• No results found

Influence of transaction costs on international equity portfolio allocations of EU 15 countries

N/A
N/A
Protected

Academic year: 2021

Share "Influence of transaction costs on international equity portfolio allocations of EU 15 countries"

Copied!
49
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Influence of transaction costs on international equity portfolio

allocations of EU 15 countries

Master thesis

MSc International Financial Management

University of Groningen

Faculty of Economics and Business

University of Uppsala

Faculty of Economics

Groningen, 2ndMay 2011

Author: Diana Frolova

Student number: s1942840

(2)

Abstract

Globalization of financial markets offers possibilities for international diversification. However, many investors allocate more of their wealth to domestic securities. Investors invest more in domestic securities due to the barriers to cross-border investment, which arise from capital controls, exchange rate volatility, information asymmetry, different legal systems, discriminatory taxes, etc. Transaction costs are one of the barriers to cross-border investment. Increased transaction costs affect the level of portfolio diversification, decrease expected returns from investment, increase market volatility, etc. I use bilateral cross-country equity portfolio holdings data of EU 15 countries among 23 developed countries and 14 emerging market economies between 2002 and 2009 to analyze whether the level of transaction costs can influence foreign equity portfolio allocations. I find that the magnitude of transaction cost has a negative significant effect on the foreign equity portfolio allocations. Therefore, countries with higher transaction costs attract less foreign equity portfolio investments. Moreover, the results indicate that EU 15 countries concentrate their equity holdings in European markets due to high integration of financial markets in Europe.

Key words:

Barriers to cross-border investment, home bias, international equity allocation, panel data, transaction costs.

JEL classification:

(3)

Table of Contents

I. Introduction ...4

II. Literature review ...7

Transaction costs: a barrier to cross-border investments...7

Factors that influence cross-border investments ...8

III. Data and methodology...12

Variables...12

Dependant variable...12

Independent variables...12

Methodology ...16

Data ...18

Descriptive statistics: key stylized facts ...20

IV. Results ...32

Robustness tests...34

V. Conclusion...40

(4)

I.

Introduction

The main goal of this paper is to investigate the impact of transaction costs on foreign equity portfolio allocations of 15 European countries across 23 developed countries and 14 emerging market economies over the period of 2002-2009. Moreover, I control for the impact of other factors on the portfolio allocations such as home bias, economic and stock market development, potential diversification, exchange rate volatility, and information asymmetry. I also conduct robustness checks to test the sensitivity of the results and test whether there is an impact of EU or the EMU on portfolio allocations of EU 15 countries.

Globalization is a driving force for economic and financial integration (Subramanian et al., 2009). Due to globalization processes, over the last years many trade barriers and tariffs have decreased, companies are increasingly producing and selling their products abroad, and many capital markets across the world have become more accessible and efficient for the investors. Financial markets, which became more accessible across the globe, provide many opportunities for the investors to diversify their investments across many markets (Chan et al., 2005).

However, investors do not exploit the opportunities of the international diversification, because they allocate a large part of their investments to the domestic markets. This phenomenon, when investors overweight the domestic markets in their portfolios, is home bias (Chan et al., 2005). Prior research finds that the investors exhibit home bias in foreign portfolio allocations due to information asymmetry between domestic and foreign investors (Chan et al., 2005). Moreover, various government policies and domestic institutions, such as capital controls, information disclosure, corporate governance, etc., influence home bias in foreign portfolio allocations (Gelos and Wei, 2005). The evidence of the home bias in foreign portfolio allocations proves that there are barriers to cross-border investments. These barriers to cross-border investments arise from discriminatory taxes, different legal status given to the foreign investors in terms of ownership restrictions, differences in accounting and information disclosure standards and investor protection regulations, capital controls and transaction costs (Thapa and Poshakwale, 2010).

(5)

Prior research shows that the magnitude of transaction cost has crucial role in international investment. It also implies that despite of the opportunities of international diversification, investors allocate more of their portfolio investments to those countries where the transaction costs are low. With the reference to prior empirical studies, I investigate whether the level of transaction cost in a country affect the foreign equity portfolio allocations to it. I also control for the effects of other factors, which previous research found important, on foreign portfolio allocations. To conduct my analysis, I use bilateral cross-border equity holdings of EU 15 countries among 23 developed and 14 developing countries over the period of 2002-2009. This paper examines whether the level of transaction cost influence the foreign equity portfolio allocations of 15 European countries.

This paper focuses on 15 Developed European countries1 being a source of investment (Austria, Belgium, Denmark, Finland, France, Germany, Greece, Italy, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and UK). European equity markets have become highly integrated since 1996 (Hasselman and Herwartz, 2008). The integration of the European markets has two stages. The first stage in 1990-1993 has dismantled all restrictions to the free movement of capital within the European Union (EU). The second stage in 1994-1998 has established the European Monetary Institute to strengthen the cooperation between the central banks of EU member states. Prior research shows that the integration of European equity markets has impact on foreign portfolio allocations. Hasselman and Herwartz (2008) find that European financial markets have changed towards more integration and intra-EMU portfolio holdings. Moreover, Berglund and Al-Khail (2002), who study equity portfolio holdings of Finish investors, note that Economic and Monetary Union (EMU) is making portfolio investments into the area more attractive because it is reducing the foreign exchange risk of investments into these countries. Similarly, De Santis and Gerard (2009) find that reduced exchange rate risk and barriers to foreign portfolio investments increase portfolio holdings in Euro zone.

I find that transaction costs have significant effect on foreign equity portfolio allocations, and that countries with higher level of transaction cost attract lower foreign equity portfolio investment. I also find that investors prefer economically developed markets, which have similar cultural background, with low exchange rate volatility and information asymmetry, and which provide strong investor protection rights. Moreover, I find that EU 15 countries have nearly 50 percent of their foreign equity portfolio holdings in European markets out of which nearly 41 percent is in Euro zone. This finding confirms that European equity markets are highly integrated because of creation of EU and the EMU.

This paper follows recent study of Thapa and Poshakwale (2010), which studies transaction costs influence on decisions of equity portfolio allocation. However, this paper adds to the current literature in two ways. First, most studies, analyzing the international portfolio allocations, focus on large and mostly single country such as US, Finland being a source of investments (Borkovec et al., 2009; Didier et al., 2010). Instead, I

(6)

focus on the European market, which includes 15 countries. The chosen sample countries are the EU member states plus Norway and Switzerland. Moreover, ten countries out of 15 are members of the EMU. Secondly, the prior literature has not investigated in depth the geography of international investments due to the lack of a consistent database on international portfolio allocations. In this paper, I use CPIS database from IMF, which reports international portfolio holdings with a geographical breakdown. In addition, I use the data from CPIS over the more recent period (2002-2009) then the other studies2, which used CPIS database.

The reminder of the paper is organized as follows. Section 2 presents the theoretical framework of the analysis and the overview of prior empirical literature. Section 3 presents the data and methodology. Section 4 investigates the impact of transaction costs on foreign equity portfolio allocations and tests robustness of the results. Section 5 concludes.

2

(7)

II.

Literature review

Investors face various barriers when they invest in foreign markets. These barriers arise from the possibility of expropriation of foreign holdings, direct controls on the import and export of capital, reserve requirement on bank deposits and other assets held by foreigners, and restrictions on the part of a business that can be foreign owned. The transaction costs are also one of the barriers to cross-border investments. These barriers to cross-border investments withhold investors from investing in foreign markets. Therefore, if investors meet greater barriers when entering a foreign market, they will hold fewer equities in that market.

