bae, Chan, and Ng (2004) find a positive relation between foreign investment
restrictions (which reflects the extent or degree of liberalization) and stock return
volatility. They argue that when stocks are highly accessible to foreign investors, the
stock returns are subject to large world market exposure and therefore are vulnerable
to world market risk. Stiglitz (1999, 2000 and 2004) further argues that premature
financial market liberalization, which occurs without the support of well-functioning
institutions and appropriate regulations, makes the liberalizing country vulnerable to
financial crisis.
3
Most of previous studies of the impact of capital market liberalization on stock market
volatility focus on the degree of regulatory restrictions on foreign investment. They
ask what would happen when a country eases restrictions and allows foreign investors
to enter the domestic stock market. The nature of capital flows (short-term versus
long-term) and its impact on volatility in emerging capital markets remain not clearly
understood. Although the economic development literature has extensively discussed
the stability aspect of foreign investment and its effect on emerging economies
(Stiglitz (1999, 2000 and 2004)), the capital market liberalization literature seems to
ignore the important role of foreign investment stability on emerging markets.
Further, investability, as a proxy for the extent of liberalization (Edison and Warnock
(2002), Bae et al. (2004), and Chari and Henry (2004)), has its own problems. First, it
fails to account for the fact that foreign investors might not invest up to the limit
allowed by the regulation due to firms not belonging to the foreign investors’ areas of
interests. Second, stocks that do not pass the size and liquidity screenings will be
given a value of 0 for the investability measure by the Emerging Market Database
(EMDB) even though there might be foreign investment in those stocks.
In this study, we collect firm-level large foreign ownership (holding 5% or more of a
firm’s share) data from the OSIRIS and Lexis/Nexis databases and merge it with the
Emerging Market Database to form a unique annual large foreign ownership data set
for 1485 firms from 32 developing countries. Using the country-effect model, we
document a negative relation between stock return volatility and large foreign
ownership after controlling for firm size, turnover, and industry. This negative
relation is robust to using alternative measures of stock return volatility and large
foreign ownership, and dealing with the endogeneity problem to some extent.
Our study appears to be related to Bae, Chan, and Ng (2004). Looking at the cross-
section of individual stock return volatility over the period January 1989 – September
2000, Bae et al. (2004) find a positive relation between return volatility and the
investability of individual stocks. In detail, they classify stocks into three groups: non-
investable (foreigners may not own any of the stock), partially investable (foreigners
may own up to 50% of the stock) and highly investable (foreigners may own more
than 50% of the stock). They find that stocks in the highly investable group exhibit
higher return volatility than those in the non-investable group. This result leads us to
4
investigate the difference between the large foreign ownership variable used in our
study and the investability variable used in Bae et al.’s. Our analysis indicates that
investablity and large foreign ownership are two very different concepts: while
investability indicates how much of a local firm foreigners can legally own and are
subject to some screening criteria as defined by EMDB1, large foreign ownership
measures the actual shareholdings of all large foreign investors in a local firm,
regardless of the degree of investability of that firm2. The correlation between the two
variables in our sample is close to zero. Therefore, our major finding makes an
incremental contribution to the literature.
We further analyse and interpret our major finding along three directions. First, we
link our finding to Stiglitz’s (2000) foreign direct investment (FDI) story3. Though
Stiglitz strongly criticizes short-run speculative foreign capital flows to emerging
markets, he hypothesizes a stabilizing role of FDI in developing economies. Stiglitz
argues that FDI provides resources, technology, and the training of human capital. All
of these would increase a firm’s operating efficiency and reduce its total risk,
resulting in lower stock return volatility. Using 10% as a cut-off point for foreign
shareholding to construct a new large foreign ownership (LFO) variable, which at
least resembles FDI, we find that so-defined LFO is still negatively related to stock
return volatility, so as to confirm the prediction about FDI made by Stiglitz. This new
LFO variable, mimicking FDI, has a strong correlation coefficient of 0.98 with our
large foreign ownership defined in the beginning. In this sense, our large foreign
ownership variable can be viewed as an approximation to FDI. This partially explains
why large foreign ownership is negatively related to stock return volatility.
1 For stocks to be included in the investable series, not only must they be able to be legally held by foreigners, but they also have to meet size and liquidity screening criteria. The size criterion requires a stock to have a minimum investable market capitalisation of $50 million or more over the 12 months prior to the addition of the stock to the investable index. The investable market capitalization is determined after applying the foreign investment rules and after any adjustments due to cross-holdings or government ownership. The size criteria require that stock must have at least $20 million in trade over the prior year, and that it must be traded on at least half of the local exchange's trading days. Therefore, even when a stock can legally be held by foreigners, it will still be classified as non-investable according to the EMDB if it fails either the size or liquidity criteria.2 The degree to which foreign investors could invest in a local firm may not accurately reflect the reality of foreign investment in that firm as foreign investors may not invest up to the limit legally allowed by local governments. This problem with investability is acknowledged by Bae et al. (2004).3 The OECD (1999) defines a foreign direct investment as a foreign ownership of 10% or more of the ordinary shares or voting power of a local enterprise.
5
Second, our study can be explained by an improvement in corporate governance
following participation by foreign investors. Stulz (2005) argues that foreign investors
provide firms in emerging markets with the tools and incentives to improve corporate
governance. In a similar line, Kelley and Woidtke (2006), Rossi and Volpin (2004)
finds that investment in a firm by foreign investors leads to increased probability that
good investor protection mechanisms be implemented in the firm. This could be the
result of foreign investors’ demand for higher transparency, improved disclosure
rules, accountability of management, and better shareholder rights (Kim and Singal
(2000)). The improved corporate governance in a firm due to large foreign ownership
should lead to a reduction in the firm’s return volatility. Further investigations of the
role of large foreign ownership in different corporate governance environments
produce very interesting results. Large foreign ownership is related to return volatility
only in better corporate governance environments. There is no relation between large
foreign ownership and stock return volatility in worse corporate governance
environments. These findings hold for country-level as well as firm-level corporate
governance. We interpret the results as evidence that the quality of corporate
governance affects the foreign investors’ role in reducing stock return volatility.
Foreign investors have more flexibility to improve a firm’s risk in better corporate
governance environments, whereas their influence on the firm’s governance and
operation is weak in bad corporate governance environments.
Last, the negative relation between large foreign ownership and stock return volatility
could be explained by Merton’s (1987) investor-base broadening hypothesis. His
hypothesis implies that in a market with incomplete information, expanding investor
base would lead to lower stock return volatility. Increasing large foreign ownership
could be roughly understood as expanding the investor base of a firm.
The advantage of using the foreign ownership variable is that it measures the actual
presence of foreign investors in emerging markets. Most of previous papers, which
investigate the openness of foreign investment regulation, could only measure the
prospect of foreign presence in these countries. Therefore, our study contributes to the
capital market liberalization literature by providing new evidence on the benefits of
capital market liberalization. In addition, the differentiation between the “actual
foreign presence” and the “prospect of foreign presence” helps solve the liberalization
6
debate by implying that liberalization is not the only (and effective) path to reduce the
riskiness of firms in developing countries (and to ultimately promote economic
growth). The evidence in our paper further shows that a relaxation in foreign
investment restrictions does not automatically guarantee actual foreign capital flows
into emerging markets. Policies or reforms designed to improve investor protection,
enhance transparency, and better reporting regulations are needed as they provide the
real incentives for foreign investors to invest in the emerging markets. Shleifer and
Wolfenzon (2002) argue that countries with better investor protection have higher
interest rates4 and are consequently more attractive to international capital flows.
The outline of the paper is as follows. Section I describes the data and the summary
statistics. Section II explores the relationship between foreign ownership and
volatility. Section III compares the two variables of large foreign ownership and
investability. Section IV provides the economic explanations of our major finding.
Section V concludes the paper.
I. Data and Descriptive Statistics
A. Data sources and sample
To measure the large foreign ownership variable, we collect firm-level ownership data
from two main sources, the OSIRIS database provided by Bureau Van Dijk and
Lexis/Nexis (through which ownership data are obtained from Worldscope, Major
Companies Database and Thompson Financial’s Extel Cards)5. From these sources,
we obtain shareholder names and percentage shareholdings reported in the year 2002
for listed domestic firms in 33 developing countries, where detailed ownership data
4 In their model, total output is determined by the production technology and by agency costs (the waste or fines resulting from diversion). Even though firms in different countries have access to the same production technology, they differ in the severity of agency costs. In countries with better investor protection, the agency problem is less severe, so the effective production technology (net of agency costs) is more efficient. Countries with better investor protection then have a higher marginal product of capital and consequently higher interest rates.5 Multiple data sources allow outlying observations to be cross-checked. To further improve our coverage, we also obtain ownership data for many firms in Mexico, Malaysia, Pakistan, Sri Lanka, Singapore and Thailand directly from their annual reports, and for firms in India and Chile from their stock exchanges’ websites.
7
are available6. We measure large foreign ownership as the sum of foreign block
holdings, where a block is defined as a holding larger than or equal to 5% of the
firm’s issued shares. In their paper, Li, Moshirian, Pham and Zein (2006) also use the
same threshold to define a block holding. We later increase the threshold to 10% as a
robustness check.
The firm-level data, which are used to calculate stock return volatility, size, turnover,
and investability, are from Standard & Poor’s Emerging Market Database (EMDB).
The firms in analysis are the constituents of the EMDB’s S&P/IFCG indices, which
are the core of Standard & Poor's family of emerging markets indices. These indices
are intended to represent the performance of the most active stocks in their respective
markets and to be the broadest possible indicator of market movements. In 2002, the
aggregate market capitalization of S&P/IFCG index constituents ranged between 60%
and 75% of the total capitalization of all domestic listed shares on the local stock
exchange. The S&P/IFCG Composite Index included 33 emerging markets and
covered 1941 firms.
In their study of the relation between stock investability and stock return volatility,
Bae et al. (2004) analysed firms across 33 emerging markets. Although the
S&P/IFCG Composite Index in 2002 covered the same number of emerging markets,
it did not include Greece and Portugal due to these countries’ graduation to the
“developed economy” status7. Replacing Greece and Portugal in the S&P/IFCG
Composite Index are Bahrain and Oman.
