Equity Home Bias In The Netherlands

Equity Home Bias In The Netherlands

by

Melle Pebesma

University of Groningen

Faculty of Economics and Business

MSc Finance

June 26

, 2014

Supervisor: drs. A. Plantinga

melle_pebesma@hotmail.com

student number: 1775383

Keywords: equity, home bias, the Netherlands, portfolio management, diversification.

JEL Classification: G11, G15, F21

Abstract:

1. INTRODUCTION

Not short after the Great Financial Crash of 2008, the Obama Administration believed that the best way to repair the financial system was to improve the performance and oversight of global banks and investment firms. However, a growing number of Americans prefers to pull out their retirement savings and invest them locally (Shuman, 2009). He suggests that the publicly traded institutions are no longer considered to be safe anymore, and in this way the local economy can be given a boost.

To some extent, this response of the public is very understandable. However, portfolio theory (Markowitz, 1952 and Sharpe, 1964) suggests that an investor should only care about expected returns and variance in their efficient portfolio. The correlation between these assets in the portfolio will lead to a lower variance. Risk averse investors should then create a diversified portfolio that maximizes expected return given the risk.

In the last four decades, this aspect of diversification is taken from a national level to an international level. Solnik (1974) has shown that investing in foreign indices reduces the volatility of U.S. market portfolios, with gains attributed to low return correlations between national equity indices. This implies that due to correlation between national economies, an investor should create a portfolio which consists of assets from the entire globe, in order to fully capture the diversification effects and thus reducing risk optimally.

This has however not been the case. Investors have a tendency to invest in familiar stocks. French & Poterba (1991) show that investors tend to invest in stocks of companies in their own country. In a more recent study, Seasholes and Zhu (2010) even prove this to be true at a local or regional level. If international diversification is an efficient way to lower the risk of a portfolio, then why is this not happening? French and Poterba (1991) show that investors expect domestic returns to be systematically higher than for a diversified portfolio. This, of course, cannot be true. The explanations for this under-diversification vary between institutional and behavioral factors, which I will address in the literature review.

suboptimal. Most authors agree that investing in local stocks is more risky and therefore costly. However, there are a few authors who argue that this isn’t true. Errunza et al. (1998) examines whether portfolios of domestically traded securities can mimic foreign indices so that trading abroad is not necessary. The results show strong evidence that for the U.S., the need to hold assets that trade only abroad to obtain international diversification is disappearing. Antoniou et al. (2010) perform an out of sample test for UK equity, and find the same results. Both papers thus suggest that home-made diversification makes it unnecessary to diversify internationally.

This controversy on the benefits of diversification warrants another look at the possibility of home-made diversification overcoming the need to buy foreign traded assets. Errunza et al (1998) investigates these issues in the USA, a country with probably the most diversified stock market in the world. The second study by Antoniou (2010) shows findings of the entire UK equity market. If homemade diversification is possible in a small country like the Netherlands, one could argue that the home-diversification puzzle doesn’t even exist. That is why this paper will examine if there is still a need for Dutch investors to diversify their portfolios internationally, or if home-made diversification can mimic the same results as foreign equity. Also, the most recent paper of Antiniou et al. (2010) only covers data up until 2003, so this might have changed over the years. What makes that even more interesting is the fact that Solnik (2001) shows that in bull markets, correlations tend to go up. In light of the recent crisis, diversification might not work at all in that case.

2. LITERATURE OVERVIEW

Traditional models of portfolio theory suggest that expected return and variance are the only factors an investor should care about (Markowitz, 1952 and Sharpe, 1964). One should maximize expected return with a constant variance or minimize risk with a constant return. Correlation between stocks can, in their theories, lower the risk of one’s portfolio. This implies that investors should seek to diversify as much as possible, in order to fully capture the gains of diversification. Therefore, Levy and Sarnat (1970) suggest that correlation between national economies could lead to further risk reduction if an investor also picks foreign assets in their portfolio.

Due to financial liberation, deregulation and relaxation of restrictions on foreign ownership of domestic equity in many countries, the equity market opportunities have increased in recent years. This provides excellent opportunities for international diversification (Antoniou et al. 2010). Although this wasn’t the case a couple of decades ago, there were enough opportunities to invest abroad. However, in spite of the theoretical superiority of international diversification, French and Poterba (1991) prove that investors still have the tendency to invest in stocks of companies in their own country, and this is again shown in a study by Huberman (2001). In recent studies, Seasholes and Zhu (2010) show that investors even pick stocks that are close to their own community. There does seem to be a shift towards more international diversification, but it remains significantly lower than researchers are suggesting (Berrill & Kearney, 2010).

