Effect of European economic integration on equity-home bias of European
investors
Name: Jasper Krikke
Student number: 10245790
Specialization: Economics and Finance
Supervisor: Lin Zhao
Table of contents
1. Introduction ………3
2. Theoratical framework…..………4
2.1 Modern portfolio theory………..4
2.2 The gravity model………...6
3. Literature review……….……..7
3.1 Explanations home bias……….….8
4. Methodology and Data ………11
4.1 Methodology..………11
4.2 Data description……….13
5. Results………..14
5.1 Analysis………...14
5.2 Discussion………...…....18
6. Conclusion………19
7. Bibliography……….22
21. Introduction
Along with the globalization of the world economy, the financial world becomes more internationally orientated as well. Technological breakthroughs and government cooperation all over the world have started to eliminate barriers to international capital flow. As it
becomes easier and less costly for capital to flow all over the world, one would expect highly internationally diversified portfolios as standard portfolio theory predicts that the optimal portfolio is the world market portfolio. This highly international diversified portfolio is however rarely seen in practice. Evidence shows that even today investors all over the world still have a strong preference for home equity. This empirical observation is known as the equity home-bias puzzle.
Benefits of portfolio diversification have been widely acknowledged for decades. The technology breakthroughs in the last decade have made it possible for individuals to invest directly in foreign equity. In addition, information on firms has become globally accessible to individual investors. Given these developments, equity home-bias would be expected to decrease sharply. Although recent empirical evidence indeed show a decreasing equity home-bias, at least for developed countries, it also reveals that investors still have a strong
preference for home equity. Research on the equity home-bias has investigated various types of barriers to international investment, reaching from explicit economic barriers, to various types of implicit barriers. Government cooperation all over the world has decreased explicit barriers to international capital flow. The question of interest in this paper is to what extent the removal of explicit barriers to international capital flow can decrease equity home-bias.
No region in the world has tried to eliminate the economic barriers across countries in the last decade as hard as the European Union has. With the formation of the European Union and the European Monetary Union, Europe has tried to create a zone without restrictions to either trade, movement of people or capital flow among its member states (Bekaert, Harvey, Lundblad and Siegel, 2013). The effect of removal of economic barriers to international capital flow, will therefore be examined by looking at the effect of the formation of the
European Union and the introduction of the Euro on cross-border equity holdings of European investors. The sample contains 21 countries from all over the world. Cross-border equity holdings between these countries over the period 2001-2008 have been used to evaluate the effect of the European economic integration on cross-border equity holdings in these
countries. To separate the effect of removal of economic barriers from other effects, economic size, distance and sharing of same border and language are used as control variables in the
performed regressions, as these variables are identified in the literature to affect cross-border equity holdings.
The structure of this paper is as follows. In section 2 a theoretical framework will be developed covering modern portfolio theory and the gravity model for international capital flow. Section 3 will be a literature review on the equity home bias. In section 4 there will be a description of the methodology and the data used to test the impact of formation of the
European union and European monetary union on cross-border equity holdings. In section 5 the results will be analyzed and discussed and section 6 concludes.
2. Theoretical Framework
In order to understand why the observed phenomenon of the home-bias is so puzzling to financial experts, standard portfolio theory is first examined. Standard portfolio theory derives the ‘optimal’ portfolio for investors to hold, so that deviations from this portfolio would mean inefficient allocation. Equity home-bias is often defined as the deviation from this optimal portfolio. As deviations from the efficient portfolio in the real world are very large, the second part examines the determinants of cross-border holdings using the traditional gravity model.
2.1 Modern portfolio theory
Modern portfolio theory is built on the idea that investors seek to maximize expected portfolio returns given its level of risk. The level of risk of a stock is given by the volatility of its return. Another equivalent way of looking at the derivation of the efficient portfolio therefore is to minimize portfolio volatility given a fixed return.
The higher the volatility of a stocks return, the higher its risk premium will be (Berk and Demarzo, 2011). Efficient diversification however can decrease portfolio volatility, without sacrificing the expected return (Markowitz, 1952). Markowitz (1952) pointed out that that not only the number of stocks and their weights in a portfolio and the volatility of the individual stocks determined the volatility of the portfolio, but that its volatility was also determined by the covariance of the included stocks. Stocks with a negative covariance would mean the returns of these stocks would move in opposite direction. This means including stocks with a low or negative covariance can be used to stabilize expected returns, or decrease portfolio volatility without sacrificing expected return (Markowitz, 1952). This decrease in portfolio volatility is due to the elimination of idiosyncratic risk (Berk and DeMarzo, 2010). Theory distinguishes two broad types of risk to which all stocks are exposed: market risk and
firm specific risk. In a large and well diversified portfolio, portfolio return is not dependent on individual stock performance, such that firm-specific risk is diversified away(Body, Kane and Marcus, 2011). The portfolio with the lowest possible risk would then be the market portfolio, where all firm-specific risk has been diversified away and the portfolio is only subject to market risk (Berk and Demarzo, 2011).
In addition, Sharpe (1964) showed that by using three simplified assumptions, it induces all investors to hold the market portfolio. The first assumption is that investors only hold efficient portfolios. Although in the real world, this is not observed, it’s reasonable to assume investors maximize utility by choosing the efficient portfolio. The second assumption is that all securities can be bought and sold at competitive market prices and investors are able to borrow and lend infinitely at a known risk-free interest rate. This is an assumption that does not apply in the real world due to the presence of transaction costs and other trading costs. The last assumption is that all investors have access to the same information and therefore all investors make the same predictions about risk and return. This is another unrealistic
assumption. As these assumptions do not accurately describe the real world, its predictions on investor behavior deviate from what is seen in the real world. However, this theory can be used to show that the market portfolio can be used as a representation of the efficient portfolio (Berk and DeMarzo, 2011).