The main reason for us to use the EMDB is that it has the investable weights for
individual firms across emerging markets. One important finding of this paper is
based on the analysis of the difference between large foreign ownership (and as
extension, actual foreign ownership) and a stock’s investability. EMDB applies two 6 During our data collection process, OSIRIS and Worldscope gradually updated their ownership database. As a result, some firms with no ownership data in 2002 were updated with such data in 2003. For the sake of providing better coverage, especially for countries with a small number of firms, we also include these firms in the sample.7 Portugal was removed from EMDB in March, 1999 and Greece was removed from the database in May, 2001. EMDB specifies two criteria for a market to graduate from S&P/IFCG coverage: i) GNP per capita must be in the high income economy range for three consecutive years, and ii) the investable market capitalization-to-GDP ratio must be in the top 25% in the emerging markets universe for three consecutive years.
8
tests to measure the investability of a stock. The first test of a stock’s investability is
to determine whether the market is open to foreign institutions. It assesses the extent
to which and the mechanisms foreign institutions can use to buy and sell shares on
local exchanges and repatriate capital, capital gains, and dividend income without
undue constraint. If foreign institutions can invest in the listed shares, then the second
test is applied by determining whether there is any corporate by-law or corporate
charter or industry limitation on foreign ownership of the stock. It then creates a
variable called the “degree open factor” with value from zero to one that indicates the
amount of the security foreigners may legally own (0.00 indicates that no share of the
stock is legally available to foreigners).
Rouwenhorst (1999) documents two problems associated with the EMDB. One
problem is the survivorship bias as EMDB select stocks based on firm size and
liquidity, which are probably correlated with the past performance of the companies.
The other is data error due to the existence of unreasonable figures in the database.
Bae et al. (2004) argue that for cross-sectional studies, the first problem is irrelevant
because they do not seek to explain the performance of the companies over time.
Neither do we in this paper. The second problem, however, is important as a few
return outlier observations could significantly bias the volatility upward. We follow
the procedure used in Bae et al.’s study to detect dubious return and volume
observations. The basic idea behind Bae et al.’s procedure in identifying the extreme
observations is that stock return and trading volume observations have to be
compatible because it is well documented that stock return volatility is positively
related to trading volume. For instance, if the absolute return of a stock in a month
falls into the top 0.1% tail of the distributions of all firm-month return observations in
a country, but the monthly turnover for the stock in that month is not in the top 10%
tail of the distributions of all firm-month turnover observations in the country, we will
cross-check the stock’s return in the month with the figure from Datastream. If the
two deviate by more than 5% or if the stock data are not available from Datastream,
the observation of the stock in that month will be discarded8.
8 For a detailed description of how the error-fixing process works, see Bae et al. (2004).
9
After deleting the outliers identified in the process described above, we merge the
data from the EMDB with those from the OSIRIS database. The merged sample
contains annual data for 2002 and covers 1485 firms across 32 emerging markets9.
B. Variables
We discuss the variables to be used in the analysis of this paper in order. For the
consideration of robustness, both the dependent variable, stock return volatility, and
the major independent variable, large foreign ownership, are measured in two
different ways. Control variables are also explained.
Stock return volatility: We calculate the return volatility of individual stocks by using
two alternative methods: the traditional standard deviation and the logarithm of
squared returns as also used in Bae et al. (2004).
Method 1: Volatility is the sample standard deviation of monthly stock returns.
where n is the number of monthly observations for each stock in 200210; is
the stock’s monthly return (after adjusting for dividends, stock-splits, etc.);
and, is the average of monthly stock returns over n months.
Method 2: Volatility is the logarithm of squared returns:
where n and are defined as above.
Large foreign ownership (LFO): We measure large foreign ownership as the sum of
foreign block holdings, where a block is defined as a holding larger than or equal to
5% of the firm’s issued shares. The large foreign ownership varies from 0% to 100%.
9 Saudi Arabia is missed out compared with the EMDB coverage.10 Normally, n equals to 12, which is the number of months in a year. However, for some stocks there are less than 12 monthly observations in a year due to delisting or late listing to the respective stock markets.
10
Large foreign ownership dummies: Stocks are classified into three groups by defining
the following two large foreign ownership dummy variables. DZO is a dummy variable
taking value 1 if LFO = 0 and taking value 0 otherwise. DPO is a dummy variable
taking value 1 if 0 < LFO ≤ 50% and assuming value 0 otherwise. If both dummies
take value 0, LFO for the corresponding observations are higher than 50%. Using
dummy variables is intended to captures a qualitative or non-linear relationship
between large foreign ownership and stock return volatility.
Size: As the market capitalizations of stocks in emerging markets are highly skewed,
the logarithm of market capitalization is used to proxy for the size variable in
regressions.
where is the stock’s monthly market capitalization.
Turnover: The monthly turnover of a share is the number of shares traded in a month
divided by the number of shares outstanding at the beginning of the month. The
turnover variable used in this paper is the average of monthly turnover:
Industry dummies: A more appropriate name should be Sector dummies as we use the
10 sectors in the Global Industry Classification System (GICS) to identify asset
classes. These sectors are Energy, Materials, Industrials, Consumer Discretionary,
Consumer Staples, Health Care, Financials, Information Technology,
Telecommunication Services, and Utilities. There are nine industry dummies as one
category of stocks is dropped out to avoid colinearity. The industry dummy variables
are set to 1 if the observation belongs to the relevant category.
Investability: The investability value of a stock is the average of the stock’s monthly
investable weights, which range from 0 to 1:
11
Investability dummies: Stocks are classified into three groups based on their
investability. Zero investability group includes stocks whose investability measures
are zero. Partial investability group consists of stocks with investability being higher
than 0 and lower or equal to 0.5. High investability group consists of stocks with
investability being higher than 0.5. There are two investability dummies as one
category of stocks is dropped out to avoid colinearity. In this study, the high
investability dummy is dropped out. The investability dummy variables are set to 1 if
the observation belongs to the relevant category.
C. Descriptive statistics
Table 1 shows the summary statistics of the emerging stock markets covered in the
sample. There are a total of 32 countries spanning across Asia, Europe, Latin
America, Africa, and Middle East. The number of stocks in each country ranges from
9 in Slovakia to 189 in China. The volatility measure, the standard deviation of
monthly stock returns in 2002, ranges from 4.95% in Morocco to 28.9% in
Zimbabwe. The median market capitalization is lowest in Sri Lanka with US$ 12.62
million and highest in Russia with US$ 3,920.96 million. With the exception of
Korea, Taiwan, and Turkey, all other emerging markets have median turnover of less
than 10% per month. South Africa and Korea are the two countries with the highest
average investable weight (0.79 and 0.73, respectively) whereas Bahrain, Colombia,
Jordan, Nigeria, Oman, Pakistan, Slovakia, Sri Lanka, Venezuela, and Zimbabwe
have an average investable weight of zero. It is noted that the statistics on the large
foreign shareholding of domestic firms are quite different from those on the
investability. The two countries with the highest average foreign shareholding are
Slovakia (43.84%) and Argentina (40.21%) and none of the countries in the sample
have the average foreign ownership of zero.
II. Regression analysis
12
This section documents a negative relationship between stock return volatility and
large foreign ownership. To do so, we form two country-effects regression models:
(1)
(2)
where i represents stock i, j represents country j, and all the variables are as defined in
part B of Section I.
The reason for us to use the country-effect regression technique versus simple cross-
section technique is that stocks in the same country are more homogeneous than
stocks from different countries. This is because different countries have different sets
of law and regulation, various stages of financial development, and different degrees
of corporate governance and macro-economic policies, etc. In addition, Stulz (2005)
argues that country attributes are critical to financial decision-makings because of the
“twin agency problems”, which are different across countries. It is well-known that
fixed effects (here, country-effects) models can deal with the heterogeneity issue well,
as they still produce consistent estimates of the mode parameters even if there are
latent and/or omitted country-related variables correlated with the explanatory
variables of the models.
A. Analysis using the standard deviation of return volatility
In this section we run regressions (1) and (2) with the volatility calculated using
Method 1 (the sample standard deviation of monthly stock returns). The results of
regression (1) are reported in columns (1a), (2a), and those of regression (2) are in
column (3a) of Table 2.
When large foreign ownership is the only explanatory variable, the coefficient on the
large foreign ownership variable is negative and significant at the 1% level (t-statistic
of -3.65 in Column (1a)). With the inclusion of control variables such as size,
13
turnover, and industry dummies, the negative relation between return volatility and
large foreign ownership still exists but it weakens. The estimated coefficient on large
foreign ownership is significant at 10% level (Column (2a)). In column (3a), where
the large foreign ownership dummies are used, the results show that stocks belonging
to the high – large foreign ownership group (i.e. large foreign ownership is greater
than 50%) have lower volatility than those belong to the other two large foreign
ownership categories. The coefficient on the zero – large foreign ownership is
positive and significant at the 5% level (t-statistic of 2.23), and the coefficient on the
partial – large foreign ownership is positive and significant at the 10% level (t-statistic
of 1.69).
B. Analysis using the logarithm of squared returns
Volatility measured by the standard deviation of monthly returns is highly skewed and
has very high kurtosis. The skewness and kurtosis of the volatility measured in this
way are 1.9764 and 10.1630, respectively. In this section, we estimate volatility by
using the logarithm of squared monthly returns (Method 2). The distribution of
volatility measured in this way is closer to normality with the skewness of -0.2871
and kurtosis of 3.8888. In their paper, Bae et al. (2004) also use the logarithm of
squared returns as a measure of stock return volatility. The results of regression (1)
are reported in columns (1b), (2b), and those of regression (2) are in column (3b) of
Table 2.
The results in columns (1b), (2b), and (3b) are quantitatively the same but are much
stronger than those in columns (1a), (2a), and (3a) in terms of the significance level.
For instance, column (2b) shows that the coefficient on large foreign ownership is
negative and significant at the 1% level. The value and t-statistic of the coefficient are
-0.0036 and -2.82, respectively. In column (2a), the value and t-statistic of the
coefficient on large foreign ownership are -0.0001 and -1.69, respectively.