There have been many studies trying to prove this home-bias, with differences in institutional factors and behavioral factors.

Institutional factors

Also, asymmetric information is given as an explanation for the bias (Fellner and Maciejovsky, 2003). It is often researched whether domestic investors have an information advantage, which could lead to higher returns. However, this informational advantage is not an argument that could rationalize the home bias (Seasholes & Zhu, 2010). In their study, there is no evidence of outperformance on investing in stocks of locally headquartered firms.

Behavioral factors

Being amongst the first to recognize the behavioral aspects, French & Poterba (1991) argue that different expectations and overconfidence in own markets is a very important factor for the home-bias. They argue that investors are more optimistic in their own market than in the foreign markets, which leads to a portfolio consisting mostly out of domestic stocks. Next to that, Huberman (2001) claims that familiarity breeds investment: people feel comfortable investing money in a business that is visible to them. Also important are cultural factors, local media exposure and community effects: all these factors tend to drive investors towards investing in the domestic market.

Many authors agree on the fact that the home bias is more a result of behavioral factors than it is of institutional factors (French & Poterba, 1991; Fellner & Maciejovsky, 2003). There is however still no consensus over these subjects. They do not necessarily have to be mutually exclusive. There is however one thing that the authors agreed on: lack of international diversification is a suboptimal choice. This has been generally accepted until the late 1990s. Errunza. et al. (1998) in their study were the first to challenge this. They argue that international diversification could be done without trading abroad. In their study, covering the U.S. equity market, they show the first results that deny the general consensus of the home-bias being suboptimal. They suggest that only gains made trough home-made diversification are needed, and further trading abroad is insignificant. They are the first to conclude that an U.S. investor does not need to hold foreign assets to achieve maximum diversification effects.

It is important to understand how it is possible to mimic the foreign portfolio with domestic securities. In the studies of Errunza et al. (1999) and Antoniou et al. (2010), two different assets are used to capture the international diversification benefits: industry indices and multinational companies.

Industry indices

One of the main sources of gains from international diversification is the diversity of industrial structures (Roll, 1992). He suggests that stock markets in a country with a number of different industries are imperfectly correlated to each other, since the firms within one industry are on average highly correlated and therefore industries (or different firms) are not. Moreover, Ferreira and Gama (2005) show that industry diversification in both the global and the local level is highly effective. However, Heston and Rouwenhorst (1994) argue the effectiveness of industry benefits in diversification. They find that the industry explains less than one percent of the differences in volatility.

The conflicting evidence led to more research opportunities in the field of industry benefits. Serra (2000) argues that the variance in the emerging markets is mainly explained by country factors, not by industrial factors. This is also true for developed markets, as Griffin and Karolyi (1998) argue. In their study, they take a look at to which extent international diversification is explained by the countries’ industry composition. They show that less than four percent of the volatility is explained by the industry, but also that this differs very much per industry. Internationally traded goods have larger industry effects than goods that are not traded internationally (e.g. consumer goods vs. healthcare). Both Serra (2000) and Griffin and Karolyi (1998) conclude that industrial diversification should not be ignored.

Multinational companies

Once again, there is still a debate on whether multinational companies create benefits for home-diversification. Jacquillat and Solnik (1978) show that investing in multinational companies is not an adequate replacement for international diversification, due to the fact that the systematic risk is hardly influenced by the foreign transactions. On the other hand, Fatemi (1984) finds that foreign economic activity is imperfectly correlated with domestic economic activity, which ensures that the stockholders of multinational companies have excess gains over stockholders of purely domestic firms.

In recent years, Berril and Kearney (2010) found that investors can obtain international diversification effects by investing in multinational companies. Moreover, the more countries a multinational company operates in, the greater the benefits are. An investor can become a ‘free-rider’ when investing in a multinational company, because he or she gets the benefits of internationality, but still engaging in the home-diversification puzzle (Berril and Kearney, 2010).