Early research by Grubel (1968) used the insights of efficient diversification put forward by Markowitz, to show the benefits from international diversification. Besides firm specific risk, stocks are also exposed to some extent to country-specific risk (Bodie, Kane and Marcus, 2011). This risk category is a market risk when looking at a specific country, but when looking at the world market, country specific risk may be diversified away just like firm-specific risk (Grubel, 1968). Although the theory of international diversification is exactly the same as theory regarding domestic diversification with the exception that the world market portfolio now represents the efficient market portfolio, there are a few
complicating factors. Firstly international investing brings two additional risks, which might be hard to predict: Exchange rate risk and political risk (Bodie et al., 2012). Exchange rate risk refers to possible fluctuations in exchange rate and political risk refers to the risk of political events in a specific country affecting the economy. A second complicating factor in the theory of international diversification is the fact that barriers, such as transaction costs, are higher to international capital flow than barriers to domestic capital flow, and therefore far more interfering the capital market. The tendency towards reduction of barriers to
international capital flow over the last decade has however largely increased possible benefits of international diversification.
2.2 Gravity model
Observed cross-border holdings are not even close to what modern portfolio theory would expect them to be (Coeurdacier and Rey, 2012). A natural follow-up question is what determinants affect cross-border equity holdings. In the international trade of goods, one of the most accurate prediction models of trade flows is known as the gravity model, in spite of its incompetency to explain trade patterns fully (Anderson, 2011). Based on empirical research, Portes and Rey (2005) showed that international financial asset holdings can be largely explained by an almost identical gravity model. The only real difference between the original gravity model for trade of goods and the gravity model for financial asset holdings used by Portes and Rey (2005) is the definition used for economic size of a country. In the original gravity model for trade of goods, economic size is often defined as the gross domestic product of a country where in the gravity model for equity flows Portes and Rey (2005) use market capitalization to represent economic size of a country.
The simple gravity model for international trade in goods is based on Newton’s law of gravity. It states that economic size of both trading countries affects bilateral trade positively while distance between the two countries has a negative effect on trade (Krugman, Obstfeld and Melitz, 2012). The simple model can be extended easily by additional determinants such as same language spoken in the trading countries, cultural and religion similarities and sharing of a common border by the two countries. Empirical research on the extended gravity model has proven the model to be one of the most accurate models in describing international trade (Fagiolo, 2010).
Portes and Rey (2005) use the original gravity model of trade as a starting point for the research on determinants of cross-border financial holdings. Although effects of the different determinants of the gravity model differ between goods and equity, Portes and Rey (2005) find that the gravity model does at least as well in explaining international equity flows as it does in explaining international trade flows. Coeurdacier and Martin (2009) confirm the view that similar determinants drive the international flow of goods and capital. They find that the traditional variables of the gravity model, including common extension of language and cultural similarities, also have significant effects on cross-border asset holdings.
Balta and Delgado (2008) also make use of the gravity model when looking at the impact of the formation of the European union on cross-border equity holdings. In their
performed regressions, they use GDP as the measure of economic size of a country and find effects of GDP of both countries and sharing of the same language and border are positive and significant, and distance between the countries has a negative effect, just like the gravity model expects. Using the gravity model for estimating the cross-border holdings of a domestic country in a country with the same economic size and distance as its own, shows that investors all over the world also invest a disproportionate large share in domestic equity when compared to a gravity model benchmark (Balta and Delgado, 2008).
3. Literature review
The home bias equity-puzzle is the observed fact that investors all over the world invest disproportionately more in domestic stocks than what theory would suggest to be the optimal allocation (Lau et al., 2010). In modern portfolio theory, it is derived that the mean-variance efficient portfolio is the market portfolio. This would mean that according to the theory, investors optimally hold a proportion of their portfolio in domestic stock equal to the proportion of their domestic market capitalization of the total world market capitalization. Holdings in excess of this proportion would be the proportion of home-bias of the investors. Explicit barriers to international capital flow and several other explanations might make it optimal to invest more in domestic stocks than theory would predict. The actual observed home-bias is however considered to be too high to be optimal even given barriers to international capital flow (French and Poterba, 1991).
The home-bias has some important implications for both investors and firms. The implications for investors can be explained by the diversification benefits highlighted by mean-variance analysis. As investors invest in foreign countries they may diversify away the country-specific risk, possibly leading to a higher expected return given the level of volatility of the portfolio is fixed (Stulz, 1999). In addition to these implications to investors, there exists a positive relationship between the degree of home-bias and a countries cost of capital (Lau et al., 2010). Different measures are used for a countries cost of capital and a countries overall home-bias, but in all cases they find that increasing international risk sharing reduces cost of capital, where risk sharing in equity market is simply the sharing of country-specific risk between domestic and foreign investors.
As benefits of international diversification have been widely recognized, there has been a tendency towards more liberalized capital markets all over the world. This led to a sharp decrease in home-bias from 1988 onwards (Coeurdacier and Rey, 2012). Besides the
tendency of governments to reduce explicit barriers to capital flows, technological
breakthroughs have made it easier for individual investors to invest abroad. Coeurdacier and Rey (2012) however also find that the home-bias is still well above 50% on average for developed countries. Sorenson et al. (2007) confirm these findings by studying the OECD countries over the period 1993-2003 where they find a sharp rise in international
diversification among these countries, but also find home-bias is still a widespread
phenomenon. Both Coeurdacier and Rey (2012) and Balli, Basher and Ozer-Balli (2010) find a greater increase in international equity holdings among European Union countries than the average increase among developed countries and contribute this to the high degree of the European economic integration. This might be an indication that the home bias of European countries is gradually being replaced by a ‘euro bias’ (Balli et al., 2010). Although the negative effects would be less severe than individual country home bias, because the Euro area represents a greater part of total world market capitalization then its individual countries, benefits of diversification are still foregone by investing disproportionately in equity issued in Euro-area(Balli et al. 2010). For Europe this might emphasize a successful economic
integration.