C. Endogeneity
C.1 Instrumental variables regressions
14
As with studies on the effect of financial liberalization on economic growth, our study
also faces the endogeneity issue. Is the investment decision by foreign investors
exogenous or do they make investment in firms based on the firms’ corporate
governance? These concerns are highly relevant because investment environment is
an important factor in the investment decision-making process.
Addressing the endogeneity issue is difficult in our context as there is no prior theory
or empirical evidence that could help us identify a suitable instrument for large
foreign ownership. Therefore, we choose an ad-hoc approach, using the corporate
governance factors that are highly correlated with the large foreign ownership
variable. We employ the following corporate governance factors as instruments. The
first corporate governance variable is the ownership of the largest domestic
shareholders. These shareholders tend to be become controlling shareholders when
their ownership in firms surpasses certain levels. La Porta et al. (2002) argue that
controlling shareholders have the power to expropriate minority shareholders within
the constraints imposed by the law. Facing with the possible expropriation by the
largest domestic shareholder, foreign investors may choose to invest in firms where
the largest domestic shareholder has weak power of expropriation. Dahlquist and
Robertsson (2001) find that foreigners tend to underweight firms with a dominant
owner. Initial analysis shows that large foreign ownership and the largest domestic
shareholder’s ownership are highly negatively correlated (coefficient correlation of -
0.4898). The second corporate governance variable is the type of the controlling
shareholder, who is defined as the largest domestic shareholder among those with
more than 20% ownership11. The type of the controlling shareholder variable is a
dummy one, taking a value of 1 if the controlling shareholder is family or
government, and 0 otherwise. This correlation between this variable and large foreign
ownership is – 0.3964. The final corporate governance variable is the pyramidal
structure of the controlling shareholder’s ownership. We call this variable pyramid
dummy, which takes a value of 1 if the controlling shareholder owns shares through a
pyramid and 0 otherwise. La Porta et al. (1999) and Wolfenzon (1998) find that
controlling shareholders use pyramids to acquire power disproportionate to their cash
flow rights. Therefore the pyramidal structure of the controlling shareholder’s
11 La Porta et al. (1999) also uses the 20% cut off point. We follow La Porta et al.’s (2002, 1999) assumption of one controlling shareholder.
15
ownership could be an important factor which foreign investors take into
consideration before making an investment. The correlation efficient between
pyramid dummy and large foreign ownership in our sample is -0.1488.
We run six instrumental variable (IV) regressions of stock return volatility on large
foreign ownership. With the first three IV regressions, the volatility is estimated by
standard deviation of monthly returns. With the last three IV regressions, the volatility
is estimated by logarithm of squared monthly returns. The results are reported in
Table 3.
Columns (1a) and (1b) of Table 3 shows the IV regression results when the instrument
is largest domestic shareholder’s ownership. The coefficient on large foreign
ownership in each column is statistically smaller than 0. The t-statistic for the
coefficient is -2.29 in column (1a) and -2.36 in column (1b). Moving to the regression
results in columns (2a) and (2b) where the instrument is type of the controlling
shareholder, the relation between large foreign ownership and stock return volatility
is also negative (t-statistic of -2.90 and -2.27, respectively).
Columns (3a) and (3b) report the IV regression results when two instruments are used.
The two instruments are type of the controlling shareholder and pyramid dummy. The
reason we do not use pyramid dummy as the only instrument in our IV regressions is
that pyramid dummy has very low correlation with large foreign ownership. This
suggests the pyramidal structure of the controlling shareholder’s ownership is not a
stand-alone factor in the foreign investors’ investment decision making process. La
Porta et al. (1999) find that controlling shareholders in large corporations are usually
State or families. These shareholders largely use the pyramidal ownership structure to
gain control rights in firms in excess of their cash flow rights. A combination of the
type of the controlling shareholder and the pyramidal property of the controlling
shareholder’s ownership would strongly affect foreign investment in the firm. The
results in columns (3a) and (3b) confirm the findings in columns (1a), (2a), (1b), and
(2b). The t-statistic for the coefficient on large foreign ownership is -2.36 in column
(3a) and -2.04 in column (3b), suggesting that the negative relation between large
foreign ownership and stock return volatility is significant at 5% level.
16
In summary, the consistent theme across all six columns in Table 3 is that large
foreign ownership leads to a reduction in stock return volatility in emerging market.
C.2 The relation between 2002 large foreign ownership and 2003 stock return
volatility
As another attempt to deal with the endogeneity issue, we conduct a test of the
relation between large foreign ownership and stock return volatility where LFO is the
2002 data and stock return volatility is calculated based on the 2003 data. This is an
ad-hoc approach to solving the endogeneity issue. The intuition is that foreign
investors as at 2002 could not observe the 2003 stock return volatility and thus did not
make investment based on the 2003 stock return volatility12.
Table 4 reports the regression results where all the variables are measured as
described in part B of section I. The volatility, size, turnover, and industry dummies
are the 2003 data, whereas large foreign ownership and the large foreign ownership
dummies are the 2002 data. Columns (1a), (2a), and (3a) report the regression results
where volatility is estimated by standard deviation of monthly returns. Columns (1b),
(2b), and (3b) report the regression results where volatility is estimated by logarithm
of squared monthly returns. In column (1a), the coefficient on large foreign
ownership is -0.0004. Its t-statistic of -3.67 indicates that it is significant at 1% level.
When size, turnover, and industry dummies are present, the coefficient on large
foreign ownership is -0.0002 and significant at 5% level (t-statistic of -2.07, column
(2a)). These results are similar to those in columns (1a) and (2a) of Table 2, except
that the coefficient on large foreign ownership in column (2a) of Table 2 is significant
at 10% level. Moving to column (3a) of Table 4, we find that the coefficient on non-
large foreign ownership dummy is positive and significant at 10% level (t-statistic of
1.81), while the coefficient on partial-large foreign ownership is insignificant (t-
statistic of 1.16). This is different from the result in column (3a) of Table 2, where the
coefficient on partial-large foreign ownership is significant at 10% level (t-statistic of
1.69).
12
17
The slightly quantitative difference between the results in column (3a) of Table 4 and
column (3a) of Table 2 may be due to the excessive skewness of return volatility
measured by standard deviation of monthly returns. When volatility is measured by
logarithm of squared monthly returns, the coefficient on the partial-large foreign
ownership dummy is positive and significant at 5% level. Its value and t-statistic are
0.2076 and 2.05, respectively (column (3b)). The coefficient on the non-large foreign
ownership is also positive and significant at 1% level. These results imply that stocks
in the high-large foreign ownership group have lower volatility than those in the non-
large foreign ownership or partial-large foreign ownership groups. This finding is
consistent with that for the 2002 sample. The regression results when the continuous
large foreign ownership variable is used also show support for the conclusion that
large foreign ownership is negatively related to return volatility. The coefficient on
large foreign ownership in both columns (1b) and (2b) is negative and significant at
1% level. The corresponding t-statistics are -4.74 and -3.23.
In summary, Table 4 shows that the 2002 LFO is negatively related to the 2003 stock
return volatility. This result strengthens the causal relationship between large foreign
ownership and stock return volatility.
III. The Comparison of Large Foreign Ownership and Investability
A popular measure related to foreign investment in emerging markets is the
investability of a stock (Bae et. al. (2004), Chari and Henry (2004), and Edison and
Warnock (2002)). This measure indicates the percentage of a firm’s capitalization that
is available to foreign investors. It is fair to ask whether our finding is driven by or
related to investability. We address this issue in this section. Before we proceed, the
major points and results are summarized below:
Large foreign ownership measures the actual presence of large foreign
investors in a firm, while investability measures the degree the firm is
accessible to foreign institutional investors.
Investability is not a good measure of accessibility to foreign investors.
Using annual data, investability is not related to stock return volatility.
18
The first point is obvious from the definitions of large foreign ownership and
investability. As defined in Part B of Section I, large foreign ownership is the sum of
foreign block holdings, which have to be in existence for large foreign ownership to
be higher than 0. On the other hand, investability is determined based on the limit on
the amount of company capital foreign investors may hold, regardless of whether
foreign investors actually invest in the company. Thus, the investability of a stock
could take a value of 1 (that is, the stock is fully investable) while there is no foreign
investment in the stock. Panel B of Table 5 shows 225 cases where the investability of
a stock is higher than 0.5, i.e. stocks are highly accessible to foreign investors, but its
large foreign ownership is 0. Although large foreign ownership, which excludes
foreign block shareholdings of less than 5%, does not exactly represent the actual total
foreign ownership, Panel B gives an indication that in reality, there could exist stocks
with high investability and zero foreign ownership. In other words, Panel B implies
that easing foreign investment restrictions does not necessarily attract foreign
investment in domestic firms. Panel A of Table 5 provides further evidence with only
16 out of 1485 stocks actually having more than 50% foreign ownership and high
investability (that is, investable weights larger than 0.5).
The second point is evidenced through Panel C of Table 5. There are a total of 55
stocks where large foreign ownership in each stock is higher than 50% but the
investability for each stock is 0. There are two possible reasons for the mismatch
between large foreign ownership and investability. Firstly, some stocks are currently
owned by foreign investors however they do not pass the size and liquidity tests as
specified by EMDB, they are assigned investable weights13 of 0 (see footnote 3).
Secondly, some stocks are owned by a local firm, which is a wholly-owned subsidiary
of a foreign entity. EMDB excludes the local firm’s ownership of those stocks in their
calculation of the stocks’ investability, whereas our large foreign ownership variable
regards the local firm as if it was a foreign firm, and accordingly, the firm’s
ownership of those stocks as foreign ownership. In addition, investability does not
exclude family or individual domestic block shareholdings. Failing to take into
account these block shareholdings overestimates the investable market capitalization
that is available to foreign investors.13 We use investability and investable weight interchangeably.