Summarizing, there are a lot of conflicting views on the effects of industry composition and multinational stock. It is not sure whether they actually capture the benefits of international diversification. Moreover, the studies of Errunza et al. (1999) and Antoniou et al. (2010) were performed in countries with relatively large equity markets, which means that their conclusions do not necessarily hold for smaller countries at all. However, based on their common findings, I expect that the home bias is not sub optimal for Dutch investors.

Research question

The research question, based upon the initial motive for this study is:

3. METHODOLOGY

I will use the same methodology as the Errunza et al. (1998) and Antoniou et al. (2010) papers.. According to Henry (2000), most emerging stock market liberation policies had been executed by 1992. Since the other two papers have taken data from resp. 1976-1973 and 1994-2003, I will use weekly data from 01/01/2003-23/05/2012. This is due to the fact that the Dutch local indices, the FTSE Local Netherlands Index and the Amsterdam Small Cap Index did not exist earlier on, as well as the consumer goods industry. Therefore, I can’t do an appropriate analysis with data from the 1990’s. Also, 7 out of the 9 industries only have data until 23/05/2012 instead of 2014. The data will be on a weekly basis.

Like Errunza et al. (1998) and Antoniou et al. (2010), I will have to address two issues in order to draw conclusions in this study. First of all whether it is possible to mimic foreign market indices with securities trading in the Netherlands. I will construct portfolios using: (1) a Dutch local index, the FTSE Local Netherlands Index. This includes companies whose business is mostly done in the Netherlands. Next to that, I will use the Amsterdam Small Cap Index, to capture size effects. (2) Industrial indexes. These are derived from the FTSE industry sectors. I will use the same industries as given by the Industry Classification Benchmark, except for the utilities sector. Unfortunately, it is not available on Datastream. The used industries are listed in the appendix. (3) Multinational companies’ equity. According to Root (1994), a multinational company is a parent company that (i) engages in foreign production through its affiliates located in several countries, (ii) exercises direct control over the policies of its affiliates and (iii) implements business strategies in production, marketing, finance and staffing that transcend national boundaries. Root (1994) concludes that multinational companies exhibit no loyalty to the country in which they are incorporated.

Therefore, I will use Orbis to acquire companies that have their headquarters based in the Netherlands, and also have foreign subsidiaries. Of course, they will have to be publicly listed. This leads to 10 multinational companies, as shown in table 2 in the appendix. However, Delta Lloyd data only has 4 years of history on the stock market, so I didn’t incorporate them. As a foreign benchmark, I will use the FTSE All-World Index, which is a measure of overall equity returns in 49 countries.

Table 1: Descriptive statistics dependent variable. Series: WORLD Sample 1/08/2003 5/23/2012 Observations 490 Mean 0.001 Median 0.002 Maximum 0.124 Minimum -0.138 Std. Dev. 0.023 Skewness -0.663 Kurtosis 7.718 Jarque-Bera 490.417 Probability 0.000

Before I continue with the methodology, it is wise to take a closer look at the available data. The table shows the mean, median, maximum, minimum, skewness, kurtosis and Jarque-Bera statistics for the world index. The mean is quite low, which confirms a period of crisis. This is also confirmed by the standard deviation of 2,3%. The table also presents values of the kurtosis and the skewness. A normal distribution would have a kurtosis of 3 and a skewness of 0. The kurtosis is much higher than 3, suggesting a peaked distribution or a leptokurtic distribution. This is very much in line with the crisis period we’re in, since the dataset has some extreme negative values. This can also be seen in the negative skewness, there is a tail with more negative values.

The descriptive statistics for the independent variables can be found in the appendix, due to the amount of space that it requires. As can be seen, the p-value of the Jarque-Bera statistic is always 0.000, which implies that we have to reject the null hypothesis of normality. However, since I have a large sample this does not have serious implications.

Mimicking portfolios

Firstly, I will perform regression analysis to develop the mimicking portfolios using the industry assets (Pi) and the local assets (Pl).

Since the diversity of industrial structures can be an important part of diversification effect - as shown in the literature review - I will use the described industries (see Appendix) to develop the mimicking portfolios. To do this, I will regress returns of the benchmark on the nine Dutch industry groups, in order to explain the return on the foreign index by returns on nine Dutch industries. So, for Pi:

where is the weekly return on the benchmark index in period t, α is the constant, are

the weights assigned to the industries, is the weekly return on Dutch industrial indices

and is an error term.