3.1 Explanations for existence of home-bias
The literature covers a wide range of research on the home-bias puzzle. Besides empirical evidence on the existence of the home-bias, there have been put forward a lot of potential explanations, which will be reviewed in this part.
3.2.1 Explicit barriers to international investment
The first and most straightforward explanation for deviations from the optimal portfolio suggested by standard portfolio theory is the existence of explicit costs and barriers for foreign investors. The effect of explicit cost and barriers to foreign investors is mostly investigated in the earlier research as there has been a tendency toward removing explicit barriers over the last decades. The most examined costs to capital flow are transaction cost and taxation (Cooper, Sercu and Vanpée, 2012).
Logically costs and barriers on foreign investment lower its benefits. As standard portfolio theory assumes no barriers to international investment, the existence of such barriers induces a deviation of the optimal portfolio. Black (1974) is the first to develop a model to find equilibrium in the international capital market which includes investment barriers across countries. In his model taxes are included as to represent all investment barriers present to
foreign investors. Solving the model leads to the conclusion that investors’ overweighing of domestic stocks is justified when investment barriers are significantly present.
Errunza and Losq (1985) focus on the explicit barriers of restrictions on foreign holdings. The limited amount of domestic stocks available to foreigners would raise the price for foreigners above the price that is paid by domestic investors. All these explicit costs decrease real return of foreign investment. In the presence of such cost, the required nominal return on foreign assets to be an efficient diversification investment exceeds actual return and holdings are biased towards domestic equity.
3.2.2 Information asymmetries
Another explanation put forward in the literature on home bias is information asymmetries between domestic and foreign investors. A source of such information asymmetries of investors is familiarity, referring to geographical or professional ties of investors with certain firms (Massa and Simonov, 2006). Coval and Moskowitz (1999) show that investors indeed heavily invest in stocks to which the investor is geographically close and attribute this to strong ties and information advantages due to easy access. The study shows that investment also tends to be concentrated locally within countries and does not stop at a domestic level. The ‘local-bias’ is attributed to superior access to information and strong ties with local companies.
The explanation of information asymmetries leading to such large biases seems a suitable explanation for the past. However, just like explicit barriers to international
investment has been gradually reduced over the last decades, information has become more globally accessible. Van Nieuwenburgh and Veldkamp (2009) state that information has indeed became globally accessible, they however argue that the information bias is still
present, due to the fact that investors choose which information to learn. Magi (2008) calls the amount of information available to investors nowadays even ‘unmanageable’ and states that it is impossible for investors to evaluate all available information to make the optimal
investment decisions. That is why investors choose what information to learn and in that process they seem to prefer to learn about known companies. Massa and Simonov (2006) acknowledge the information picking and in addition state that picking familiar stocks is optimal to investors as a cheap information source is used.
3.2.3. Hedging motives
Lewis (1999) identified three main hedging motives in the literature. At first hedges against domestic inflation, secondly hedges against wealth that is not traded in capital markets and thirdly hedges of diversification benefits by investing in companies that have significant operations overseas, also called ‘domestic multinationals’. Hedging against domestic inflation as well as against exchange rate risk can be explained by the invalidity of the purchasing power parity (PPP). The purchasing power parity is one of the assumptions underlying the CAPM and states that the exchange rate between two countries’ currencies will equal the ratios of their domestic price levels (Krugman, Obstfeld and Melitz, 2012). If PPP holds this would mean exchange rate would equalize real returns of domestic and foreign investors, but in empirical evidence shows PPP rarely holds except for occasions in the long run. Because PPP rarely holds real returns of domestic and foreign investors differ and portfolio
composition should also differ (Adler and Dumas, 1983).
With regard to hedging against wealth that is not traded in capital markets, Lewis (1999) highlights that in standard portfolio theory it is implicitly assumed that all wealth is liquid and tradable. As this is not the case in the real world it might be an explanation for the observed deviation towards home equity. No consensus is reached in the literature that
investment in home equity provides a better hedge for non-tradable wealth than foreign equity does. Massa and Simonov (2006), in addition, conclude investors tend to do exactly the opposite of hedging non-tradable wealth as they show investors tend to invest in stocks closely related to any non-financial income. This might even indicate that home-bias worsens the hedging of non-financial income of investors. The large observed home-bias cannot be justified by hedging motives regarding non-financial income.
The third hedging motive states that investors can domestically hedge against diversification benefits. Investors have a strong preference for domestic stocks, but when looking at what domestic stocks they pick it can be shown they also overweigh large multinational firms over real domestic firms (Cai and Warncok, 2006). Although these so-called ‘domestic multinationals’ may be headquartered in a certain country, they do give international diversification benefits due to international operations carried out all over the world. The true international diversification benefits however can only be obtained when such ‘domestic multinationals’ are in fact present. The study of Cai and Warnock (2006) is one based on data of US investors, a country where such domestic multinationals are present on a
large scale. For countries where less multinationals are present, a large home-bias cannot be justified by this hedging motive.
3.2.4. Behavioral Biases
A fourth explanation addresses behavioral biases among investors. Strong and Xu (2003) investigate the relative optimism of fund managers of different countries about their home market performance over a five-year period. In the results of the research it is found that fund managers are more optimistic regarding their home market than towards foreign markets and that correlations between their sentiment and their actions in different markets are
significant and positive. By using a survey on proudness to be a national citizen Morse and Shive (2011) show that more patriotic countries invest more heavily in domestic stock. This phenomenon was already known among consumption behavior as consumers prefer domestic brands over foreign brands, but this research shows it can also be applied to equity holdings. Cultural differences between countries also seem to play a role in international investment behavior. According to research by Beugelsdijk and Frijns (2010) societies that are more uncertainty avoiding, invest more in domestic stock where societies where more
individualistic behavior is seen, tend to invest more abroad. It is concluded not only geographic distance, but also cultural distance between countries is a determinant of investment of one country in another.