Transaction costs: a barrier to cross-border investments

International Capital Asset Pricing Model (CAPM) implies that international investors should hold assets of each country in proportion to the country’s share in the world’s market portfolio (Gelos and Wei, 2005). Nevertheless, according Cooper and Kaplanis (1986), in presence of transaction costs portfolio holdings of international investors start to deviate from the world’s market portfolio.

Prior literature assumes that each investor acts as an expected return maximiser for a given level of risk (Cooper and Kaplanis, 1986; Thapa and Poshakwale, 2010). Therefore, the investors, considering expected returns, transaction costs and the tolerated level of risk, seek to obtain the individually optimal portfolio.

Each investor holds the same two funds, when they have no barriers to foreign investment, the world’s market portfolio and the minimum risk portfolio (Cooper and Kaplanis, 1986). However, if investors have barriers to cross-border investment, they hold an additional fund, which is specific to the individual investor. This fund has the minimum level of risk for a specified level of transaction cost. Therefore, if there are no barriers to cross-border investment or the transaction costs are zero, then each investor holds the world’s market portfolio. When all investors hold the world’s market portfolio, where each country portfolio is weighted by its market capitalization, then the market equilibrium is achieved (De Santis and Gerard, 2009). But if the barriers to cross-border investment exist or the transaction costs of any country/investor pair are not equal to zero, then the portfolio holdings of each investor will differ from the world’s market portfolio (Chan et al., 2005).

(8)

(country j) (Cooper and Kaplanis, 1986; Chan et al., 2005; Thapa and Poshakwale, 2010). Therefore, the foreign equity portfolio allocation from country i to country j will depend on the difference between the transaction costs for investor i investing in domestic market and the transaction cost for investing in country j. Moreover, the higher transaction costs for country i investing in country j, the more is underweighted country j in investor i portfolio (Chan et al., 2005).

Factors that influence cross-border investments

According to theoretical model developed by Cooper and Kaplanis (1986), investors’ portfolio choice depends on barriers to cross-border investments. In presence of barriers to cross-border investments, investors allocate less wealth to foreign markets and invest more in domestic market. There are some empirical studies in the field of international investment and portfolio choice. Most of these studies focus on home bias, as a key issue in international portfolio choice, and try to explain why this phenomenon exist and what the determinants of home bias are. Generally, prior research focuses on the determinants of international investments such as economic growth and development, real exchange rate volatility, diversification potential, transaction costs, and information asymmetries. Table 1 presents an overview of most relevant empirical literature for my study.

Faruqee et al. (2004) studies the determinants of international equity portfolio holdings. They test the relationship between the bilateral equity holdings of 23 countries and transaction costs. They found that transaction costs have significant and negative relationship with international equity portfolio holdings. Other study, done by Chan et al. (2005), also includes transaction costs in their model. Analyzing how mutual funds from 23 countries allocate their equity portfolios across 48 countries, they found that transaction costs are associated with portfolio allocations. Chan et al. (2005) found that transaction costs bear significant negative sign, and, thus, investors prefer to invest more equity in countries where transaction costs are low. Furthermore, recent research conducted by Thapa and Poshakwale (2010) analyse the influence of different measures of transaction cost on international equity portfolio allocations. They found a robust and significant impact of transaction costs on international equity portfolio investments. Their findings show that countries with lower transaction costs attract greater foreign investments.

(9)

TABLE 1

Overview of the results of prior empirical literature

This table shows the main characteristics of the studies used as references in my analysis. The presented signs of variables are based on the results of the studies. A “+” means that a positive coefficient is found, a “-” means that a negative coefficient is found. “NA” means that variable is not included in a study done by those authors.

Authors Research question Sample

years Sample description

Model description D ep en d a n t v a ri a b le Independent variables T ra n sa ct io n co st s H o m e b ia s E co n o m ic g ro w th E co n o m ic D ev el o p m en t S to ck m ar k et d ev el o p m en t In v es to r p ro te ct io n E x ch an g e ra te v o la ti li ty D iv er si fi ca ti o n B il at er al d is ta n ce L in g u is ti c li ai so n Faruqee et al. (2004)

Study tests the determinants of international portfolio holdings. 1997 Cross-border equity holdings from 23 countries Holdings of country i in country j - NA NA NA NA NA NA - - +

Gelos and Wei (2005)

Do and how country transparency affect portfolio allocations?

1996-2000 (monthly)

Holdings of international and global funds with investments in emerging markets Weight of country i in fund j's portfolio NA NA NA + NA + NA NA NA NA Chan et al. (2005)

How mutual funds in different countries allocate their portfolios between domestic and foreign stocks?

1999-2000

Mutual fund equity holdings from 26 countries across 48 countries

Weight of domestic assets in fund holdings; Weight of foreign assets in fund holdings

- NA + + + + NA - - +

Fidora et al. (2007)

What is the impact of real exchange rate volatility on cross-border investment?

1997, 2001-2003

Global equity and bond holdings of 40 countries across 120 countries Home bias NA NA NA NA NA - + + + -Thapa and Poshakwale (2010)

Study analyses the influence of transaction costs on international equity allocations.

2001-2006

Bilateral equity holdings from 16 countries across 36 countries

Weight of stock holdings from country i in country j

(10)

Thapa and Poshakwale (2010) analyse investment holdings from 16 investor countries to 36 destination countries and find that home bias is statistically significant and bears negative sign. Their estimations show that 1 percent increase in home bias decreases portfolio holdings by nearly 1 percent.

When analysing international portfolio allocations, prior empirical studies include economic and stock market development in their models. A country’s level of advancement in economic and stock market development affects its ability to draw foreign investments to the country. Chan et al. (2005), studying the factors which determine allocations of financial, show that the percentage of mutual fund holdings in a particular country is related to economic and stock market development. Chan et al. (2005) find that economic and stock market development has significant and positive relation to international equity portfolio allocations. Therefore, investors tend to allocate more of their equity investment in a country, which is more economically developed and has large and liquid stock market. Gelos and Wei (2005) also find that economic development positively affects portfolio allocations of international and global equity funds with investment in the emerging markets. Additionally, Thapa and Poshakwale (2010) include economic and stock market development in their set of variables. Their results show that investors are concerned with the level of economic development in the countries and prefer to invest in countries where stock markets are large and less volatile.

La Porta et al. (1998) examine laws protecting rights of investors in 49 countries. Their results show that English common-law countries (Canada, Australia, India, US, and UK) generally have the strongest protection of investors’ rights and French civil-law countries (Belgium, France, the Netherlands, Poland, Italy, Portugal, Spain, etc.) have the weakest protection. German (Austria, Germany, Greece, Hungary, Switzerland, Japan, etc.) and Scandinavian civil-law countries (Norway, Sweden, and Finland) have moderate investor protection. Moreover, La Porta et al. (1998) find that investors are unlikely to be important in countries that poorly protect their rights. Other research also find that portfolio allocations have positive relation to investor protection, because investors prefer investing in those countries which provide better protection of investors’ rights (Chan et al., 2005; Gelos and Wei, 2005; Thapa and Poshakwale, 2010).