19
The third point comes from the regression results in Table 6. In columns (1a) and (2a)
of Table 6, which correspond to regression (1) and where investability is treated as a
continuous variable (see part II), the coefficients on investability in both columns are
insignificant. The corresponding t-statistics are 0.36 and 0.76, respectively. When
volatility is measured by logarithm of squared returns, the coefficient on investability
is also insignificant. Its t-statistic is 0.74 in column (1b) and 0.73 in column (2b). The
regression results in columns (1a), (2a), (1b), and (2b) show that investability is not
related to stock return volatility.
Bae et al. (2004), however, find that investability is positively related to stock return
volatility. In their paper, they do not use the continuous investability. Instead, they use
three investability dummies, non-investability where investable weight is equal to 0,
partial-investability where investable weight ranges from above 0 and up to 0.5
(inclusive), and high-investability where investable weight is higher than 0. To
examine whether our findings in columns (1a) and (2a) are driven by the use of
continuous investability, we also create three investability dummies based on Bae et
al.’s methodology. The regression results are reported in column (3a) and (3b). Note
that the high-investability dummy is dropped out to avoid linearity among the
investability dummies. In column (3a), the coefficient on the non-investability dummy
is -0.006. Its t-statistic is only -1.28, indicating that stocks in the high-investability
and non-investability groups have similar return volatility. The coefficient on the
partial-investability dummy and its t-statistic are -0.0048 and -1.05, respectively. The
low absolute value of t-statistic shows that stocks in the high-investability and partial-
investability groups have similar return volatility. Column (3b) reveals similar
findings to those in column (3a). In summary, the results in column (3a) and (3b)
imply that investability is not related to return volatility14.
Overall, the results from Table 2 and Table 6 prove that large foreign ownership and
investability are two different variables. The next issue is whether investability will
14 Using monthly data for the period Jan 1989 – Sep 2000, Bae et al. find a positive relation between investability and stock return volatility. This prompts us to test the investability – volatility relation using the monthly data from firms in our sample. The regression result confirms the positive relation between stock investability and stock return volatility. This, coupled with the results from Table 4, means that the relation between investability and stock volatility is significant at smaller interval (monthly) rather than larger interval (yearly).
20
take away all (or part) of large foreign ownership’s explanatory power of the variation
in stock volatility if both variables appear in the same regressions. We study this issue
by including both investability and foreign ownership as independent variables in
regressions. The volatility measure used in the regressions is calculated using
logarithm of monthly return15.
Table 7 reports four different regressions with different combinations of large foreign
ownership and investability variables. Looking across the first two columns (1) & (2)
where large foreign ownership is a continuous variable, the coefficient on large
foreign ownership is negative and significant at the 1% level. The t-statistic for the
coefficient on large foreign ownership is -2.78 in column (1) and -2.80 in column (2).
The values of the coefficient on large foreign ownership in those two columns are
both -0.0036. This figure is the same as the coefficient on large foreign ownership in
column (2b) of Table 2, suggesting that the investability measures, whether
continuous or dummies, do not diminish the explanatory power of large foreign
ownership.
The results in columns (3) and (4) support the conclusion above. In these columns, we
use large foreign ownership dummies instead of continuous large foreign ownership.
The coefficient on non-large foreign ownership in column (3) is 0.3016 and
significant at 1% level (t-statistic of 3.24). Its counterpart in column (4) has similar
value (0.3005) and t-statistic (3.21). The coefficient on partial-large foreign
ownership in column (3) is 0.2192 and significant at 5% level (t-statistic of 2.17). Its
counterpart in column (4) also has similar value (0.2143) and t-statistic (2.11).
Comparing the values of the coefficients on non-large foreign ownership and partial-
large foreign ownership dummies in columns (3) and (4) of Table 7 with their
counterparts in column (3b) of Table 2, we find that they almost the same. This
suggests that the presence of investability measures in the regression models do not
weaken the power of large foreign ownership in explaining stock return volatility.
IV. Economic Explanations
15 Although we choose to demonstrate the regression results where the logarithm of squared monthly return is used as a measure of volatility, those regression results where the standard deviation of monthly returns is used as a measure of volatility are quantitatively similar.
21
Given the negative relation between large foreign ownership and stock return
volatility, it is important and interesting to ask why this should be the case. In this
section, we analyse this issue along three directions. First, we interpret our finding as
to be consistent with the foreign direct investment (FDI) story by Stiglitz (2000).
Second, we investigate the influence of large foreign ownership on stock return
volatility under different scenarios of corporate governance environments. Third, we
interpret the finding through Merton’s (1987) model of capital market equilibrium
with incomplete information. However, we admit that there might be other possible
explanations and likely we raise more questions than what we are trying to answer in
this section. Consequently, we invite more future research on these issues.
A. Link to Stiglitz’s (2000) FDI story
A.1 Stiglitz’s FDI story
The experience of the 1997 Asian financial crisis leads many people to blame foreign
investors for the excessive volatility. Stiglitz (1998) says that “developing countries
are more vulnerable to vacillations in the international flows than ever before”. Bae et
al. (2004) provide supporting evidence, which is based on the measure of
investability. The argument against foreign investment is that short-term capital flows,
or “hot money”, come and goes quickly, generating large price fluctuations. However,
Stiglitz (2000) makes a strong distinction between (short-term) speculative foreign
capital flows and (long-run) foreign direct investment. An FDI represents a strategic
investment by a foreign investor in a local firm and “brings with it not only resources,
but technology, access to markets, and (hopefully) valuable training, an improvement
in human capital” (Stiglitz (2000)). Stiglitz develops a hypothesis on the stabilizing
role of the FDI. We first examine his hypothesis within our sample.
A.2 Constructed FDI and stock return volatility
To align with the discussion of FDI above, we change the cut-off point for a block
shareholding from 5% to 10%. By this way, a block shareholding will resemble a
22
foreign direct investment16. The large foreign ownership variable, under the new
definition of block shareholding, is likely more stable than the old one as an FDI is
renowned for its stability.
With the new large foreign ownership variable, we repeat the country-effect
regressions (1) and (2) and report the results in Table 8. It is clear to see that Table 8
is almost a mirror of Table 2 with the coefficients and the relevant statistics in each
column are similar to those in the counterpart column. In other words, Table 8
confirms the earlier results that large foreign ownership is negatively related to stock
return volatility.
B. Corporate governance role
The results in section III are also consistent with Stulz’s (2005) theory of twin agency
problems. Stulz argues that foreign investment will lower the agency costs of
corporate insider discretion and those of state ruler discretion by providing the
governance mechanisms and incentives for firms to improve their corporate
governance17. In this section, we investigate how the role of foreign investors varies
varies in different macro corporate governance environments.
B.1 Firm-level governance environments
We examine the role of large foreign ownership in three different scenarios. In the
first scenario, we classify firms into two groups based on the presence of the
controlling shareholders in the firms. One group includes firms with a controlling
shareholder and the other consists of firms without a controlling shareholder. We then
run the country-effect regressions of stock return volatility on large foreign ownership
for each group. The regression results for the non-controlling shareholder group are 16 We do not call a block shareholding defined in this way a foreign direct investment because in many cases, block shareholdings are owned by foreign portfolio managers. We notice, however, that standard definition of foreign direct investment implies a foreign block shareholding a foreign direct investment.17 The first problem occurs when corporate insiders, or those who control firms, expropriate private benefits to maximize their own welfare at the expense of the firms’ investors. The second problem occurs when state rulers use the powers of the state to expropriate investors, such as confiscating the firms’ assets, changing regulations in favour of the constituents of the current rulers of the state, etc. In the environment of severe twin agency problems, ownership concentration becomes common to counteract the problems. Claessens, Djankov, and Lang (2000), Faccio and Lang (2002) show that outside the United States and United Kingdom, firms are typically controlled by large shareholders.
reported in column (1a) of Panel A and Panel B (Table 9). The results for the other
group are reported in column (1b) of these panels.
Similarly, in the second scenario, we classify firms into two groups based on the
pyramidal structure of the controlling shareholder’s ownership. If the controlling
shareholder owns the shares through a pyramid, the firm is assigned to the pyramid
group. All other firms, including those which do not have a controlling shareholder,
are assigned to the non-pyramid group. We then run the country-effect regressions of
stock return on large foreign ownership for each group. The regression results for the
non-pyramid group are reported in column (2a) of Panel A and Panel B. The results
for the pyramid group are reported in column (2b) of these panels.
In the final scenario, we classify firms into two groups based on the type of the
controlling shareholder. Firms belong to the family group if the controlling
shareholders are individuals, families or the State. All other firms belong to the non-
family group. We then run the country-effect regressions of stock return on large
foreign ownership for each group. The regression results for the non-family group are
reported in column (3a) of Panel A and Panel B. The results for the family group are
reported in column (3b) of these panels.
Panel A shows that the coefficient on large foreign ownership is statistically
indifferent from 0 across almost all groups. The exception is with the non-pyramid
group, in which the coefficient on large foreign ownership is negative and significant
at the 10% level (t-statistic of -1.80, column (2a)). However, moving to Panel B,
where stock return volatility is estimated by logarithm of squared returns, we find that
the coefficient on large foreign ownership is negative and significant for one group
and insignificant for the other group in each scenario. The regression results for firms
in the non-controlling shareholder group show that large foreign ownership is
negatively related to stock return volatility, while the results for firms in the other
group show no relation between those two variables. The t-statistic for the coefficient
on large foreign ownership is -2.67 in column (1a) and 0.14 in column (1b). These
results imply that when there is a presence of a controlling shareholder, the influence
of foreign investment is weakened.
Columns (2a) and (2b) show that large foreign ownership is negatively related to
stock return volatility in the non-pyramid group, but not in the pyramid group. The
regression for the non-pyramid group produces a large foreign ownership coefficient
of -0.0043 and a corresponding t-statistic of -3.39. In contrast, the regression for the
pyramid group returns a large foreign ownership coefficient of -0.0023 and a
corresponding t-statistic of -0.22. These results indicate that for cases in which
controlling shareholders employ pyramids to gain control in firms, the influence of
large foreign ownership is weakened.