4. Methodology and data
In this section the methodology for testing the effect of European economic integration will be examined first. Secondly the data used will be described.
4.1 Methodology
To investigate the effect of the formation of the European union and the European monetary Union on international diversification and the home-bias among European investors, the view that the variables of the gravity model of trade also are determinants of international capital flow is followed. This view is consistent with findings of Portes and Rey (2005), who conclude that a very similar gravity model can be just as accurately used
predicting the flow of capital as the gravity model for goods can be used for predictions on the flow of goods. Although the gravity is not able to predict exact flows of capital, it can provide variables that have a significant influence on cross-border holdings.
Portes and Rey (2005) identify four broad determinants of cross-border asset holdings:
Economic size, distance, information variables and technology variables. Variables added in the regression are consistent with these four broad categories. Economic size is included via logarithm of GDP of the domestic country and logarithm of GDP of the foreign country. Although it is argued market capitalization might be a better indicator for economic size, when cross-border equity holdings are concerned, Balli et al. (2010) find no significant differences in results using either of the two measures. Distance is included using the logarithm of distance between the two countries which is measured using latitude and longitudes of the most important cities in terms of population of the two countries.
Information variables would include variables that create information asymmetries between domestic and foreign investors and technology variables are referred to as variables indicating the general development of financial markets. Exact variables falling into these categories are hard to identify and quantify, but in order to capture the effects of these variable categories the following two dummy variables are added. At first, there is a consensus in the literature that distance creates information asymmetry, therefore the included distance variable already captures some information effect. A dummy for common language is added to further capture information effect, as the sharing of a language makes it easier to evaluate obtained
information. To capture the effect of technological development a dummy variable for contiguousness is added, as it assumes that contiguous countries have similar developed financial markets. Although the two included dummies are very broad and might not capture the whole information and technology effect, the fact that they are easily observable makes them the best variables to use to capture at least part of the effects.
Consistent with the method used by Balta and Delgado (2008), two dummies are added to measure the effects of European economic integration. The first dummy is one on intra-EU holdings, which equals one when both countries are members of the European union and zero otherwise. The second dummy is a dummy on intra-EMU holdings, that equals one when both countries have adopted the euro as their official currency and zero otherwise. The dependent variable in the regression is the logarithm of cross-border equity holdings of domestic investors in the foreign country. The ultimate regression estimated is:
𝐿𝐿𝐿𝐿𝐿𝐿 (𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑓𝑓𝑒𝑒𝑒𝑒𝑓𝑓𝑒𝑒𝑒𝑒 ℎ𝑓𝑓𝑜𝑜𝑜𝑜𝑓𝑓𝑓𝑓𝑓𝑓𝑜𝑜) =
𝛼𝛼 + 𝛽𝛽1 LOG 𝐿𝐿𝐺𝐺𝐺𝐺 (𝐺𝐺𝑓𝑓𝐷𝐷𝑓𝑓𝑜𝑜𝑒𝑒𝑓𝑓𝐷𝐷) + 𝛽𝛽2 LOG 𝐿𝐿𝐺𝐺𝐺𝐺 (𝐹𝐹𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓) + 𝛽𝛽3 𝐿𝐿𝐿𝐿𝐿𝐿 𝑜𝑜𝑓𝑓𝑜𝑜𝑒𝑒𝑑𝑑𝑓𝑓𝐷𝐷𝑓𝑓 + 𝛽𝛽4 𝐺𝐺𝑒𝑒𝐷𝐷𝐷𝐷𝑒𝑒 𝐷𝐷𝑓𝑓𝐷𝐷𝐷𝐷𝑓𝑓𝑓𝑓 𝑜𝑜𝑑𝑑𝑓𝑓𝑓𝑓𝑒𝑒𝑑𝑑𝑓𝑓𝑓𝑓 + 𝛽𝛽5 𝐺𝐺𝑒𝑒𝐷𝐷𝐷𝐷𝑒𝑒 𝐷𝐷𝑓𝑓𝐷𝐷𝐷𝐷𝑓𝑓𝑓𝑓 𝑏𝑏𝑓𝑓𝑓𝑓𝑜𝑜𝑓𝑓𝑓𝑓 +
𝛽𝛽6 𝐺𝐺𝑒𝑒𝐷𝐷𝐷𝐷𝑒𝑒 𝐼𝐼𝑓𝑓𝑒𝑒𝑑𝑑𝐼𝐼𝐼𝐼 + 𝛽𝛽7 𝐺𝐺𝑒𝑒𝐷𝐷𝐷𝐷𝑒𝑒 𝑓𝑓𝑓𝑓𝑒𝑒𝑓𝑓𝑑𝑑𝐼𝐼𝑖𝑖𝐼𝐼 + ɛ
The regression is estimated using panel data of a sample of 21 countries from all over the world covering the period 2001-2008. The following countries are included in the sample: Australia, Austria, Belgium, Brazil, Canada, Cyprus, Czech Republic, Denmark, France, Germany, Hungary, Iceland, Italy, Japan, the Netherlands, Norway, Slovak Republic, Spain, the United Kingdom, the United States and Switzerland. Four regressions are performed using different number of panels and observations. The first two regressions cover data of all countries in the sample, they however differ in the way panels are included. As some data on cross-border equity holdings is not available, not all panels have complete data. Sometimes equity-holdings from one country in another particular country are fully missing, in which case the panel is omitted from the regression. In most cases of incomplete data however, only data of cross-border holdings in particular years is missing. With this missing data is dealt in two different ways, using two regressions. The first regression omits all panels that do not have complete data and so only includes panels that have complete data on cross-border equity holdings from 2001 to 2008. This means that some data that is available is omitted as well, but it leads to a balanced regression as there are no gaps in the panels. Using this method for including panels, the regression includes 364 panels and 2912 observations. In the second regression all available data is included. This means now 407 panels and 3032 observations are used. This way more observations are used, but the actual regression is unbalanced as not all included panels cover the full period 2001-2008.