(11)

Further, Driessen and Laeven (2007) investigate potential benefits of international portfolio diversification for 52 countries. They find that there are statistically and economically significant regional and global diversification benefits for investors around the world. When the correlation between countries j and i is small, investors in country i enjoy larger diversification gain from investing in country j, and they have a greater desire to increase their equity holdings in country j. Previous studies find that the lower correlations should positively influence international portfolio investments (Chan et al., 2005; Thapa and Poshakwale, 2010). Moreover, Fidora et al. (2007) find that investors invest less in foreign financial assets in countries where their returns have strong positive correlation with domestic financial assets because then they cannot diversify the risk.

Finally, previous empirical research finds that investors are less familiar with foreign financial markets. Chan et al. (2005) find that with less familiarity, investors face greater information asymmetry that discourages them from investing abroad. Moreover, their results show that investors are more likely to trade stocks of firms that share the same language, cultural background, and closer geographical proximity. Thapa and Poshakwale (2010) find that common language has positive relation to foreign equity portfolio allocations because common language lowers the information asymmetry between foreign and domestic investors. On the other hand, they find that bilateral distance has negative relation to foreign equity portfolio allocations because the more distant the market is the more unfamiliar it is for foreign investors.

(12)

III.

Data and methodology

Variables

This subsection defines the dependant and independent variables used in this analysis.

Dependant variable

Following Thapa and Poshakwale (2010), the dependant variable is the allocation weight of foreign equity portfolio investment of country i in country j. For each of the EU 15 countries, I calculate the allocation weight of equity holdings in 36 countries at the end of each year as follows

(1)

where wijt is the equity investment in country j by investors from country i as a share of the total foreign

equity investments of investors from country i for the year t and FPIijtis the actual foreign equity portfolio

investment in millions US dollars from country i to country j at time t (36 recipient countries). Thapa and Poshakwale (2010) have denominated foreign equity investments in country i’s local currency, but results of the regressions do not change. Therefore, I use the foreign portfolio investments denominated in US dollars.

Independent variables

I will define the independent variables of this study. The first independent variable is transaction costs in country j at the time t. Transaction costs are a payment for making a transaction when buying or selling securities. Transaction costs in the stock market fall into two broad categories (Domowitz et al., 2001; Green et al., 2000). The first category is the direct costs. These costs include the fees, commissions, and any transaction taxes. The second category is the indirect costs. The indirect costs include the market impact costs, bid-ask spreads (are the price variances between buying and selling at any point time), costs of acquiring and processing information about stock values, companies, market movements and any other information which maybe relevant to the decision to buy or sell financial assets (Kasten, 2007).

I use two measures of transaction cost in this analysis. The first measure of transaction cost is commission paid. Commission is a service charge assessed by a broker or investment advisor in return for providing investment advice and/or handling the purchase or sale of a security (Borkovec et al., 2009). The second measure of transaction cost is implementation shortfall (IS) costs. Implementation shortfall costs is the sum

(13)

of timing delay costs and the market impact costs (ITG, 2009). Timing delay costs are any delay cost incurred between initial decision (open on Day 1) and the broker placement price (ITG, 2010). Market impact costs are price change between the time when the broker place the order and the eventual trade price (ITG, 2010).

Despite the growing international investments, the international portfolios keep on holding a big share of domestic assets (Kirabaeva and Razin, 2010). In addition, Chan et al. (2005) suggest that investors allocate a relatively large fraction of their wealth to domestic equities and underweight the foreign markets. This phenomenon is widely known as home bias. Home bias occurs when investments are to larger extent concentrated in the home country than optimal portfolio diversification implies (Berglund and Al-Khail, 2002). Home bias between two countries measures how much the actual weight of financial assets of country j in portfolio of any given country i differs from the benchmark weight this country should receive (Fidora et al., 2007). I follow Thapa and Poshakwale (2010) and include home bias as one of the variables. I calculate a measure of home bias as follows3

(2)

where HBijtis a home bias observed by the investor country i for a country j at time t; wijtis the share of

country j’s assets in country i’s portfolio. The benchmark weight is a world’s market capitalization weight, which is the share of the market capitalization of the country j’s equity market relative to the market capitalization of all equity markets. The benchmark weight is calculated:

(3)

where BWTjtis the benchmark weight of country j at time t, and MCjtis the market capitalization of the issuer

j country at time t measured in millions US dollars.

Investors tend to invest more in well-developed stock markets such as UK or US because of higher liquidity and lower volatility. Since there are no measures of stock market development, I follow Chan et al. (2005) and include to two measures of stock market development. Chan et al. (2005) use size of the stock market of each country as a proxy of stock market development because usually well-developed countries have large stock markets. Therefore, I use size of stock market measured by the stock market capitalization as a percentage of the country’s gross domestic product (GDP) as the first measure of stock market development. The second measure of stock market development is an emerging market dummy variable because emerging markets has not as well-developed stock markets as developed countries (Thapa and Poshakwale, 2010).

Further, the share of foreign investment in a particular country has relation to the economic development of that country (Chan et al., 2005). I include two economic development related variables. The first measure of country’s economic development is the log value of gross domestic product (GDP) per capita measured in US

3I calculate a measure of home bias following Fidora et al. (2007) and Thapa and Poshakwale (2010).

(14)

dollars (Thapa and Poshakwale, 2010). The second measure of economic development is the real growth rate of the gross domestic product (GDP) (Chan et al., 2005).

Investors tend to invest in countries, which provide strong protection of investors’ rights, and investors do not tend to invest in countries with poor protection of investors’ rights (La Porta et al., 1998). According to prior research, common-law countries protect investor rights better than in civil-law countries. Following Thapa and Poshakwale (2010), I use a legal dummy variable as a measure of investor protection. A dummy variable takes a value of one for common law countries and zero otherwise.

Further, following empirical literature I include variables of exchange rate volatility and potential diversification in my analysis. The returns on foreign assets vary not only because of assets’ specific risk, but also because of unpredictable fluctuations in exchange rates (De Santis and Gerard, 2009). I measure exchange rate volatility by a three-year moving average standard deviation of the trade weighted Real Effective Exchange Rate4 (REER) (Thapa and Poshakwale, 2010). De Santis and Gerard (2009) find that a need to diversify the risk of foreign assets across several countries affects foreign equity portfolio allocations. To measure the diversification potential between two countries, I use a correlation coefficient between country j and i based on the six years’ monthly returns (Thapa and Poshakwale, 2010).

Finally, Kirabaeva and Razin (2010) show that the information asymmetry between foreign and domestic investors leads to the inefficient portfolio investment allocations. I include two variables to capture the effects of the information asymmetry. The first variable is linguistic liaison because investors prefer to invest in foreign countries that share a common language (Faruqee et al., 2004). I construct a linguistic liaison dummy variable taking a value of one if a pair country shares a common language and zero otherwise. Kirabaeva and Razin (2010) summarize that international investments have negative relation to the distance between two countries, which is a proxy for informational asymmetry. Therefore, I include a variable of bilateral distance, which is the log value of the distance between two capital cities of country j and i (Faruqee et al., 2004). Linguistic liaison and bilateral distance predict the likelihood of information flow between countries, measure the barriers that foreign investors face when accessing information overseas, and indicate the extent of information asymmetry between foreign and domestic investors (Chan et al., 2005). Table 2 presents the definitions of the dependent and independent variables.

I add two measures of transaction cost and observed country specific and bilateral variables to the model to find whether they affect the cross-country variation of foreign equity portfolio allocations. I expect portfolio allocations to have negative relation to both measures of transaction cost, because foreign investors prefer to invest in markets with lower costs (Thapa and Poshakwale, 2010).