For the second portfolio, I will use the weekly return on two Dutch market indices, the weekly return on nine Dutch industrial indices and the stocks of nine multinational companies. This is due to the diversification effect in literature that is associated with the multinational companies. So, for Pl:

where is the weekly return on the benchmark index in period t, is the constant term, are the weights assigned to resp. the market indices, the industrial indices and the multinational companies, , and are the weekly returns on resp. the market indices, the industrial indices and the multinational companies and is the error term. The weights, as in previous studies, will come from regressing the weekly returns of the benchmark on the other indices and these will be used for rebalancing on a yearly basis. So for 2004, I will use the data from January 2003 until December 2003.

As in the previous studies, there are many securities. I will therefore use a stepwise regression since there are a large number of potential independent variables and there is no underlying theory on which to base model selection (Brooks, 2008). This stepwise regression will keep on adding independent variables one by one, and they are all being checked whether they are significant at a 10% level. The stopping criteria for the stepwise forward regression is a p-value of 0.2. The portfolio weights can of course be negative, since short selling is no restriction.

After the construction of the portfolios, I will assess the benefits of home-made diversification in two ways. I will use a correlation analysis to see if home-made diversification can mimic international diversification. After that, I will assess whether the mimicking portfolios can exhaust the international diversified portfolios. For this, I will use the information ratio and a mean-variance spanning test.

Correlation analysis

Antoniou et al. (2010) use a correlation analysis to assess whether home-made diversification can mimic international diversification, once the portfolios are constructed. The lower the correlation, the greater the benefits of international diversification. So in this study, I will measure pairwise correlations between the returns of the home-made portfolios and a target benchmark portfolio. The higher the correlation, the stronger ability to mimic foreign portfolios.

The pairwise correlation are calculated for the AEX, Pi and Pl. The null hypothesis here is

and from Pi to Pl. This is done to check whether Pl can improve on Pi by adding multinational

equity, and to check whether Pi can improve on the AEX by adding industry equity. To test

the null hypothesis, the following Z-statistic is being used:

Where ρ1 and ρ2 are the correlation coefficients of the first and second mimicking portfolios with their underlying foreign indices, n1 and n2 are the number of observations in each portfolio. The value of is calculated as . The critical value at 5% level of significance at which the null hypotheses can be rejected is zc = 1.96. A

Z-score larger than 1.96 suggests that the correlation does not remain constant, and in that case home-made diversification can mimic international diversification.

The possibility of exhausting the benefits of home-made diversification is tested with the Sharpe ratio and the mean variance spanning test.

Sharpe ratio

This tests the risk/return trade-off for a risk averse investor. It provides valuable information regarding the diversification benefits of the home-made diversified portfolios and the internationally diversified portfolios. The Sharp ratio is calculated as the ratio of the mean portfolio divided by the standard deviation annualized:

Where is the average expected return over the period, is the standard deviation of the portfolio. If the Sharpe ratio is higher, it implies a better strategy.

Mean variance spanning test

minimum-variance frontier of the K + N assets. If we extrapolate this towards this study, including foreign equity in the domestic portfolios will improve the efficient frontier if the foreign index is not spanned by the mimicking portfolios (Antoniou et al., 2010). So I will test the joint restrictions in the equation as follows:

where is the weekly return on the benchmark portfolio, where is the constant term,

is the weekly return the mimicked portfolio and is the error term. This will be

estimated using OLS, with a null hypothesis of:

4. RESULTS

First in Eviews I created the home-made-diversification portfolios. I did this as presented in the methodology section, and will provide one example of this here for reasons of limited space in this study. For the portfolio based on industry indices, I stepwise regressed the industry indices on the world index per year. For 2004, I used the data from January 2003 – December 2003. This gave the following result:

Table 2: How To Create The Homemade Portfolios.