As there are four countries (Cyprus, Czech republic, Hungary and Slovak republic) included in the sample of 21 countries that joined the EU during the sample period, two additional regressions are performed in order capture the effect of joining the EU on cross-border equity holdings. The dummy for same language has been omitted in these two
regressions as the included countries do not have a language that is shared by any of the other countries in the sample. The dummy on intra-EMU holdings is omitted as only Cyprus
adopted the Euro during the sample period and in Cyprus the euro was only adopted in the last year of the sample period. The regressions are performed in the same way as the regression covering all data, one including only complete panels and one including all available data. The regression using only fully complete panels includes 65 panels and 499 observations and regression using all available data includes 77 panels 557 observations.
4.2 Data
Data on cross-border equity holdings is difficult to obtain due to the fact that individual investor holdings are difficult to track down. In the literature on home-bias two
main sources are used: mutual funds holdings and the coordinated portfolio investment survey(CPIS). The relative share of total investment in a country invested by mutual funds will differ across countries which makes it hard to accurately compare the data. The CPIS is a voluntary survey conducted by the IMF in which between 67 and 76 countries participated between 2001 and 2008. The CPIS provides guidelines for collecting the data, but leaves the decision of collection method up to the countries, such that they can choose a method fitting their investment climate. Consistent with earlier research on home-bias, data on cross-border equity holdings is obtained from the CPIS.
Data on GDP of the included countries is obtained from the world bank, which has collected data on country GDP since 1960 and measures GDP in American dollars. Data on distance, common language and contiguousness are obtained from the CEPII. CEPII includes variables which were identified in a gravity model for trade flows by Mayer and zignago (2005). Distance is measured in four different ways by CEPII. The distance measure used is the simple measure using latitudes and longitudes of the most important cities in terms of population of the two countries included in the panel. Also different dummies for ‘common language’ are identified on the bases of the percentage of people in a country speaking a certain language. In this regression the dummy on common language equals one if a language is spoken by at least 20% of the population in both countries. The dummy for contiguousness is easily observable and non-arbitrary.
5. Results
In this part the results of the performed regression analysis will be reviewed. First the results will be analyzed. Secondly the regression method and the implications of the
regression output will be reviewed.
5.1 Analysis
To test the effect of the formation of the European union and the European monetary union on cross-border equity holdings, four regressions have been performed. In all four regressions I have controlled for the effect of economic size of the countries, distance between the countries and the effect of sharing the same border on cross-border equity holdings. In the first two regressions there is also controlled for the effect of a common language among countries on cross-border equity holdings, leading to the following estimated model:
𝐿𝐿𝐿𝐿𝐿𝐿 𝐹𝐹𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑓𝑓𝑒𝑒𝑒𝑒𝑓𝑓𝑒𝑒𝑒𝑒 ℎ𝑓𝑓𝑜𝑜𝑜𝑜𝑓𝑓𝑓𝑓𝑓𝑓𝑜𝑜 = 𝛼𝛼 + 𝛽𝛽1 𝐿𝐿𝐿𝐿𝐿𝐿 𝐺𝐺𝑓𝑓𝐷𝐷𝑓𝑓𝑜𝑜𝑒𝑒𝑓𝑓𝐷𝐷 𝐿𝐿𝐺𝐺𝐺𝐺 + 𝛽𝛽2 𝐿𝐿𝐿𝐿𝐿𝐿 𝐹𝐹𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝐿𝐿𝐺𝐺𝐺𝐺 + 𝛽𝛽3 𝐿𝐿𝐿𝐿𝐿𝐿 𝐺𝐺𝑓𝑓𝑜𝑜𝑒𝑒𝑑𝑑𝑓𝑓𝐷𝐷𝑓𝑓 + 𝛽𝛽4 𝐺𝐺𝑒𝑒𝐷𝐷𝐷𝐷𝑒𝑒 𝐷𝐷𝑓𝑓𝐷𝐷𝐷𝐷𝑓𝑓𝑓𝑓 𝑜𝑜𝑑𝑑𝑓𝑓𝑓𝑓𝑒𝑒𝑑𝑑𝑓𝑓𝑓𝑓 + 𝛽𝛽5 𝐺𝐺𝑒𝑒𝐷𝐷𝐷𝐷𝑒𝑒 𝐷𝐷𝑓𝑓𝐷𝐷𝐷𝐷𝑓𝑓𝑓𝑓 𝑏𝑏𝑓𝑓𝑓𝑓𝑜𝑜𝑓𝑓𝑓𝑓 + 𝛽𝛽6 𝐺𝐺𝑒𝑒𝐷𝐷𝐷𝐷𝑒𝑒 𝑓𝑓𝑓𝑓𝑒𝑒𝑓𝑓𝑑𝑑𝐼𝐼𝐼𝐼 + 𝛽𝛽7 𝐺𝐺𝑒𝑒𝐷𝐷𝐷𝐷𝑒𝑒 𝑓𝑓𝑓𝑓𝑒𝑒𝑓𝑓𝑑𝑑𝐼𝐼𝑖𝑖𝐼𝐼 + ɛ
To obtain meaningful estimates logarithms have been used for cross-border equity holdings, GDP of both countries and distance. Because there has been used a logarithm for cross-border holdings, all estimated effects refer to percentage changes in cross-border equity holdings. As logarithms have been taken for GDP and distance as well, estimated effect for these variable refer to the percentage change in cross-border equity holdings given a one percent change in GDP or distance.