4Fidora et al. (2007) also use real effective exchange rate in their study. Thapa and Poshakwale (2010) use both real effective

(15)

TABLE 2

Variables and descriptions used in the regression analysis

This table presents the dependent and independent variables included in the regression analysis. The hypothesized signs represent expected relationship of the independent variable with dependent variable, derived from prior research.

Variable Hypothesized sign Variable description

ijt

w N/A

The equity investment in country j by investors from country i as a share of the total foreign equity investments of investors from country i for the year t

TC1jt - Implementation shortfall costs of country j measured in basis points

TC2jt - Commission of country j measured in basis points

HBijt - Home bias between country j and i

Growthjt + Economic development of country j measured by real growth rate of GDP in percentages

Investor

Protectionjt +

Legal dummy variable, which takes value of 1 for common-law countries and 0 for civil-law countries

Stock Marketjt + Size of stock market measured by the stock market capitalization as a percentage of the country’s GDP

Diversificationij +/- Potential diversification measured by correlation coefficients of country i’s and country j’s return data

Languageij + Common language dummy variable, which takes a value of 1 if two countries share the same language and 0 otherwise

EXj - Exchange rate volatility measured by a three year average standard deviation of country j’s real effective exchange rate

Developmentjt + Economic development of country j measured by log value of

GDP per capita in US dollars

Distanceij - Bilateral distance measured by log value of the distance between

two capital cities Emerging

Marketjt

-An emerging market dummy variable, which takes value of 1 for emerging markets and 0 otherwise

(16)

Methodology

The goal of this study is to find out whether transaction costs in country j influence foreign equity portfolio allocations from country i to country j. The dataset contains multiple cross-sections and time series. The data include information about portfolio holdings of 15 investor countries across 36 recipient countries over a period of 8 years (2002-2009). Therefore, the model contains 540 cross-sections and has 4320 observations. I use panel data regression models, because it captures the cross-sectional effects in the dataset while taking into account the effects of time variation.

Panel data are repeated observations over the time for the same set of cross-sections (Lee, 2005). There are two types of panel data regression models: fixed effects models and random effects models. Fixed effects models allow intercept of the regression model to vary cross-sectionally or over time, while the slope estimates are fixed both cross-sectionally and over time (Brooks, 2008). Random effects models allow intercept to differ cross-sectionally, but to be constant over time. The slope estimators, similarly to fixed effects models, are constant cross-sectionally and over time. However, unlike fixed effects models, random effects models assume that the intercepts for each cross-section arise from a common intercept, which is the same for all cross-sections and over time (Brooks, 2008).

One of the differences between fixed effects and random effects models is that the random effects models do not remove the time invariant explanatory variables from the regression estimation. Consequently, the impact of time invariant variables on the independent variable I estimate using random effects models. I use the random effects in most of my estimations because the regression model includes a number of time invariant explanatory variables, such as time dummies, country specific and bilateral variables. Additionally, regarding prior literature, I use fixed effects in some of my estimations because the key explanatory variables are time varying (Thapa and Poshakwale, 2010).

I use the following regression equation for the analysis

ijt ij jt ij jt j ij ij jt j jt ijt jt jt ijt

Year

rket

EmergingMa

ce

Dis

t

Developmen

EX

Language

ation

Diversific

t

StockMarke

otection

Growth

HB

TC

TC

w

' 13 12 11 10 9 8 7 6 5 4 3 2 1

tan

Pr

2

1

(4)

where wijtis foreign equity portfolio allocation weight from country i to country j at time t (the proportion of

country i’s total wealth invested in equities of country j). TC1jt is implementation shortfall costs of country j

at time t. TC2jt is commission of country j at time t. HBijt is home bias between country i and j at time t.

Growthjt is economic growth of country j at time t. Protectionj is investor protection in country j. Stock

Marketjt is stock market development of country j at time t. Diversificationij is diversification potential

(17)

volatility in a country j. Developmentjtis economic development of a country j at time t. Distanceijis distance

between country i and country j. Emerging Marketjtis a dummy variable for the emerging markets. Year is a

dummy variable for the year in which the data is recorded. εij is random variable that measures random

deviation of each section’s intercept term from the common intercept term (α), and varies cross-sectionally but is constant over time (Brooks, 2008). vijt is individual observation error term, which varies

cross-sectionally and over time (Brooks, 2008).

I perform a correlation matrix to test for multicollinearity between explanatory variables. Appendix A presents correlation matrix. The highest correlation coefficient is 0.65 between variables of economic development and emerging market dummy. I should remove explanatory variables with correlation coefficients above 0.7 or -0.7 from the regression model (Brooks, 2008). All correlation coefficients in Appendix A are below 0.7 or -0.7, therefore, explanatory variables have no correlation with each other. Thus, I add and remove variables from this model and it will not cause any biased results (Brooks, 2008).

The regression model should fulfil the assumption of homoscedasticity. In other words, the variance of error terms should be constant (Brooks, 2008). If error terms in the regression model do not have constant variance, they are heteroscedastic (Brooks, 2008). To control for possible heteroscedasticity in this model, I use White’s consistent standard errors.

Further, the error terms in the regression model are assumed to have no correlation with each other (Brooks, 2008). If error terms have correlation with each other, then they are autocorrelated. To test for autocorrelation in the regression model, I use Durbin-Watson test. Durbin-Watson test can take one of three important values: zero, two, and four (Brooks, 2008). If Durbin-Watson test takes value of two, there is no autocorrelation in explanatory variables. If Durbin-Watson test takes value of zero, there is perfect positive autocorrelation. In addition, if test takes value of four, there is perfect negative autocorrelation. Therefore, if Durbin-Watson statistic of my regression estimations is near two, there is little evidence of autocorrelation (Brooks, 2008).

(18)

Data

The data set of my analysis covers foreign equity portfolio holdings of EU 15 countries in 36 developed and developing countries between 2002 and 2009. The sample includes 18 European (16 EU members, Norway and Switzerland), two North American (US and Canada), 11 Asian (Turkey, Israel, China, Hong Kong, India, Indonesia, Japan, Korea, Malaysia, Singapore, and Thailand), two South American (Chile and Brazil), two Pacific (Australia and New Zealand), and Eastern European (Russia) countries. Appendix C presents the regional groups. I obtain the data from the International Monetary Fund’s (IMF) Coordinated Portfolio Investment Survey (CPIS)5 over period of 2002-2009. Bilateral investment holdings are denominated in US dollars.

In CPIS, 70 countries provide information about their foreign portfolio investment assets. CPIS presents portfolio investments according to instruments (equity and debt) and residence of issuer, which provides information about the destination of portfolio investment. The survey includes the financial market participants, the primary end-investors (e.g. banks, security dealers, pension funds, insurance companies, mutual funds, nonfinancial corporations, households), and primary custodians, who hold or manage securities on behalf of others.