Dependent Variable: WORLD Method: Stepwise Regression Date: 06/12/14 Time: 10:50 Sample: 1/08/2003 12/31/2003 Included observations: 52

Number of always included regressors: 1 Number of search regressors: 9

Selection method: Stepwise forwards

Stopping criterion: p-value forwards/backwards = 0.2/0.2 Variable CoefficientStd. Error t-Statistic Prob.*

C 0,000 0,001 0,137 0,891 INDUS 0,113 0,052 2,186 0,034 HC 0,228 0,064 3,550 0,001 CON_GDS 0,435 0,139 3,134 0,003 TELE 0,077 0,033 2,312 0,026 TECH 0,138 0,052 2,669 0,011 BAS_MAT -0,135 0,054 -2,491 0,017 OIL_GAS 0,137 0,063 2,168 0,036 R-squared 0,920 Mean dependent var 0,003 Adjusted R-squared0,907 S.D. dependent var 0,029 S.E. of regression0,009 Akaike info criterion -6,490 Sum squared resid0,003 Schwarz criterion -6,190 Log likelihood176,751 Hannan-Quinn criter. -6,375 F-statistic 72,242 Durbin-Watson stat 2,170 Prob(F-statistic)0,000 Selection Summary Added INDUS Added HC Added CON_GDS Added TELE Added TECH Added BAS_MAT Added OIL_GAS

This means that for the 2004 portfolio, I will have to use the following industries and their weights to create the portfolio: Industrials (0.113), healthcare (0.228), consumer goods (0.435), telecommunications (0.077), technology (0.138), basic materials (-0.135) and oil and gas (0.137). The negative value in front of the basic materials-coefficient means I will short this one. The p-value of all the variables are significant at a 10% level, so I will include them all. This is done for all the years and also for the local portfolio. The data is available in Excel.

Correlation analysis

To check whether it is possible to mimic international diversification by creating homemade diversification portfolios, we need to create pairwise correlations between the foreign market index and the domestic portfolios.

Table 3: Correlation table.

World Local Small AEX Pi Pl

World 1.000 Local 0.726 1.000 Small 0.757 0.610 1.000 AEX 0.862 0.727 0.817 1.000 Pi 0.804 0.666 0.784 0.878 1.000 Pl 0.651 0.581 0.568 0.697 0.787 1.000

The table above reports the correlations between all the indices and portfolios. The important ones are underlined, since we only need to look at the pair wise correlations of the benchmark with the home-made portfolios. As can be seen, the AEX is also included in the analysis. The AEX is the Dutch index for the biggest Dutch companies and often used as a benchmark for the Dutch market. It is a useful way to interpret the mimicking ability of the homemade portfolios. If the correlation between the AEX and the foreign benchmark is higher than the correlations between the homemade portfolios and the benchmark, an investor can simply invest in the AEX and achieve better results.

study are the correlations for the Pi and Pl portfolios with the benchmark. In order to analyze whether Dutch investors can mimic the benchmark, I calculated the Z-statistic as shown by the equation in the methodology section. The results can be found below:

Table 4: Z-statistics

The correlation coefficient of the Pi portfolio with the benchmark index (as can be shown in table 3) is 0.804. To analyze the ability to mimic the benchmark, we have to look at the Z-statistic as shown in table 4. Adding Pi to the AEX has a Z-Z-statistic of 0.659. This means that the correlation remains constant at a significance level of 5%. Moreover, the correlation between the AEX and the benchmark is higher than the correlation between Pi and the benchmark (as can be seen in table 3: 0.862 > 0.804). So the home-made portfolio Pi, where

z-Test: Two Sample for Means

AEX Pi

Mean 0,0008054 0,0001785

Known Variance 0,0008556 0,0000291

Observations 438 438

Hypothesized Mean Difference 0

z 0,4410532

P(Z<=z) one-tail 0,3295872 z Critical one-tail 1,6448536 P(Z<=z) two-tail 0,6591745 z Critical two-tail 1,959964

z-Test: Two Sample for Means

Pi Pl

Mean 0,0001785 0,0002845

Known Variance 0,0000291 0,0000363

Observations 438 438

Hypothesized Mean Difference 0

z -0,2742303

the industry effects of the Netherlands have been added has a lower correlation with the world index than the AEX has, albeit non-significant.

The next step is adding the Pl portfolio to the Pi portfolio, and to check whether this would induce benefits to the Dutch investor I once again looked at the correlation and the Z-statistic. The Z-statistic can be found in table 4, which equals to 0.784. This means that also adding the Pl portfolio, the correlation change is still not significant at a 5% level. Also, more important, the correlation itself is shown to be much lower than the previous ones: 0.651. This implies that it is not necessary to change the significance levels of the Z-statistic in order to check whether the correlations would still be constant at a 10% level, because it is always wiser to invest in the AEX index due to the higher correlation with the world index.