The results of the first two regressions have been summarized in the following table:
(1) Panels: 360 Observations: 2797 Overall R²: 0.6627 (2) Panels: 407 Observations: 3032 Overall R²: 0.6467 Constant -52.71148** (1.55472) -53.06098** (1.593204) LOG GDP Home-country 1.167959** (0.0477326) 1.100657** (0.0495491) LOG GDP Foreign country 1.158113**
(0.0501128) 1.221283** (0.0504587) LOG distance -0.5693339** (0.1012192) -0.5299217** (0.1034089) Common language 1.76503** (0.3730162) 1.817007** (0.3912275) Common border -0.1388696 (0.4148657) -0.0450822 (0.4373038) EU-membership 0.3728554** (0.0866189) 0.29134** (0.0895537) 15
Euro 0.0957644 (0.2155512)
0.2213037 (0.233333)
Significant indication: * statistically significant at 5% and ** statistically significant at 1% level. Numbers between parenthesis indicate standard deviations of the estimates.
The first row shows the number of panels and the number of observations used for estimating the regression model and in addition shows the overall R² of the performed regressions. The subsequent rows show the estimated effects of the seven variables used. As predicted by the literature, the estimated coefficients for economic size of both countries and common language are positive and the coefficient of distance is negative. These coefficients are significant at a 1% level in both regressions. The coefficient of sharing a common border is negative, but insignificant. Looking at the effect of European economic integration on cross-border equity holdings, two conclusions can be drawn from the performed regressions. Firstly EU-membership has a positive effect on cross-border equity holdings. Although the estimated effects differ between the two performed regressions, the effect in both regressions is positive and significant at a 1% level. Secondly, no significant effect has been found in any of the two performed regressions regarding the effect of having the Euro on cross-border equity
holdings.
The last two regressions have been performed only using data of the four countries in the sample that joined the EU in 2004: Cyprus, Czech republic, Hungary and Slovak
Republic. The estimated regression model differs from the original model as the dummy for common language and the dummy for intra-EMU holdings have been omitted. The dummy for common language has been omitted as no language spoken in any of the four countries is spoken in any other country included in the sample. The dummy intra-EMU holdings is excluded because Cyprus is the only country that adopted the Euro as their official currency during the sample period and only did this in the last year of the sample period, which means there are too little observations to estimate the effect of introducing the Euro. The estimated regression model for the last two regression therefore is as follows:
𝐿𝐿𝐿𝐿𝐿𝐿 𝐹𝐹𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑓𝑓𝑒𝑒𝑒𝑒𝑓𝑓𝑒𝑒𝑒𝑒 ℎ𝑓𝑓𝑜𝑜𝑜𝑜𝑓𝑓𝑓𝑓𝑓𝑓𝑜𝑜 = 𝛼𝛼 + 𝛽𝛽1 𝐿𝐿𝐿𝐿𝐿𝐿 𝐺𝐺𝑓𝑓𝐷𝐷𝑓𝑓𝑜𝑜𝑒𝑒𝑓𝑓𝐷𝐷 𝐿𝐿𝐺𝐺𝐺𝐺 + 𝛽𝛽2 𝐿𝐿𝐿𝐿𝐿𝐿 𝐹𝐹𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝐿𝐿𝐺𝐺𝐺𝐺 + 𝛽𝛽3 𝐿𝐿𝐿𝐿𝐿𝐿 𝐺𝐺𝑓𝑓𝑜𝑜𝑒𝑒𝑑𝑑𝑓𝑓𝐷𝐷𝑓𝑓 + 𝛽𝛽4 𝐺𝐺𝑒𝑒𝐷𝐷𝐷𝐷𝑒𝑒 𝐷𝐷𝑓𝑓𝐷𝐷𝐷𝐷𝑓𝑓𝑓𝑓 𝑏𝑏𝑓𝑓𝑓𝑓𝑜𝑜𝑓𝑓𝑓𝑓 + 𝛽𝛽5 𝐺𝐺𝑒𝑒𝐷𝐷𝐷𝐷𝑒𝑒 𝑓𝑓𝑓𝑓𝑒𝑒𝑓𝑓𝑑𝑑𝐼𝐼𝐼𝐼 + ɛ
The results of the these two regressions are summarized in the following table:
(3) Panels: 65 Observations: 499 Overall R²: 0.4419 (4) Panels: 77 Observations: 557 Overall R²: 0.3880 Constant -58.38428** (4.521867) -52.98548** (4.808026) LOG GDP home-country 1.366605** (0.1716352) 1.216167** (0.1758893) LOG GDP Foreign country 1.095491**
(0.1654307) 1.038016** (0.1372473) LOG Distance -0.5184293* (0.2246773) -0.5289857* (0.2371831) Common border 2.190871** (0.8543545) 1.462067 (0.8966978) EU-membership 0.4955192** (0.1592554) 0.5230149** (0.1770774)
Significant indication: * statistically significant at 5% and ** statistically significant at 1% level. Numbers between parenthesis indicate standard deviations of the estimates.