In the third scenario, in which stocks are sorted based on the type of the controlling
shareholder, we find that large foreign ownership is negatively related to stock return
volatility for the non-family group. The t-statistic of the coefficient on large foreign
ownership is -2.22 (column (3a)). In contrast, we find no significant relation between
large foreign ownership and return volatility for the family group. The relevant t-
statistic is only -0.57 (column (3b)). These results suggest that for cases in which the
controlling shareholder is individuals, families, or the State, the influence of large
foreign ownership is weakened.
The results in Panel B are consistent with findings in La Porta et al. (2002, 1999) and
Wolfenzon (1999). These authors find that controlling shareholders have the power to
expropriate minority shareholders. The power of expropriation is higher through the
use of pyramids. Controlling shareholders are often families, who participate in the
management of the firms they own and whose control of firms is unchallenged by
other equity holders. We show that in those cases the role of large foreign ownership
almost disappears.
Why do Panel A and Panel B give different implications on the role of foreign
investors in different governance environment? We argue that it is because volatility
estimated using the standard deviation method is far from normally distributed. The
skewness and kurtosis of the volatility measured in this way are 1.9764 and 10.1630,
respectively. In contrast, the distribution of the volatility estimated using the
logarithm method is closer to normality with a skewness of -0.2871 and kurtosis of
3.8888.
B. Country-level governance environments
In this section, we analyse four different country-level corporate scenarios based on
four alternative measures of macro corporate governance environment. First, we use
the Minority Rights index to measure both the existence and the degree of
enforcement of shareholder rights. This index is formulated based on the survey of
world business leaders in the World 14 Economic Forum’s Global Competitiveness
Report 2003. Second, we use the Financial Disclosure index to measure the ability to
access sufficient, accurate, and timely corporate information by shareholders. This
index is also from the survey data from the World Economic Forum’s Global
Competitiveness Report 2003. Third, we use La Porta anti-director rights index to
measure the degree to which a macro governance environment protects voting rights
of minority shareholders and offers them avenues to challenge insiders in the
corporate decision making process. This measure is labelled La Porta 2002 and taken
from Pagano and Volpin (2005). Final, we use a law-origin indicator to differentiate
between countries that have civil law origin and those that have common law origin.
This indicator is denoted Common Law and taken from La Porta et al. (1998, 2000).
With the first three governance indices, a higher score means better macro corporate
governance environment. With the last index, the common law countries are
considered to have better corporate governance environment than the civil law ones
(La Porta et al. (2000)).
In the first scenario, we assign countries to two corporate governance groups based on
the median Minority Rights index. We then run country-effect regressions of stock
return volatility on large foreign ownership for firms in the high Minority Rights
group and for those in the low Minority Rights group separately. We repeat the same
procedure with Financial Disclosure index, La Porta 2002 index, and Common Law
indicator take turn to be the basis for division of countries into high versus low
governance groups. The country-effect regression results are reported in Panel A and
Panel B of Table 10.
For each scenario in Panel A, the regression results consistently show that in better
corporate governance environment, large foreign ownership is negatively related to
stock return volatility. The t-statistics for the coefficient on large foreign ownership in
high Minority Rights, high Financial Disclosure, high La Porta 2002, and Common
Law group are -2.26, -2.35, -1.97, and -3.10, respectively. Nevertheless, in weaker
corporate governance environment, large foreign ownership is not related to stock
return volatility. The t-statistics for the coefficient on large foreign ownership in low
Minority Rights, low Financial Disclosure, low La Parta 2002, and Civil Law group
are -0.12, 0.47, 0.28, and 0.25, respectively.
The first three scenarios in Panel B show a similar picture to the four scenarios
reported in Panel A quantitatively. Large foreign ownership is negatively related to
stock return volatility in better corporate governance environment, but there is no
relation between these two variables in weaker corporate governance environment.
The fourth scenario exhibit an exception to the conclusion above. Not only large
foreign ownership is negatively related to stock return volatility in Common Law
countries, but also it is negatively related to stock return volatility in Civil Law
countries. The t-statistic for the coefficient on large foreign ownership in the Civil
Law group is -1.87, indicating the coefficient is statistically significant lower than 0 at
the 10% level.
As argued above, when the regression results are different due to alternative measures
of stock return volatility, we will base our conclusion on the results where stock
return volatility is estimated by logarithm of squared returns. In this section, we
therefore use the results from Panel B to reach a conclusion on the relation between
stock return volatility and large foreign ownership under different macro corporate
governance scenarios.
In summary, the role of foreign investors disappears in weak corporate governance
environments, where the quality of corporate governance is measured based on
Minority Rights, Financial Disclosure, and La Porta 2002 indices. Nevertheless, we
find that the influence of foreign investors on firms’ risks exists in both common law
and civil law countries.
C. Relation to Merton’s (1987) Model
The negative relation between large foreign ownership and stock return volatility is
also consistent with Merton’s (1987) investor-base broadening hypothesis, which
implies that a larger foreign investor base would lead to lower stock return volatility.
Our findings are consistent with the prediction of Merton’s model if we measure the
increase of investor base by the actual large foreign ownership.
V. Conclusion
Past financial crises have raised concerns on the impact of capital market
liberalization on the market volatility. Many studies have investigated this issue but
do not reach a conclusion on what effect capital market liberalization might have on
emerging market volatility. This paper studies the issue by looking at the role played
by large foreign investors on stock return volatility.
We find a negative relation between large foreign ownership and stock return
volatility. The result is robust to alternative definitions of stock return volatility as
well as alternative definition of large foreign ownership.
We have three stories to understand our major finding. Firstly, in many cases, a large
foreign ownership is a foreign direct investment (FDI). Stiglitz (2000) argues that
coming with FDI are resources, technology, and valuable training of human capital.
All of these would increase the firm’ operating efficiency and reduce its specific risk,
resulting in lower stock return volatility. Secondly, foreign investors of large
investment demand higher transparency, improved disclosure rules, accountability of
management, and better shareholder rights (Kim and Singal (2000)). In addition,
financial liberalization benefits the liberalizing countries by reducing the cost of the
twin agency problems proposed by Stulz (2005). It provides the governance
mechanisms and incentives for firms to improve their corporate governance. Finally,
Merton’s (1987) model implies that a larger foreign investor base would lead to lower
stock return volatility. Roughly speaking, our finding is consistent with the prediction
of Merton’s model if we measure the increase of investor base by the actual large
foreign ownership.
The use of large foreign ownership in our paper helps address two problems that
many of previous research face. First, they focus on regulatory barriers and ignore the
nature (stability versus non-stability) of investment. Second, previous studies ignore
the fact that opening markets is not sufficient for foreign investors to make investment
in domestic stock markets. Our study therefore provides a new dimension of studying
capital market liberalization.
The study also shows that large foreign ownership of a domestic firm is different from
the degree of openness, or investability of that firm. Firms with high foreign
ownership may have very low investable weights, while firms with high investable
weights may have low foreign investment. This is due to two possible reasons. Firstly,
there are problems with the investability measures, such as it does not reflect the true
investability of a stock when the stock is small and illiquid. Secondly, foreign
investors may not invest up to the legal limit in companies that are not their targets.
Furthermore, our paper reveals the varying role of foreign investors in different
corporate governance environments. We find that the role of foreign investors is
strong in better corporate governance environments, whether at firm-level or country-
level corporate governance, but it disappears in weaker corporate governance
environments. Except for the case of civil law versus common law countries, the role
of foreign investors exists in both groups of countries.
The results in this paper have important implications for policy makers and
international financial theorists alike. For policy makers, designing policies that could
attract foreign investors, such as better regulations, more investor protection, more
transparency, etc. is more important than just opening up the markets. For
international financial theorists, foreign investment factor needs to be taken into
account in their model of international investment and risk.
Reference
Aggarwal, R., C. Inclan, and R. Leal, 1999. Volatility in emerging stock markets. Journal of Financial and Quantitative Analysis 34, 33-55.
Alexander, Gordon, Cheol Eun, and Sundaram Janakiramanan, 1987, Asset pricing and dual listing on foreign capital markets: A note. Journal of Finance 42, 151-158.
Bae, Kee-Hong, Kalok Chan, and Angela Ng, 2004. Investibility and return volatility. Journal of Financial Economics 71, 239-263.
Bekaer, Geert, and Campbell R. Harvey, 1997. Emerging equity market volatility. Journal of Financial Economics 43, 29-77.
Bekaert, Geert, and Campbell R. Harvey, 2000. Foreign speculators and emerging equity markets. Journal of Finance 55, 565-614.
Bekaert, Geert, and Campbell R. Harvey, 2002. Research in emerging markets finance: looking to the future. Emerging Markets Review 3, 429-448.
Bekaert, Geert, and Campbell R. Harvey, 2003. Emerging markets finance. Journal of Empirical Finance 10, 3-55.
Bekaert, Geert, and Campbell R. Harvey, 2005. Does financial liberalization spur growth?. Journal of Financial Economics 77, 3-55.
Bekaert, Geert, Campbell R. Harvey, and Robin L. Lumsdaine, 2002. Dating the integration of world equity markets. Journal of Financial Economics 65, 203-247.
Boyd, John, and Bruce Smith, 1996. The coevolution of the real and financial sectors in the growth process. World Bank Economic Review 10, 371-396.
Buckberg, Elaine, 1995. Emerging stock markets and international asset pricing. World Bank Economic Review 9, 51-74.
Chari, Anusha, and Peter Blair Henry, 2004. Risk sharing and asset prices: Evidence from a natural experiment. Journal of Finance 59, 1295-1324.
Claessens, S., S. Djankov, and L. H. P. Lang, 2000. The separation of ownership and control in East Asian corporations. Journal of Financial Economics 58, 81-112.
Clark, John, and Elizabeth Berko, 1997. Foreign investment fluctuations and emerging market stock returns: The case of Mexico, Staff Reports, Researve Bank of New York Staff Reports, No. 24
Dahlquist, Magnus, and Goran Robertsson, 2001. Direct foreign ownership, institutional investors, and firm characteristics. Journal of Financial Economics 59, 413-440.
Edison, H., and F. Warnock, 2003. A simple measure of the intensity of capital controls. Journal of Empirical Finance 10, 81-104.