While the CPIS provides the most comprehensive survey of international portfolio investment holdings, it is still subject to a number of important caveats. Most importantly, the CPIS is not able to address the issue of third country6 (can be intermediate holding companies or special investment holding companies) holdings and round tripping (is the practice of two or more companies trading securities back and forth at approximately the same price). For example, Company A may sell securities to Company B and agree to buy them back at the same price later. Round tripping creates the impression of a high trading volume, suggesting interest in assets or securities that may not actually be there. Due to round-tripping, German equity investment alone in Luxembourg was reported to be USD 152 billion in 2003, when Luxembourg’s stock market capitalization was less than USD 40 billion (Fidora et al., 2007). Moreover, the CPIS shows a very low degree of cross-border holdings by emerging market economies. In the absence of other financial data especially for this country group, it is difficult to check whether this reflects reality or is due to reporting omissions. Finally, the CPIS does not provide a currency breakdown, do not report investment under USD 500 000 and does not identify domestic security holdings.7

Fidora et al. (2007), who use CPIS, note that investment from developing countries into developed countries is almost negligible and most not reported by IMF. Thus, the developed country is a source of investment,

5

CPIS website is http://www.imf.org/external/np/sta/pi/datarsl.htm. On this website, I choose segment of Individual Economy Tables (By participant). Then from the list of countries, select a country that I am interested in and download data from Table - Reported Portfolio Investment Assets by Country of Nonresident Issuer: Equity Securities.

6

Multinational corporations use third country to conduct their investment strategies.

(19)

and I consider portfolio weights from the point of view of 15 investor countries based in Developed Europe. Ten countries are members of European Monetary Union (Austria, Belgium, Finland, France, Germany, Greece, Italy, Netherlands, Portugal, and Spain). Denmark, UK and Sweden are members of European Union (EU), but they keep their own currencies. Norway and Switzerland have limited integration into European Union’s single market.

TABLE 3

Summary of data, sources, and measure of the data

Data Source Measure

Foreign equity portfolio investment from country i

to country j IMF (CPIS) US dollars

Market capitalization of country j World Bank (WDI) US dollars

Implementation shortfall costs ITG Basis points

Commission ITG Basis points

Real growth rate of GDP World Bank (WDI) Percentages

Stock market capitalization (as percentage of GDP) World Bank (WDI) Percentages

Real Effective Exchange Rate BIS

-Performance of various country indices MSCI US dollars

The data availability dictates the selection destination countries, which are 24 developing countries and 13 emerging market economies. I note that due to the mentioned above caveats of the CPIS and following Fidora et al. (2007) I exclude some countries from the analysis, in particular financial centres such as Ireland and Luxembourg.

I retrieve the data for measures of transaction costs from Investment Technology Group (ITG) Global Trading Cost Review8based on ITG’s Peer Group Database. ITG’s Peer Group Database is database of ITG’s proprietary clients and comprises client execution data from more than 120 different large buy-side institutions such as mutual funds, pension funds, and insurance companies. However, ITG report individual transaction costs for only US, UK, Japan, and Canada. The rest of the countries, ITG divide into three major groups (Developed Europe ex. UK, Developed Asia ex. Japan, and All Emerging Markets) and report transaction costs specific for regional group. This disadvantage of ITG reporting can cause bias in results of this analysis. Transaction costs are denominated in basis points.

8

(20)

I obtain the data for market capitalization (US dollars), real growth rate of GDP, GDP per capita, and stock market capitalization (as percentage of GDP) from World Development Indicator (WDI) of the World Bank. Moreover, I source the data for Real Effective Exchange Rate (REER) from the Bank of International Settlement (BIS). The BIS REER basket includes 52 economies, including the emerging countries. Finally, I retrieve the data to calculate diversification from Morgan Stanley Capital International (MSCI) website9, which reports historical performance of various country indices. Table 3 presents a summary of data, its sources and measure used in this study.

Descriptive statistics: key stylized facts

I will present the descriptive statistics of the variables and talk about observed key facts of the data set in this section. Appendix D provides descriptive statistics of the dependent and independent variables of this study over the period of 2002-2009. Further, I will discuss each variable in more detail.

I collect the foreign equity portfolio holding data for EU 15 countries. The data on holdings contain annual observations between 2002 and 2009. EU 15 countries invest in 24 developed and 13 developing countries all over the world. Table 4 presents an overview of foreign equity portfolio holdings for the total sample in the year 2009.

Table 4 shows that the foreign equity portfolio holdings exhibit some interesting facts. The table indicates that in the year 2009 total foreign equity holdings of EU15 countries are 3,257,843 million US dollars out of which the biggest share of nearly 50 percent are located in Developed European markets. Interestingly, the table shows that the EMU countries attract nearly 41 percent of the EU 15 countries’ total holdings and other EU countries 9 percent. Further, the table indicates that holdings in US and Canada are 979,101 million US dollars and account for nearly 30 percent of the total holdings. The holdings in Emerging markets are 282,339 million US dollars, which are nearly 9 percent of the total holdings. In addition, the holdings in Developed Asia are 369,841 million US dollars, which are nearly 11 percent of the total holdings.

Next, foreign equity portfolio holdings change over time. Appendix E illustrates how median foreign equity portfolio holdings of EU 15 countries change over the period of 2002-2009. The graph in Appendix E shows the median value of EU 15 countries’ holdings. In general, the median value of holdings is steadily rising over the 8-year period across all EU 15 countries. However, there is a sharp fall in foreign holdings in 2008 for all EU 15 countries. This decrease in foreign holdings can be a consequence of financial crisis in 2008.

(21)

TABLE 4

Total holdings and holdings of EU15 by region in the year 2009

This table presents total holdings of EU 15 countries in 2009 in Column 2. The rest of the columns present total holdings across different regions in 2009. The holdings are in millions US dollars.

Country

Foreign equity portfolio holdings

T o ta l U S U K Ja p an C an ad a D ev el o p ed E u ro p e ex . U K D ev el o p ed A si a ex . Ja p an A ll E m er g in g M ar k et s Netherlands 443,154 205,520 58,063 18,721 6,442 94,985 26,141 33,282 Austria 42,397 6,151 4,073 765 658 25,477 995 4,277 Belgium 99,384 13,668 7,378 2,069 972 71,446 1,092 2,760 Denmark 95,851 31,606 11,724 5,078 1,145 34,279 3,476 8,543 Finland 54,260 12,369 7,977 536 156 29,228 1,040 2,955 France 415,588 60,798 55,263 23,072 5,462 221,409 10,917 38,668 Germany 292,694 48,455 36,953 9,926 5,729 167,448 7,768 16,416 Greece 13,340 7,733 4,551 3 52 766 13 222 Italy 116,238 17,680 8,286 2,877 760 80,017 1,703 4,915 Norway 319,255 91,032 47,228 16,741 9,034 114,037 10,888 30,295 Portugal 21,839 3,706 2,632 259 102 12,190 76 2,874 Spain 63,444 10,242 5,255 0 191 42,955 0 4,802 Sweden 188,321 58,924 26,285 8,328 3,961 69,302 6,872 14,651 Switzerland 191,367 61,884 21,526 10,644 7,074 73,172 6,992 10,075 UK 900,711 299,813 - 110,163 7,782 292,660 82,687 107,606 Total 3,257,843 929,581 297,193 209,181 49,520 1,329,370 160,660 282,339

Source: Author’s calculations based on data from CPIS (IMF)

Table 5 presents the distribution of EU15 countries’ average foreign equity portfolio allocations (in percent) across 36 countries worldwide. It also reports the average total foreign equity portfolio holdings of EU15 countries (second row) over entire 2002-2009 period.

(22)

TABLE 5

Foreign equity portfolio allocations

This table presents the distribution of 15 investor countries’ average foreign equity portfolio allocations (in percent) across 36 destination countries over the period of 2002-2009. The second row presents the average total foreign equity portfolio holdings of 15 investor countries. The second column presents a country j’s average stock market capitalization weight in the world market portfolio.