Looking at the results of the last two regressions, similar conclusions can be drawn by analyzing estimated effects, as has been drawn from the estimated effects in the first two regressions. Again economic size of both countries has a positive effect on cross-border equity holdings, which is significant at a 1% level in both regressions. Distance again shows a negative effect, which is significant at a 5% level. For the effect of sharing a common border, an ambiguous effect has been found by using these two regressions. The regression using only complete panels finds a positive effect that is significant at a 1% level. The regression using all available data however finds no significant effect. More detailed research is needed to really evaluate the effect of sharing a common border. The effect of European economic integration is again consistent with the findings in the first two regressions. In both regressions it is found that EU-membership has a positive effect on cross-border equity holdings, which is significant at a 1% level. The effect seems somewhat larger than the effect estimated in the first two regressions. This might indicate a boost in international
diversification to countries that join the European Union, but the small sample size might
have biased results. Using this sample, it is not possible to estimate the effect of introduction of the Euro.
5.2 Discussion
There are four regressions used to evaluate the effect of the European economic
integration on cross-border equity holdings. The first two evaluate the effect using a sample of 21 countries of all over the world. The second two regressions estimate the effect of joining the EU on cross-border equity holdings of investors of the joining countries using a sample of only four countries; Cyprus, Czech Republic, Hungary and Slovak republic. Using any of the four samples, all effects except the effect of sharing the same border show an effect predicted by the literature. For economic size and sharing the same language, this yields a positive effect; for distance, this is a negative effect. Sharing a common border shows, in contrast with what the literature would predict, a negative effect in the last two regressions and also shows an insignificant effect in three of the four performed regressions. Looking at the effect of European economic integration on cross-border equity holdings, it can be concluded that investors from European union members hold on average 37% more equity in other European member countries than in non-European union member countries according to the regression using only complete panels. When using the panel with all available data, this estimate is 29%, which although somewhat lower, still indicates a significant positive effect of European union membership. Looking at the effect on investor behavior in joining countries, the effect of EU membership seems even larger. Using any of the two methods, investors in these countries hold on average close to 50% more equity in other European member countries then in other foreign countries. These findings together with the observed decrease in equity home-bias, show a significant contribution of removal of economic barriers to the decrease in home-bias. Possible explanations for the higher observed effect in joining countries, is that the four countries used in this sample are relatively poor European countries, for which the relative benefits of the removal of economic barriers to international capital flow are higher than for some richer European countries. In contrast with what was expected, further economic integration, by the introduction of the Euro as a single currency does not show a significant effect in any of the two regressions in which the effect is estimated.
Although the estimates of the effect of European union membership in the second set of regressions is higher than in the first two regressions, these regressions also show a lower overall R², indicating lower explanatory power. This might be partially due to the low number
of sample countries used in these regressions. The use of only four countries in the sample leads to less accurate estimation results. The reason only four out of the ten countries that joined the EU in 2004 are used in the sample, is the lack of data on cross-border equity holdings in these other countries. The observed R² in the last two regressions, indicate that less than 50% of the observed variation is explained by included variables. The overall R² of close to 0,7 in the first two regressions, show a quiet strong explanatory power of these regression results, despite the omitted variables affecting cross-border equity holdings because of difficulties in quantification.
Two possible sources of bias in the estimated effects need to be taken into account when evaluating estimated results. Firstly there is the possibility of inaccurate estimates, because of the use of inaccurate data. Data on cross-border equity holdings is not widely available. A lot of countries do not keep track of cross-border equity holdings at all. Countries that do keep track of cross-border equity holdings, often use different methods and experience difficulties in the tracking of exact individual investor holdings, therefore reported cross-border equity holdings might deviate from real cross-cross-border holdings. The CPIS is
constructed to produce as accurate data on cross-border holdings as possible, but reported holdings can still deviate from real holdings. A second possible source of bias in the estimates, is omitted variable bias. There have been difficulties in exact identification of variables affecting cross-border equity holdings. In addition some identified variables in the literature are difficult to quantify and therefore difficult to include fully. Variables are included to represent variables affecting cross-border equity holdings as good as possible.
5. Conclusion
Theory tells that internationally diversified portfolios are optimal to hold for investors. Empirical evidence in contrast shows investors all over the world prefer to hold portfolios highly biased toward home-securities. Looking at equity holdings, investors also tend to invest disproportionally more in domestic equity, leading to a paradox known in finance as the equity home-bias. The most likely explanation for this observed home-bias is the
existence of significant barriers to international capital flow, which decrease expected benefits from international diversification leading to more domestically orientated portfolios. Even when acknowledging the fact that explicit barriers to international capital flow make full international diversification suboptimal, the overall observed equity home-bias seems far too high to be justified just by explicit barriers to trade. The question is to what extent the
observed equity home-bias is created by explicit barriers to international capital flow and to
what extent it must be attributed to other implicit barriers. To examine this question, the effect of removal of economic barriers by the formation of the European union an European
monetary union on cross-border equity holdings of European investors have been
investigated, as the European integration may be seen as the greatest effort to remove explicit barriers to trade and capital flow seen over the last decade.
Results of the performed regressions over the whole sample data show that cross-border equity holdings are on average between 29% and 37% higher among European Union members. When looking at a sample of countries that joined the EU within the sample period the effect of EU membership seems somewhat larger, between 49% and 52%. The higher effect on investors might be due to the fact that the countries included as joining countries, are relatively poor European countries, making the relative benefits of the removal of economic barriers to international capital flow higher for investors in these countries. The sample however only consists of four countries due to lack of data on other joining countries, so the estimated effects might not be fully accurate due to the limited number of observations. Estimates of the effect of formation of the European union on cross-border equity holdings of European investors are positive and significant in all performed regressions. Looking at the effect of introduction of the Euro as a single currency, in contrast to what would be expected, no significant effect has been found on cross-border equity holdings.
The estimated effects of European integration on cross-border equity holdings might be biased due to two factors. At first inaccurate data might bias results. Data on cross-border equity holdings is scarce. At first there are a lot of countries that do not record cross-border equity holdings at all. Secondly difficulties in tracking down individual investor holdings, make available data not fully accurate. A second possible factor of bias in estimated effects is omitted variable bias. Some variables affecting cross-border equity investments are difficult to identify or quantify, such that it is difficult to include them in the regression. Variables have been included to cover effects identified in the literature, but these variables are likely to cover only part of the effects. In order to get fully accurate estimates it is therefore needed to identify and quantity specific effects further.