Errunza, Vihang, and Etienne Losq, 1985. International asset pricing under mild segmentation: Theory and test. Journal of Finance 40, 105-124.
Eun, Cheol, and Sundaram Janakiramanan, 1986. A model of international asset pricing with a constraint on foreign equity ownership. Journal of Finance 41, 897-914.
Faccio, M., and L. H. P. Lang, 2002. The ultimate ownership of Western European corporations, Journal of Financial Economics 65, 365-395.
Henry, Peter Blair, 2000. Stock market liberalization, economic reform, and emerging market equity prices. Journal of Finance 55, 529-564.
Henry, Peter Blair, 2003. Capital account liberalization, the cost of capital, and economic growth. American Economic Review, 93(2): 91-96.
Johnston, Jack, & John DiNardo, 1997. Econometric Methods, 4th edn, The MacGraw-Hill Companies, New York.
Kelley, Eric, and Tracie Woidtke, 2006. Investor protection and real investment by U.S. multinationals. Journal of Financial and Quantitative Analysis, forthcoming.
Kim, E. Han, and Vijay Singal, 2000. Stock market openings: Experience of emerging economies. Journal of Business 73, 25-66.
La Porta, Rafael, Florencio Lopez-de-Silanes, and Andrei Shleifer, 1999. Corporate ownership around the world. Journal of Finance 54, 471-517.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny, 2002. Investor protection and corporate valuation. Journal of Finance 57, 1147-1170.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny, 2000. Investor protection and corporate governance. Journal of Financial Economics 58, 3-28.
Levine, Ross, and Sara Zervos, 1996. Stock market development and long-run growth. World Bank Economic Review 10, 323-339.
Li, Donghui, Fariborz Moshirian, Peter Kien Pham, and Jason Zein, 2006. When financial institutions are large shareholders – The role of macro corporate governance environments. Journal of Finance, forthcoming.
Merton, Robert C., 1987. A simple model of capital market equilibrium with incomplete information. Journal of Finance 42, 483-510.
Milton, Todd, 2006. Stock market liberalization and operating performance at the firm level. Journal of Financial Economics, forthcoming.
Organisation for Economic Cooperation and Development (1999), OECD Benchmark definition of foreign direct investment, 3rd edn, http://www.oecd.org/dataoecd/10/16/2090148.pdf
Pagano, Marco, and Paolo Volpin, 2005. Political economy of corporate governance. American Economic Review, forthcoming.
Rajan, Raghuram G., and Luigi Zingales, 1998. Financial dependence and growth. American Economic Review 88, 559-86.
Rossi, Stefano, and Paolo F. Volpin, 2004. Cross-country determinants of mergers and acquisitions. Journal of Financial Economics 74, 277-304.
Rouwenhorst, G., 1999. Local return factors and turnover in emerging stock markets. Journal of Finance 54, 1439-1464.
Shleifer, Andrei, and Daniel Wolfenzon, 2002. Investor protection and equity markets. Journal of Financial Economics 66, 3-27.
Stapleton, Richard, and Marti Subrahmanyam, 1977. Market imperfections, capital market equilibrium, and corporate finance. Journal of Finance 32, 307-319.
Stiglitz, Joseph E., 1999. Reforming the Global Economic Architecture: Lessons from Recent Crises. Journal of Finance 54, 1508-1521.
Stiglitz, Joseph E., 2000. Capital market liberalization, economic growth, and instability. World Development 28, 1075-1086.
Stiglitz, Joseph E., 2004. Capital-market liberalization, globalization, and the IMF. Oxford Review of Economic Policy 20, 57-71.
Stulz, Rene M., 1999a. International portfolio flows and security markets, Working paper, Dice Center for Financial Economics, The Ohio State University.
Stulz, Rene M., 1999b. Globalization and the cost of equity capital, Working paper, The New York Stock Exchange.
Stulz, Rene M., 2005. Presidential address: The limits of financial globalization. Journal of Finance 60, 1595-1639.
Wolfenzon, Daniel, 1998. A theory of pyramidal ownership. Mimeo, Havard University.
Table 1. Summary Statistics of the Foreign Ownership Sample
In 2002 observations from the Standard and Poor’s Emerging Markets Database are merged with those from the OSIRIS database. Standard deviation, investable weight, and foreign ownership are the means of firms’ standard deviation, investable weight, and foreign ownership across all firms in each country. Size and turnover are the medians of firm size and turnover across all firms in each country. A firm’s standard deviation, investable weight, foreign ownership, size, and turnover are computed as followed. A firm’s standard deviation is the standard deviation of monthly U.S dollar stock returns. A firm’s size, turnover and investable weight are the (time series) averages of monthly market capitalization, monthly turnover, which is the number of shares traded in a month divided by the number of shares outstanding at the beginning of the month, and investable weight, respectively. A firm’s foreign ownership is the annual percentage of block shareholdings by all foreign investors.
Country No. of Stocks
Standard Deviation
(%)
Size(mil. USD)
Turnover(%)
Foreign Ownership
(%)
InvestableWeight
Argentina 18 26.75 67.24 4.94 40.21 0.45Bahrain 12 6.95 120.46 0.33 18.93 0.00Brazil 50 19.92 258.64 2.44 19.44 0.54Chile 37 8.17 515.64 0.70 21.38 0.43China 189 9.72 358.46 6.40 1.77 0.18Colombia 11 9.52 253.61 0.43 7.33 0.00Czech Republic 15 12.43 132.98 0.10 30.06 0.17Egypt 49 7.36 32.17 0.99 6.79 0.14Hungary 17 9.77 122.43 3.22 35.03 0.50India 119 12.83 172.79 4.31 12.18 0.16Indonesia 54 17.50 54.29 1.62 16.65 0.19Israel 43 12.43 275.37 2.93 7.47 0.54Jordan 28 8.04 55.19 2.22 4.70 0.00Korea 98 15.20 352.44 25.82 5.89 0.73Malaysia 105 10.29 308.77 1.56 4.61 0.40Mexico 49 11.81 578.34 1.40 13.37 0.55Morocco 20 4.95 300.25 0.49 23.37 0.28Nigeria 26 10.12 102.05 0.62 28.98 0.00Oman 14 11.12 48.39 0.93 9.40 0.00Pakistan 34 15.58 49.72 9.65 10.31 0.00Peru 19 14.87 34.42 0.67 21.73 0.23Philippines 53 14.22 114.02 0.88 7.37 0.09Poland 25 12.57 224.20 2.19 36.83 0.46Russia 14 12.59 3920.96 1.77 2.24 0.39Slovakia 9 13.10 56.66 3.94 43.84 0.00South Africa 56 13.60 493.77 4.00 8.25 0.79Sri Lanka 41 11.87 12.62 1.83 19.47 0.00Taiwan 91 16.02 928.80 19.84 2.63 0.43Thailand 55 13.95 238.04 8.36 11.04 0.21Turkey 22 19.12 43.70 13.41 20.94 0.33Venezuela 11 16.94 88.01 0.36 11.03 0.00Zimbabwe 20 28.90 130.20 2.01 5.05 0.00
Average 44 13.38 326.39 4.07 15.88 0.26
Table 2. Country-effect Regressions of Volatility on Large Foreign Ownership
Data are from the Foreign Ownership database and in the panel data format: one dimension is across countries and the other across individual stocks within each country. In Regressions 1a, 2a, and 3a, volatility is the standard deviation of stock monthly returns over the 12 months in 2002. In Regression 1b, 2b, and 3b, volatility is the average of logarithm of squared monthly returns in 2002. Large foreign ownership is sum of block shareholdings in a firm. Three large foreign ownership dummies are created: zero-large foreign ownership dummy takes a value of 1 if large foreign ownership is equal to 0% and 0 otherwise; partial-large foreign ownership dummy takes a value of 1 if large foreign ownership is higher than 0% but less than and up to 50% and 0 otherwise; high-large foreign ownership dummy takes a value of 1 if foreign ownership is higher than 50% and 0 otherwise. Only two large foreign ownership dummies are used in order to avoid the collinearity problem. Size is the (time-series) average of monthly market capitalization for each firm. Turnover is the (time-series) average of monthly stock turnover, which is the number of shares traded in a month divided by the number of shares outstanding at the beginning of the month. Industries are classified based on the GICS methodology. Wald test tests the null hypothesis that the coefficients on the zero-large foreign ownership and partial-large foreign ownership dummies are the same.
Dependent variable Standard Deviation of Monthly Stock Returns Logarithm of squared returns(1a) (2a) (3a) (1b) (2b) (3b)
Independent variablesConstant 0.1304 0.1761 0.1645 3.6085 4.4570 4.1494
t-stat 80.76 27.64 19.92 121.73 38.06 27.52Large foreign ownership -0.0003 -0.0001 -0.0061 -0.0036
t-stat -3.65 -1.69 -4.63 -2.82Large foreign ownership dummies Foreign ownership = 0% 0.0114 0.3041
t-stat 2.23 3.27 0% < Foreign ownership <= 50% 0.0093 0.2273
t-stat 1.69 2.27Size -0.0082 -0.0081 -0.1488 -0.1472
t-stat -7.78 -7.69 -7.65 -7.55Turnover 0.0234 0.0233 0.4913 0.4876
t-stat 4.31 4.29 4.81 4.78Industry dummies Yes Yes Yes YesWald test 0.3700 0.4500
p-value 0.54 0.23R-squared 0.0097 0.0937 0.0951 0.0146 0.1074 0.1093
Table 3. Instrumental variable regressions of volatility on large foreign ownership
Data are from the Foreign Ownership database and in the panel data format: one dimension is across countries and the other across individual stocks within each country. In Regressions 1a, 2a, and 3a, volatility is the standard deviation of monthly stock returns over the 12 months in 2002. In Regression 1b, 2b, and 3b, volatility is the average of logarithm of squared monthly returns in 2002. Large foreign ownership is the sum of block shareholdings in a firm. Three instruments for the foreign ownership are Largest domestic shareholder’s ownership, Controlling shareholder type, and Pyramid dummy. Largest domestic shareholder’s ownership is the ownership of the largest shareholders in firms (excluding foreign shareholders). Controlling shareholder type is a dummy variable. It takes a value of 1 if the controlling shareholder is family or government, and takes a value of 0 otherwise. The controlling shareholder is defined as the largest domestic shareholder among those with more than 20% ownership of a firm. Pyramid dummy takes the value of 1 if the controlling shareholder owns shares through a pyramid, and 0 otherwise.