(23)

TABLE 5 - Continued

M a rk et ca p it a l w ei g h t N et h er la n d s A u st ri a B el g iu m D en m a rk F in la n d F ra n ce G er m a n y G re e ce It a ly N o rw a y P o rt u g a l S p a in S w ed en S w it ze rl a n d U K Hong Kong 2.50 1.15 0.63 0.42 1.36 0.79 0.80 0.58 0.08 0.63 0.72 0.08 0.08 0.98 0.67 3.21 Korea 1.49 1.11 0.51 0.24 1.47 0.57 0.87 0.59 0.04 0.82 1.40 0.04 0.03 0.71 0.69 2.61 Norway 0.43 0.41 0.38 0.39 1.31 2.04 0.37 0.59 0.16 0.26 0.05 0.13 2.90 0.39 0.82 Austria 0.24 0.33 0.35 0.42 0.40 0.45 1.66 0.10 1.31 0.36 0.50 0.31 0.23 1.07 0.40 Denmark 0.41 0.42 0.28 0.29 1.64 0.26 0.43 0.09 0.17 1.97 0.08 0.05 1.21 0.32 0.57 India 1.63 0.30 0.61 0.19 0.88 0.43 0.49 0.18 0.02 0.26 0.26 0.11 3.30 0.20 0.11 0.89 South Africa 1.26 0.47 0.19 0.18 0.43 0.10 0.24 0.23 2.37 0.27 0.41 1.65 0.03 0.11 0.38 0.62 China 5.28 0.39 0.52 0.19 0.86 0.14 0.87 0.41 0.52 0.47 0.55 0.04 0.43 0.41 0.30 1.43 Portugal 0.20 0.33 0.10 0.26 0.12 0.16 0.39 0.27 0.09 0.56 0.25 3.47 0.10 0.10 0.34 Singapore 0.63 0.64 0.29 0.11 0.33 0.47 0.23 0.30 0.00 0.27 0.46 0.02 0.01 0.41 0.37 1.20 Greece 0.32 0.45 0.26 0.50 0.24 0.24 0.44 0.48 0.47 0.37 0.09 0.19 0.33 0.22 0.46 Mexico 0.59 0.35 0.10 0.06 0.62 0.06 0.18 0.09 0.02 0.20 0.34 0.01 0.91 0.19 0.24 0.72 Turkey 0.33 0.14 0.77 0.10 0.25 0.00 0.08 0.19 0.81 0.28 0.22 0.00 0.22 0.32 0.18 0.36 Poland 0.23 0.11 0.89 0.14 0.35 0.29 0.19 0.43 0.23 0.10 0.25 0.29 0.03 0.26 0.06 0.12 Hungary 0.07 0.09 0.99 0.06 0.23 0.08 0.19 0.11 0.32 0.06 0.13 0.04 0.02 0.15 0.07 0.13 Thailand 0.30 0.21 0.13 0.07 0.30 0.21 0.08 0.16 0.01 0.19 0.10 0.00 0.00 0.06 0.12 0.49 Malaysia 0.52 0.21 0.09 0.04 0.28 0.13 0.10 0.05 0.00 0.14 0.08 0.00 0.00 0.08 0.09 0.44 Israel 0.32 0.17 0.10 0.08 0.30 0.05 0.06 0.10 0.06 0.14 0.10 0.01 0.01 0.05 0.23 0.20 Indonesia 0.25 0.16 0.08 0.03 0.15 0.03 0.07 0.08 0.00 0.11 0.05 0.00 0.03 0.05 0.08 0.27 Chile 0.34 0.08 0.01 0.02 0.05 0.00 0.03 0.00 0.00 0.05 0.02 0.00 0.26 0.03 0.02 0.08 New Zealand 0.10 0.03 0.01 0.01 0.04 0.00 0.03 0.02 0.00 0.08 0.06 0.00 0.00 0.03 0.03 0.10 Developed countries 85.1 95.3 91.9 98.1 92 95.5 95.7 96.1 95.9 96 94.9 92.6 92.8 95 95.9 90.7 Developing countries 14.9 4.7 8.1 1.9 8 4.5 4.3 3.9 5.1 4 5.1 7.4 7.2 5 4.1 9.3 Total: 100. 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100

(24)

Table 5 includes 23 developed countries and 14 developing countries. The table shows that foreign equity holdings do not spread evenly across countries. Nearly 95 percent of the foreign equity holdings are concentrated in developed countries and only 5 percent of foreign equity holdings are in developing countries. Moreover, only developed countries are in the top ten markets where EU 15 countries allocate the biggest number of holdings (US, UK, France, Germany, the Netherlands, Japan, Spain, Switzerland, Italy, and Sweden). Among developing countries, Brazil receives the highest foreign equity portfolio allocation weight (~1%), and among developed countries, US the highest weight (~26%). In the bottom ten countries receiving the lowest portfolio allocation weights, there are eight developing countries but only two developed countries (New Zealand, Chile, Indonesia, Israel, Malaysia, Thailand, Hungary, Poland, Turkey, Mexico.

To highlight how EU15 countries overweight or underweight foreign markets, I present the average world’s market capitalization weight of each country in Column 2 (Table 5). In general, EU15 countries tend to underweight foreign markets in their equity portfolios. The table shows that the portfolio allocations generally are smaller than the world’s market capitalization weights of foreign markets. However, there is some evidence that EU15 countries also tend to overweight some foreign markets in their equity portfolios. Mostly, EU15 countries overweight countries within European Union. For example, the table indicates that Austrian investors overweight Germany in their equity portfolios (28.58% vs. 3.21%), Belgian investors overweight the French market (34.95% vs. 4.42), and Finish investors overweight the Swedish market (23.49% vs. 0.95%). Appendix F presents the average, median, maximum, minimum, and standard deviation in foreign equity portfolio allocations to 24 developed and 13 developing countries between 2002 and 2009.

Figure 1 illustrates how foreign equity portfolio allocations to the developed and developing countries change over time. This figure shows that there is only a small shift in portfolio allocation weights over entire 2002-2009 period. A share of foreign equity allocated to developing countries by EU 15 countries decreased by nearly 5 percent from 2002 to 2009. As a result, a share of foreign equity allocated to developing countries increased by nearly 65 percent.

(25)

FIGURE 1

Median value of foreign equity portfolio allocations

This figure shows the median value of foreign equity portfolio allocations to the developed and developing countries over the period of 2002-2009.

Source: Author’s calculations based on CPIS (IMF)

The table shows that both measures of transaction cost are lower in developed countries. The median implementation shortfall costs range from 38 basis points (UK) to 94 basis points (Emerging markets). The median commission ranges from 9.6 basis points (Japan) to 29 basis points (Emerging markets). This indicates that the lowest implementation shortfall costs are in UK and the lowest commissions are in Japan, while the highest implementation shortfall costs and commissions are in developing countries.

Further, the table reports that the median total transaction costs in UK are 50 basis points, in US are 58.5 basis points, in Canada are 62.9 basis points, and in Japan are 79.5 basis points. The medians for Developed Europe (ex. UK) and Developed Asia (ex. Japan) are 60.9 basis points and 86.6 basis points, respectively, while median total transaction costs in developing countries are 121.5 basis points. The medians show that UK has the lowest total transaction costs followed by US, Canada, and Developed European countries. Overall, transaction costs in developing countries are higher comparing to developed countries on the average by 45 percent.