To minimize these sources of bias, one of the most reliable sources of data on cross-border equity holdings conducted by the IMF has been used and variables are included such that at least part of variables affecting cross-border holdings identified in the literature is covered. Therefore two broad conclusions can be drawn from the estimated regression models. Firstly, EU-membership has a positive effect on cross-border equity holdings, indicating that the elimination of explicit economic barriers within the European Union has
contributed significantly to the decreased home bias among European investors. Secondly, no significant effect of the introduction of the Euro on cross-border equity holdings is found.
6. Bibliography
Adler, M. and Dumas, B. 1983. International portfolio choice and corporation finance: A synthesis. The Journal of Finance, 38 (3): 925-984.
Anderson, J. E. 2011. The gravity model. The Annual Review of Economics, 3: 133-160. Balli, F., Basher, S. A. and Ozer-Balli, H. 2010. From home bias to Euro bias: Disentangling
the effects of monetary union on the European financial markets. Journal of Economics and Business, 62: 347-366.
Balta, N. and Delgado, J. 2009. Home bias and market integration in the EU. CESifo Economic Studies, 55 (1): 110-144.
Bekaert, G., Harvey, C.R., Lundblad, C.T. and Siegel, S. 2013. The European union, the Euro, and equity market integration. Journal of Financial Economics, 109: 583-603.
Berk, J. and DeMarzo, P. 2011. Corporate Finance. Harlow: Pearson Education Limited 2011.
Beugelsdijk, S. and Frijns, B. 2010. A cultural explanation of the foreign bias in international asset allocation. Journal of Banking and Finance, 34: 2121-2131.
Black, F. 1974. International capital market equilibrium with investment barriers. Journal of Financial Economics, 1: 337-352.
Bodie, Z., Kane, A. and Marcus, A. J. 2011. Investment and portfolio management. New York: The McGraw Hill Companies 2011.
Cai, F. and Warnock, F.E. 2006. International diversification at home and abroad. Working paper (National Bureau of Economic Research).
CEPII. GeoDist. Retrieved from: http://www.cepii.fr/%5C/anglaisgraph/bdd/distances.htm Coeurdacier, N. and Martin, P. 2009. The geography of asset trade and the euro: Insiders and
outsiders. Journal of Japanese and International Economies, 23: 90-113.
Coeurdacier, N. and Rey, H. 2012. Home bias in open economy financial macroeconomics. Journal of Economic Literature, 51 (1): 63-115.
Cooper, I., Sercu, P. and Vanpée, R. 2012. The equity home bias puzzle: A survey. Foundation and Trends in Finance, 7 (4): 289-416.
Coval, J.D. and Moskowitz, T.J. 1999. Home bias at home: Local equity preference in domestic portfolios. The Journal of Finance, 54 (6): 2045-2073.
Duenas, M. and Fagiolo, G. 2013. Modeling the international-trade network: a gravity approach. Journal of Economic Interaction and Coordination, 8 (1): 155-178. Errunza, V. and Losq, E. 1985. International asset pricing under mild segmentation: Theory
and test. The Journal of Finance, 40 (1): 105-124.
French, K.R. and Poterba, J.M. 1991. Investor diversification and international equity markets. The American Economic Review, 81 (2): 222-226.
IMF. Coordinated Portfolio Investment Survey. Retrieved from: http://elibrary-data.imf.org/public/FrameReport.aspx?v=3&c=26593546
Krugman, P.R., Obstfeld, M. and Melitz, M.J. 2012. International economics. Harlow: Pearson education limited 2012.
Lau, S.T., Ng, L. and Zhang, B. 2010. The world price of home bias. Journal of Financial Economics, 97: 191-217.
Lewis, K.K. 1999. Trying to explain home bias in equities and consumption. Journal of Economic Literature, 37 (2): 571-608.
Magi, A. 2008. Portfolio choice, behavioral preferences and equity home bias. The Quarterly Review of Economics and Finance, 49: 501-520.
Markowitz, H. 1952. Portfolio selection. The Journal of Finance, 7 (1): 77-91.
Massa, M. and Simonov, A. 2006. Hedging, familiarity and portfolio choice. The Review of Financial Studies, 19 (2): 633-685.
Mayer, T. and Zignago, S. 2005. Market access in global and regional trade. Working paper (CEPII).
Morse, A. and Shive, S. 2011. Patriotism in your portfolio. Journal of Financial Markets, 14: 411-440.
Portes, R. and Rey, H. 2005. The determinants of cross-border equity flows. Journal of International Economics, 65: 269-296.
Sharpe, W.F. 1964. Capital asset prices: A theory of market equilibrium under conditions of risk. The Journal of Finance, 19 (3): 425-442.
Sørensen, B.E., Wu, Y.T., Yosha, O. and Zhu, Y. 2007. Home bias and international risk sharing: Twin puzzles separated at birth. Journal of International Money and Finance, 26 (4): 587-605.
Strong, N. and Xu, X. 2003. Understanding the equity home bias: evidence from survey data. The Review of Economics and Statistics, 85 (2): 307-312.
Stulz, R.M. 1999. Globalization of equity markets and the cost of capital. Working paper (National Bureau of Economic Research).
Van Nieuwenburgh, S. and Veldkamp, L. 2009. Information immobility and the home bias puzzle. The Journal of Finance, 64 (3): 1187-1215.
World bank. GDP (Current US Dollar). Retrieved from:
http://data.worldbank.org/indicator/NY.GDP.MKTP.CD