Columns (1a) and (1b): Instruments are Largest domestic shareholder’s ownership, Size, Turnover, and Industry dummiesColumns (2a) and (2b): Instruments are Controlling shareholder type, Size, Turnover, and Industry dummiesColumns (3a) and (3b): Instruments are Controlling shareholder type, Pyramid dummy, Size, Turnover, and Industry dummies
Dependent variable Standard Deviation of Monthly Returns Logarithm of squared monthly returns
(1a) (2a) (3a) (1b) (2b) (3b)
Independent variables
Constant 0.1713 0.1714 0.1715 4.4566 4.4603 4.4602
t-stat 26.84 26.59 26.66 38.05 37.99 38.03
Large foreign ownership -0.0003 -0.0005 -0.0005 -0.0054 -0.0070 -0.0062
t-stat -2.29 -2.90 -2.69 -2.36 -2.27 -2.04
Size -0.0079 -0.0075 -0.0076 -0.1452 -0.1425 -0.1440
t-stat -7.38 -6.76 -6.87 -7.33 -7.04 -7.13
Turnover 0.0229 0.0225 0.0226 0.4857 0.4825 0.4838
t-stat 4.20 4.08 4.11 4.75 4.71 4.72
Industry dummies Yes Yes Yes Yes Yes Yes
R-squared 0.0909 0.0748 0.0792 0.1061 0.1033 0.1052
35
Table 4. Country-effect Regressions of Volatility on Large Foreign Ownership – 2003 volatility
Data are from the Foreign Ownership database and in the panel data format: one dimension is across countries and the other across individual stocks within each country. In Regressions 1a, 2a, and 3a, volatility is the standard deviation of monthly stock returns over the 12 months in 2003. In Regression 1b, 2b, and 3b, volatility is the average of logarithm of squared monthly returns in 2003. Large foreign ownership is the sum of block shareholdings in a firm. Three large foreign ownership dummies are created (but only two are used in order to avoid the collinearity problem): zero-foreign ownership dummy takes a value of 1 if foreign ownership is equal to 0% and 0 otherwise; partial-foreign ownership dummy takes a value of 1 if foreign ownership is higher than 0% but less than and up to 50% and 0 otherwise; high foreign ownership dummy takes a value of 1 if foreign ownership is higher than 50% and 0 otherwise. Size is the (time-series) average of monthly market capitalization for each firm. Turnover is the (time-series) average of monthly stock turnover, which is the number of shares traded in a month divided by the number of shares outstanding at the beginning of the month. Industries are classified based on the GICS methodology. Wald test tests the null hypothesis that the coefficients on the zero foreign ownership and partial foreign ownership dummies are the same.
Dependent variable Standard Deviation of Monthly Stock Returns Logarithm of squared returns(1a) (2a) (3a) (1b) (2b) (3b)
Independent variablesConstant 0.1331 0.2156 0.2009 3.6115 4.4407 4.1481
t-stat 51.83 19.10 14.45 119.33 33.52 25.48Large foreign ownership -0.0004 -0.0002 -0.0065 -0.0043
t-stat -3.67 -2.07 -4.74 -3.23Large foreign ownership dummies Foreign ownership = 0% 0.0146 0.2903
t-stat 1.81 3.06 0% < Foreign ownership <= 50% 0.0100 0.2076
t-stat 1.16 2.05Size -0.0165 -0.0165 -0.1724 -0.1722
t-stat -9.28 -9.27 -8.28 -8.25Turnover 0.0456 0.0455 0.8171 0.8161
t-stat 3.97 3.96 6.04 6.03Industry dummies Yes Yes Yes YesWald test 0.7400 0.6900
p-value 0.39 0.19R-squared 0.0105 0.1177 0.1172 0.0173 0.1287 0.1282
36
Table 5. Stocks and their countries of origins in different combinations of investability and foreign ownership
High investability group is the group of stocks where the stocks’ investable weights > 0.5. Partial investability group is the group of stocks where the stocks’ investable weights are higher than 0 but less than or equal to 0.5. Non investability group is the group of stocks where the stocks’ investable weights equal to 0. High foreign ownership group is the group of stocks where the stocks’ large foreign ownership is higher than 50%. Partial foreign ownership group is the group of stocks where the stocks’ large foreign ownership is higher than 0% but less than or equal to 50%. Zero foreign ownership group is the group of stocks where the stocks’ large foreign ownership equal to 0.
Panel A Panel B Panel CHigh investability and high foreign ownership High investability but zero foreign ownership Zero investability but high foreign ownershipMarkets No. of
StocksPct Markets No. of
StocksPct Markets No. of
StocksPct
Argentina 1 6.25 Argentina 2 0.89 Argentina 4 7.27Brazil 5 31.25 Brazil 19 8.44 Chile 2 3.64Chile 1 6.25 Chile 8 3.56 Czech Republic 1 1.82Egypt 1 6.25 China 21 9.33 Egypt 1 1.82Hungary 1 6.25 Egypt 2 0.89 Hungary 3 5.45Indonesia 1 6.25 Hungary 4 1.78 India 4 7.27Mexico 4 25 India 1 0.44 Indonesia 4 7.27Poland 2 12.5 Indonesia 3 1.33 Morocco 2 3.64Total 16 100 Israel 14 6.22 Nigeria 10 18.18
Korea 51 22.67 Oman 1 1.82Malaysia 35 15.56 Pakistan 4 7.27Mexico 14 6.22 Peru 3 5.45Morocco 2 0.89 Philippines 1 1.82Peru 2 0.89 Slovakia 3 5.45Poland 3 1.33 Sri Lanka 9 16.36Russia 5 2.22 Thailand 1 1.82South Africa 37 16.44 Turkey 1 1.82Turkey 2 0.89 Venezuela 1 1.82Total 225 100 Total 55 100
37
Table 6. Country-effect Regressions of Volatility on Investability
Data are from the Foreign Ownership database and in the panel data format: one dimension is across countries and the other across individual stocks within each country. In Regressions 1a, 2a, and 3a, volatility is the standard deviation of monthly stock returns over the 12 months in 2002. In Regression 1b, 2b, and 3b, volatility is the average of logarithm of squared monthly returns in 2002. Investability is the legal limit on foreign investment in a domestic firm and is reported by EMDB. Three investability dummies are created (but only two are used in order to avoid the collinearity problem): non-investability dummy takes a value of 1 if investability is equal to 0 and 0 otherwise; partial-investability dummy takes a value of 1 if investability is higher than 0 but less than and up to 0.5 and 0 otherwise; high investability dummy takes a value of 1 if investability is higher than 0.5 and 0 otherwise. Size is the (time-series) average of monthly market capitalization for each firm. Turnover is the (time-series) average of monthly stock turnover, which is the number of shares traded in a month divided by the number of shares outstanding at the beginning of the month. Industries are classified based on the GICS methodology. Wald test tests the null hypothesis that the coefficients on the non investability and partial investability dummies are the same.
Dependent variable Standard Deviation of Monthly Stock Returns Logarithm of squared returns(1a) (2a) (3a) (1b) (2b) (3b)
Independent variablesConstant 0.1268 0.1753 0.1806 3.5188 4.4438 4.5469
t-stat 57.27 27.18 23.97 88.38 37.46 32.89Investability 0.0020 0.0041 0.0748 0.0705
t-stat 0.36 0.76 0.74 0.73Investability dummies Investability = 0 -0.0060 -0.1108
t-stat -1.28 -1.32 0 < Investability <= 0.5 -0.0048 -0.0642
t-stat -1.05 -0.77Size -0.0085 -0.0085 -0.1570 -0.1592
t-stat -8.09 -7.50 -8.06 -7.61Turnover 0.0234 0.0233 0.4921 0.4900
t-stat 4.29 4.29 4.80 4.78Industry dummies Yes Yes Yes YesWald test 0.0700 0.3100
p-value 0.79 0.58R-squared 0.0001 0.0922 0.0930 0.0004 0.1028 0.1036
38
Table 7. Country-effect Regressions of Volatility on Large foreign ownership and Investability
Data are from the Foreign Ownership database and in the panel data format: one dimension is across countries and the other across individual stocks within each country. In Regressions 1a, 2a, and 3a, volatility is the standard deviation of monthly stock returns over the 12 months in 2002. In Regression 1b, 2b, and 3b, volatility is the average of logarithm of squared monthly returns in 2002. Large foreign ownership is the sum of block shareholdings in a firm. Investability is the legal limit on foreign investment in a domestic firm and is reported by EMDB. Three large foreign ownership dummies are created (but only two are used in order to avoid the collinearity problem): zero-foreign ownership dummy takes a value of 1 if foreign ownership is equal to 0% and 0 otherwise; partial-foreign ownership dummy takes a value of 1 if foreign ownership is higher than 0% but less than and up to 50% and 0 otherwise; high foreign ownership dummy takes a value of 1 if foreign ownership is higher than 50% and 0 otherwise. Similarly, three investability dummies are created (but only two are used in order to avoid the collinearity problem): non-investability dummy takes a value of 1 if investability is equal to 0 and 0 otherwise; partial-investability dummy takes a value of 1 if investability is higher than 0 but less than and up to 0.5 and 0 otherwise; high investability dummy takes a value of 1 if investability is higher than 0.5 and 0 otherwise. Size is the (time-series) average of monthly market capitalization for each firm. Turnover is the (time-series) average of monthly stock turnover, which is the number of shares traded in a month divided by the number of shares outstanding at the beginning of the month. Industries are classified based on the GICS methodology. Wald test tests the null hypothesis that the coefficients on the two relevant dummies are the same.