97.3% 96.6% 96.3% 95.2% 94.3%

92.9% 94.7% 92.4%

2.7% 3.4% 3.7% 4.8% 5.9% 7.1% 5.3% 7.8%

2002 2003 2004 2005 2006 2007 2008 2009

(26)

TABLE 6

Statistics on transaction costs in the different regions over period of 2002-2009

This table reports the average, median, maximum, minimum, and standard deviation in measures of transaction costs over the entire period of 2002-2009. The top panel shows data on implementation shortfall costs; the middle panel shows data on commissions. The bottom panel shows data on total transaction costs, which are the sum of both measures of transaction cost. Implementation shortfall costs, commissions, and total costs are in basis points.

Region

Transaction Costs Implementation shortfall costs

Average Median Maximum Minimum Standard deviation

UK 41.4 38.0 78.7 17.0 13.8

US 54.3 49.0 124.0 26.0 22.5

Developed Europe ex. UK 56.8 47.7 161.0 35.0 25.4

Canada 57.7 50.3 95.4 26.9 25.1

Developed Asia ex. Japan 71.3 68.0 126.0 35.0 20.4

Japan 74.4 70.3 146.0 49.0 22.5

All Emerging Markets 95.6 94.0 143.0 66.0 16.4

Commissions

Average Median Maximum Minimum Standard deviation

Japan 9.5 9.6 12.0 6.0 1.6

US 11.0 9.9 19.0 7.0 3.7

Canada 11.4 11.6 13.2 9.0 1.4

UK 11.6 12.0 14.0 8.0 1.3

Developed Europe ex. UK 13.0 12.5 18.0 8.0 3.0

Developed Asia ex. Japan 17.3 16.5 22.0 12.0 3.5

All Emerging Markets 41.7 29.0 189.0 18.0 39.6

Total costs

Average Median Maximum Minimum Standard deviation

UK 53.1 50.0 88.0 31.0 13.2

US 65.3 58.5 143.0 33.0 25.4

Canada 69.1 62.9 106.7 38.6 24.5

Developed Europe ex. UK 69.8 60.9 178.0 45.4 26.7

Japan 83.9 79.5 157.0 57.7 23.3

Developed Asia ex. Japan 88.5 86.6 147.0 54.0 20.6

All Emerging Markets 137.3 121.5 292.0 96.0 45.1

Source: Author’s calculations based on ITG’s Global Trading Costs Review 2006-2009

(27)

transaction costs declined due to implementation of automated trading systems and higher completion in stock markets. Moreover, the figure shows that transaction costs are higher in developing countries comparing to developed countries.

FIGURE 2

Total transaction costs in different regions over the period of 2002-2009

This figure shows total transaction costs in US, UK, Japan, Developed European countries (ex. UK), Developed Asian countries (ex. Japan), and Emerging markets from 2002 to 2009. Total transaction costs are a sum of implementation shortfall costs and commissions. Total transaction costs are in basis points.

Source: ITG’s Global Trading Costs Review 2006-2009

Table 7 presents the descriptive statistics of all explanatory variables for each country between 2002 and 2009. The table reports data on home bias, variables of economic development and stock market development in Panel A. The table also provides data on variables of information asymmetry, investor protection, exchange rate volatility, and potential diversification in Panel B. The table presents average values for each country in the sample over the period of 2002-2009.

Column 2 of Table 7 (Panel A) reports data on home bias variable. As defined earlier, home bias shows the extent to which foreign markets are underweight in domestic investors’ equity portfolio holdings (Chan et al., 2005). I calculate a measure of home bias according to Equation (2) described in Section 3. Table 7 shows the distribution of the average home bias of EU 15 countries across the developed and developing countries in the sample. Generally, average home bias varies significantly across countries. Home bias ranges from 0.5 (the Netherlands) to 2.16 (Chile).

25 75 125 175 225 275 2 0 0 2 ( Q 3 ) 2 0 0 2 ( Q 4 ) 2 0 0 3 ( Q 1 ) 2 0 0 3 ( Q 2 ) 2 0 0 3 ( Q 3 ) 2 0 0 3 ( Q 4 ) 2 0 0 4 ( Q 1 ) 2 0 0 4 ( Q 2 ) 2 0 0 4 ( Q 3 ) 2 0 0 4 ( Q 4 ) 2 0 0 5 ( Q 1 ) 2 0 0 5 ( Q 2 ) 2 0 0 5 ( Q 3 ) 2 0 0 5 ( Q 4 ) 2 0 0 6 ( Q 1 ) 2 0 0 6 ( Q 2 ) 2 0 0 6 ( Q 3 ) 2 0 0 6 ( Q 4 ) 2 0 0 7 ( Q 1 ) 2 0 0 7 ( Q 2 ) 2 0 0 7 ( Q 3 ) 2 0 0 7 (Q 4 ) 2 0 0 8 ( Q 1 ) 2 0 0 8 ( Q 2 ) 2 0 0 8 ( Q 3 ) 2 0 0 8 ( Q 4 ) 2 0 0 9 ( Q 1 ) 2 0 0 9 ( Q 2 ) 2 0 0 9 ( Q 3 ) 2 0 0 9 ( Q 4 ) T ra n sa ct io n c o st s (b as is p o in ts ) US UK Japan

(28)

TABLE 7

Summary statistics for the explanatory variables over the period of 2002-2009

This table shows seven sets of explanatory variables for each country. Panel A presents home bias, two variables of economic development, and two variables of stock market development. Home bias reflects the deviation of the share of country j in foreign equity portfolio holdings for each country i from the world’s market capitalization weight of country

j. Variables of Economic development include gross domestic product (GDP) per capita (in US dollars) and real GDP

growth. Variables of Stock market development include stock market capitalization as a percentage of GDP and emerging market dummy variables. Panel B presents two variables of information asymmetries and variables of investor protection, potential diversification, and exchange rate volatility. Variables of information asymmetries include average common language dummy variable and average distance (in kilometers). Investor protection is a legal system dummy variable. Potential diversification is average return correlation coefficients, and exchange rate volatility is average standard deviation. The table shows average values for the sample period of 2002-2009.

Panel A

Home bias

Economic development Stock market development

(29)

Portugal 0.94 18,215 0.37 41.31 0 Russia 1.48 6,417 4.89 65.34 1 Singapore 1.42 30,937 6.02 190.07 0 South Africa 1.71 4,860 3.68 218.92 1 Spain 0.96 26,877 2.19 88.15 0 Sweden 0.87 41,871 1.68 101.59 0 Switzerland 0.75 50,988 2.06 239.63 0 Thailand 1.58 3,011 4.21 62.75 1 Turkey 1.45 6,812 4.58 27.91 1 UK 0.72 37,216 1.36 125.17 0 US 1.23 42,672 1.71 123.24 0 (Continued)

TABLE 7 - Continued

Panel B

Information asymmetries Investor protection Potential

(30)

Norway 1,397 0.00 0 0.72 3.42 Poland 1,246 0.00 0 0.66 4.68 Portugal 2,584 0.00 0 0.68 0.62 Russia 2,424 0.00 0 0.60 3.68 Singapore 10,371 0.07 1 0.72 1.17 South Africa 9,388 0.07 0 0.67 4.80 Spain 1,723 0.00 0 0.73 1.08 Sweden 1,464 0.00 0 0.72 2.77 Switzerland 1,760 0.36 0 0.70 1.08 Thailand 9,006 0.00 0 0.60 1.89 Turkey 2,346 0.00 0 0.56 3.97 UK 1,141 0.00 1 0.76 3.58 US 6,570 0.07 1 0.73 3.52

Source: Author’s calculations based on data from World Bank (WDI), MSCI, and BIS

Columns 3 and 4 of Table 7 (Panel A) show variation in the two measures of economic development across 36 countries over the period of 2002-2009. Chan et al. (2005) suggest that cross-sectional variation in GDP per capita and real GDP growth capture different aspects of economic development in each country. The developed countries have the highest values of average GDP per capita. The table reports that the top countries in terms of average GDP per capita are Norway (67,680 US dollars), Switzerland (50,988 US dollars), Denmark (48,762 US dollars), US (42,672 US dollars), and the Netherlands (40,854 US dollars). On the other hand, the mostly developing countries have the highest average real GDP growth. The top countries in average real GDP growth are China (10.76%), India (7.70%), Singapore (6.02%), Russia (4.89%), and Turkey (4.58%).

Columns 4 and 5 (Panel A) show two variables of stock market development. The first variable of stock market development for each country is the stock market capitalization as a percentage of country’s GDP, which measures the stock market size. The table shows that the stock market size ranges from 25.33 percent in Hungary to 438.76 percent in Hong Kong. Following Hong Kong, the top stock markets are in Switzerland (239.63%), South Africa (218.92%), and Singapore (190.07%). In addition, the smallest stock markets are in Turkey (27.91%), Mexico (27.47%), and Poland (29.95%). The second variable of stock market development is emerging market dummy variable. The table indicates that there are 14 developing countries and 23 developed countries in the sample.

(31)

indicates that European countries enjoy the closest distance between capital cities comparing with other countries. The distance between two European countries ranges from average 911 km to 2584 km. The most remote countries are Chile (11,354 km), Australia (16,236 km), and New Zealand (17,160 km).

Next, Column 4 (Panel B) reports the variable of investor protection, which is a legal system dummy variable. The dummy variable takes value of one for common-law countries and zero for civil-law countries. The table shows that eight countries in the sample have common law systems and 29 countries have civil law systems.

(32)

IV.

Results

To research whether transaction costs influence foreign equity portfolio allocations of EU15 countries, I run several regressions. Following the prior empirical research, I have also included a number of control variables that could explain the variation in equity allocations across different countries. Independent and control variables are regressed against foreign equity portfolio allocations for the total sample over the period of 2002-2009. I employ random effects model to the most of the regression estimations, because my model includes a number of dummies and rarely changing variables (Thapa and Poshakwale, 2010).

The results in Table 8 show that, when using random effects, all independent and control variables are significant at 1 percent significance level except common language dummy variable, which is not significant when regressed together with implementation shortfall costs (TC1). Adjusted R-squared values (Table 8) indicate that the model explains nearly 84 percent of deviations in foreign equity portfolio allocations of EU15 countries. Thapa and Poshakwale (2010) find similar adjusted R-squared values (nearly 0.80) in their research. The slight difference in values of adjusted R-squared can be due to the differences in the sample period and number of observations. Thapa and Poshakwale (2010) conduct their research over the period of 6 years (2001-2006) and include 3290 observations. On the other hand, my period of analysis is 8 years (2002-2009) and I include 4320 observations.

The table shows that both measures of transaction cost carry expected negative sign, which are consistent with results of prior research. The results indicate that an increase in transaction costs will decrease the foreign equity portfolio allocations. This finding confirms the effect of transaction costs on foreign equity portfolio allocations and is consistent with the findings of Chan et al. (2005), Thapa and Poshakwale (2010).In general, investors of EU15 countries prefer to allocate their portfolio investments to the markets with lower transaction costs.

Home bias has a negative relation to the foreign equity portfolio allocations and is consistent with the view that if investors exhibit high bias towards domestic securities, they will invest less in foreign markets. The negative sign of home bias is consistent with the finding of Thapa and Poshakwale (2010). The results (Table 8) indicate that an increase in home bias decreases the weight of foreign equity portfolio allocation by nearly 1.3 percent.

(33)

TABLE 2

Regression analysis of foreign equity portfolio allocations

In all regressions the dependent variable is the log value of country wise bilateral foreign equity portfolio allocation from country i in country j at time t (wijt). The independent variables are the two measures of transaction cost in basis

points, all control variables and time dummy variables. The second column shows the expected sign based on prior empirical literature. Test-statistics are given in parentheses.

Expected sign TC1 (IS costs) TC2 (Commission)

Transaction costs - -0.002 *** (-6.78) -0.003*** (-3.97) Home bias - -1.288 *** (-149.77) -1.276*** (-90.66) Economic growth + 0.014 *** (5.26) 0.017*** (3.79)

Stock market development + 0.001

*** (9.09) 0.002*** (14.43) Investor protection + 0.167 ** (2.52) 0.116*** (4.26) Diversification + 0.783 *** (2.93) 0.781*** (7.42)

Common language dummy + 0.096

(1.03)

0.114*** (3.11)

Exchange rate volatility - -0.041

*** (-3.10) -0.042*** (-7.67) Economic development + 0.095 *** (6.08) 0.071*** (3.07)

Emerging market dummy - -0.324

*** (-5.03) -0.157*** (-4.50) Bilateral distance - -0.258 *** (-4.13) -0.260*** (-9.98) Year dummy 2003 + 0.098 *** (5.34) 0.020 (1.35) Year dummy 2004 + 0.125 *** (5.50) 0.010 (0.63) Year dummy 2005 + 0.160 *** (6.62) 0.032** (2.03) Year dummy 2006 + 0.147 *** (6.26) 0.029* (1.71) Year dummy 2007 + 0.159 *** (6.92) 0.047*** (2.87) Year dummy 2008 + 0.247 *** (12.99) 0.169*** (10.94) Year dummy 2009 + 0.305 *** (11.53) 0.180*** (9.34) Overall adjusted R2 0.84 0.83 Number of observations 4311 4311

Referenties

GERELATEERDE DOCUMENTEN

In the Jabodetabekpunjur spatial planning evaluation process, the coordination between the government agencies in national, provincial and local levels concerning

The effectives of (high) social spending on (high) antipoverty effects of social transfers and taxes faded away during last decade. Less targeting partly offers an explanation

The table shows the results of the spanning and efficiency tests performed on country indexes created from domestic government bond returns and industry indexes created

A study of a case in the Netherlands, the Second Coentunnel showed how transaction costs in practice appear, in which stage of the purchasing process these cost arise and

Instead of taking the risk of price variations in the day-ahead market, purchasers seek protection, depending on their risk appetite, and manage a portfolio of derivative contracts

routines - aware of the existing recruitment routines and taking these into account regarding recruitment - fully aware of consequences of the existing recruitment routines in the

Dat Claes’ roman een intertekstueel hoogstaand borduurwerk- je is, verrast me niet; ik ren daar niet voor naar de boekwinkel, zelfs niet na lezing van de vak-

De inhoud uit deze module mag vrij gebruikt worden, mits er gebruik wordt gemaakt van een bronvermelding:. MBO module Mondzorg, ZonMw project “Mondzorg bij Ouderen; bewustwording