Dependent variable Logarithm of squared returns(1a) (2a) (3a) (4a)
Independent variablesConstant 4.5468 4.4460 4.1413 4.2458
t-stat 32.96 37.57 27.37 25.07Large foreign ownership -0.0036 -0.0036
t-stat -2.78 -2.80Investability 0.0625 0.0652
t-stat 0.65 0.67Large foreign ownership dummies Foreign ownership = 0% 0.3016 0.3005
t-stat 3.24 3.21 0% < Foreign ownership <= 50% 0.2192 0.2143
t-stat 2.17 2.11Investability dummies Investability = 0 -0.1036 -0.1046
t-stat -1.23 -1.23 0 < Investability <= 0.5 -0.0408 -0.0376
t-stat -0.49 -0.44Size -0.1541 -0.1502 -0.1485 -0.1527
t-stat -7.36 -7.67 -7.58 -7.29Turnover 0.4840 0.4869 0.4827 0.4795
t-stat 4.73 4.76 4.72 4.69Industry dummiesWald test of the coefficients on the foreign ownership dummies being equal
1.7900
p-stat 0.18Wald test of the coefficients on the investability dummies being equal
0.5600 1.6300 0.6400
p-stat 0.45 0.20 0.42R-squared 0.1084 0.1077 0.1095 0.1102
Table 8. Country-effect Regressions of Volatility on Large Foreign Ownership: Alternative definition of large foreign ownership
Data are from the Foreign Ownership database and in the panel data format: one dimension is across countries and the other across individual stocks within each country. In Regressions 1a, 2a, and 3a, volatility is the standard deviation of stock monthly returns over the 12 months in 2002. In Regression 1b, 2b, and 3b, volatility is the average of logarithm of squared monthly returns in 2002. Large foreign ownership is sum of block shareholdings in a firm, with a block now defined as an ownership of 10% or more. Three large foreign ownership dummies are created: zero-large foreign ownership dummy takes a value of 1 if large foreign ownership is equal to 0% and 0 otherwise; partial-large foreign ownership dummy takes a value of 1 if large foreign ownership is higher than 0% but less than and up to 50% and 0 otherwise; high-large foreign ownership dummy takes a value of 1 if foreign ownership is higher than 50% and 0 otherwise. Only two large foreign ownership dummies are used in order to avoid the collinearity problem. Size is the (time-series) average of monthly market capitalization for each firm. Turnover is the (time-series) average of monthly stock turnover, which is the number of shares traded in a month divided by the number of shares outstanding at the beginning of the month. Industries are classified based on the GICS methodology. Wald test tests the null hypothesis that the coefficients on the zero-large foreign ownership and partial-large foreign ownership dummies are the same.
Dependent variable Standard Deviation of Monthly Returns Logarithm of squared monthly returns(1a) (2a) (3a) (1b) (2b) (3b)
Independent variablesConstant 0.1301 0.1761 0.1604 3.6031 4.4566 4.1667
t-stat 82.36 27.65 19.39 124.08 38.06 27.59Large foreign ownership -0.0003 -0.0001 -0.0062 -0.0038
t-stat -3.70 -1.81 -4.67 -2.90Large foreign ownership dummies Foreign ownership = 0% 0.0107 0.2884
t-stat 2.08 3.08 0% < Foreign ownership <= 50% 0.0105 0.2138
t-stat 1.77 1.97Size -0.0082 -0.0082 -0.1492 -0.1489
t-stat -7.80 -7.80 -7.68 -7.67Turnover 0.0234 0.0234 0.4908 0.4888
t-stat 4.31 4.31 4.81 4.79Industry dummies Yes Yes Yes YesWald test 0.0000 0.9800
p-value 0.96 0.32R-squared 0.0100 0.0940 0.0948 0.0148 0.1077 0.1085
Table 9. The varying role of foreign shareholders under different micro corporate governance environments
Data are from the Foreign Ownership database and in the panel data format: one dimension is across countries and the other across individual stocks within each country. The table reports the country-effect regressions of volatility on large foreign ownership and controlled variables. We analyse three different scenarios. First, firms are classified into two groups based on the presence of controlling shareholders. Second, firms are classified into two groups based on the pyramidal structure of the controlling shareholder’s ownership. Third, firms are classified into two groups based on the type of controlling shareholders. In particular:Column (1a): The controlling shareholder is not presentColumn (1b): The controlling shareholder is presentColumn (2a): There is no pyramid ownership structureColumn (2b): There is a pyramid ownership structureColumn (3a): The controlling shareholder is not present or is a widely held corporationColumn (3b): The controlling shareholder is family/individual or governmentControlling shareholder is the largest domestic shareholder with an ownership of more than 20% of the firm’s equity. Pyramidal structure of the controlling shareholder’s ownership means indirect ownership obtained through one or many third parties. Large foreign ownership is sum of block shareholdings in a firm. Size is the (time-series) average of monthly market capitalization for each firm. Turnover is the (time-series) average of monthly stock turnover, which is the number of shares traded in a month divided by the number of shares outstanding at the beginning of the month. Industries are classified based on the GICS methodology.Panel A.
Dependent variable Standard deviation of monthly returns
(1a) (1b) (2a) (2b) (3a) (3b)
Independent variables
Constant 0.1826 0.1571 0.1732 0.1597 0.1677 0.1721
t-stat 18.16 18.87 25.27 8.33 17.99 19.08
Large foreign ownership -0.0002 0.0000 -0.0001 -0.0002 0.0000 -0.0001
t-stat -1.53 0.13 -1.80 -0.35 -0.41 -0.50
Size -0.0090 -0.0072 -0.0084 -0.0085 -0.0074 -0.0092
t-stat -5.41 -5.22 -7.31 -2.78 -4.82 -6.12
Turnover 0.0190 0.0732 0.0216 0.1820 0.0204 0.0357
t-stat 3.04 4.97 3.91 3.16 3.18 3.11
Industry dummies Yes Yes Yes Yes Yes Yes
R-squared 0.1310 0.0956 0.0986 0.1481 0.0957 0.0961
41
Panel B.Dependent variable Logarithm of squared monthly returns
(1a) (1b) (2a) (2b) (3a) (3b)
Independent variables
Constant 4.6644 4.2797 4.4154 4.2571 4.6246 4.2754
t-stat 25.01 27.96 37.62 12.93 26.99 26.10
Large foreign ownership -0.0049 0.0007 -0.0043 -0.0023 -0.0037 -0.0031
t-stat -2.67 0.14 -3.39 -0.22 -2.22 -0.57
Size -0.1540 -0.1474 -0.1341 -0.1588 -0.1547 -0.1346
t-stat -4.96 -5.79 -6.84 -3.03 -5.39 -4.95
Turnover 0.3677 1.5582 0.4385 3.5706 0.4173 0.7375
t-stat 3.09 5.72 4.63 3.61 3.43 3.50
Industry dummies Yes Yes Yes Yes Yes Yes
R-squared 0.1412 0.1189 0.1283 0.1753 0.1345 0.0868
42
Table 10. The varying role of foreign shareholders under different macro corporate governance environments
Data are from the Foreign Ownership database and in the panel data format: one dimension is across countries and the other across individual stocks within each country. Panel A reports the country-effect regressions where the dependent variable is measured by the standard deviation of monthly returns. Panel B reports the country-effect regressions where the dependent variable is measured by the logarithm of squared monthly returns. Minority rights, financial disclosure, and LaPorta 2002 are the corporate governance scores. The sample is divided into high versus low corporate governance score groups based on the median value for each score. In the common law column, yes refers to a group of countries which have a common law system and no refers to the other countries. Large foreign ownership is sum of block shareholdings in a firm. Size is the (time-series) average of monthly market capitalization for each firm. Turnover is the (time-series) average of monthly stock turnover, which is the number of shares traded in a month divided by the number of shares outstanding at the beginning of the month. Industries are classified based on the GICS methodology.
Panel A.Dependent variable Standard deviation of monthly returns
Minority Rights Financial disclosure LaPorta 2002 Common Law
High Low High Low High Low Yes NoIndependent variables
Constant 0.1673 0.1751 0.1792 0.1523 0.1769 0.1158 0.1833 0.1643t-stat 19.89 17.67 20.84 15.98 25.98 6.12 17.51 20.39
Large foreign ownership -0.0002 0.0000 -0.0002 0.0000 -0.0002 0.0000 -0.0004 0.0000t-stat -2.26 -0.12 -2.35 0.47 -1.97 0.28 -3.10 0.25
Size -0.0088 -0.0071 -0.0085 -0.0077 -0.0090 -0.0006 -0.0088 -0.0080t-stat -6.51 -4.12 -6.22 -4.63 -7.99 -0.19 -4.90 -6.09
Turnover 0.0362 0.0131 0.0224 0.1151 0.0222 0.0870 0.0248 0.0223t-stat 4.53 1.76 3.92 3.48 4.02 2.70 2.87 3.16
Industry dummies Yes Yes Yes Yes Yes Yes Yes YesR-squared 0.1127 0.0922 0.1087 0.1046 0.1074 0.1057 0.1512 0.0795
43
Panel B.Dependent variable Logarithm of squared monthly returns
Minority Rights Financial disclosure LaPorta2002 Common Law
High Low High Low High Low Yes NoIndependent variables
Constant 4.2920 4.6876 4.5374 4.2015 4.5008 4.0001 4.5260 4.4356t-stat 26.72 27.30 28.22 24.94 35.35 13.14 21.98 31.23
Large foreign ownership -0.0047 -0.0015 -0.0047 -0.0014 -0.0039 -0.0011 -0.0054 -0.0029t-stat -2.86 -0.69 -2.61 -0.77 -2.71 -0.41 -2.33 -1.87
Size -0.1549 -0.1436 -0.1449 -0.1527 -0.1551 -0.0864 -0.1752 -0.1360t-stat -5.93 -4.85 -5.63 -5.22 -7.36 -1.66 -4.93 -5.87
Turnover 0.7168 0.2985 0.4531 2.2973 0.4626 1.2452 0.6616 0.3612t-stat 4.57 2.29 4.12 3.92 4.37 2.37 3.81 2.85
Industry dummies Yes Yes Yes Yes Yes Yes Yes YesR-squared 0.1216 0.0968 0.1102 0.1376 0.1112 0.1294 0.1334 0.0951
44