From table 3 and 4 the results are (unfortunately) very clear regarding the Dutch stock market. Dutch investors cannot mimic the world index by using stocks in the Dutch market only. The correlation between the AEX and the world index is 0.862 and therefore higher than the homemade portfolios’ correlations with the world index. Also, the changes from adding Pi to the AEX and the Pl to the Pi are not significant at a 5% level, but that doesn’t even matter when an investor is better off investing in the AEX market itself. One important thing that does stand out is the much lower correlation of the Pl and the world (0.651). It seems that adding multinational stock makes the mimicking ability of the Pi portfolio much worse, albeit that it’s not significant at a 5% level.

These empirical findings are on the on hand very important, since it is different from previous studies and therefore has implications for Dutch investors and the questions regarding the international diversification puzzle, but it also forms a problem for the remainder of the study. As can be seen in the methodology section, previous studies found significant correlation changes that allowed them to test the exhausting ability of these portfolios using (amongst others) Sharpe ratios and a mean-variance spanning test.

5. CONCLUSION

In this study, I have taken a further look at the historical view in the world of Finance that the home bias is suboptimal for investors. Specifically, whether creating homemade portfolios in the Netherlands is a suboptimal choice for Dutch investors, or that the diversification benefits from international assets can also be gained by constructing mimicking portfolios consisting of Dutch equity. To do this, I used weekly data from 01/01/2003 up to 23/05/2012 of the world index (as the foreign benchmark), Dutch industries, Dutch multinational companies and domestic indexes. First, I examined whether it was possible to mimic the foreign benchmark and if successful, whether I could exhaust the benefits.

Based on the literature review it appears that until the study by Errunza et al. (1999), the common view was that international diversification induces positive benefits to any investor’s portfolio. However, according to many authors (French and Poterba, 1991; Huberman, 2001; Berrill & Kearney, 2010) there is still a tendency to invest in domestically traded assets, for both institutional and behavioral reasons. Errunza et al. (1999) has made a breakthrough with his study, showing that for the U.S. equity market this tendency isn’t suboptimal at all. Antoniou et al. (2010) has done the same for the U.K. equity market, showing that for industry indices, investing domestically isn’t suboptimal.

To recap, my research question was as follows:

Is the home-bias a suboptimal choice for Dutch investors or can the same benefits from international diversification be gained by mimicking portfolios that consist of domestically traded securities?

(and thus with less diversified stock markets) international diversification is still very relevant for a lot of investors. This provides an important conclusion, because as can be seen in the literature review, the home equity puzzle is still common today. For Dutch investors, this means that they are making sub optimal choices when they do not diversify internationally. More importantly, if an investor in a small country like the Netherlands could get international diversification benefits while only investing in domestically traded equity, the entire problem of the home equity bias wouldn’t exist anymore. This however is not the case.

When looking at the Dutch homemade portfolios, it can be seen that both are not as good as the AEX. However, the Pi (with industry composition only) play a bigger role in diversification benefits than the Pl (with multinational stock). Adding multinational stock makes the correlation between the homemade portfolio somewhat (but not significantly) worse in comparison with when I add the industry effects to the AEX. However, I find that neither the industry effects nor the multinational stock have an effect on diversification benefits. This confirms the findings of Heston and Rouwenhorst (1994) and Jacquillat and Solnik (1978).

This study however also has its limitations. Because the study is focused on the Netherlands, I did not get to work with the amount of data that I would like to, both in time-periods as in numbers of industries and multinationals. I do feel that I have enough observations, but the lack of data could explain some of the differences in my study and the one of Errunza et al. (1999) and Antoniou et al. (2010). Secondly, the Dutch equity market isn’t as diversified as the markets tested in previous studies. This could explain the different results as well, because there aren’t as many multinationals as there would be in the U.S, both in term of quantity and size. In the U.S. it might be easier to capture international diversification than it is in the Netherlands. Also, in previous studies they used country funds to test the overall performance of the domestic market, and whether their results would hold when taking periodic variation into account. I did not do this due to limited time and resources, but it could be that the Dutch market underperformed which led to these findings. Lastly, due to the fact that I was unable to mimic the foreign benchmark with homemade portfolios, I couldn’t exhaust potential benefits either.

REFERENCES

Antoniou, A., Olusi, O. and Paudyal, K. 2010. Equity Home-Bias: A Suboptimal Choice For UK Investors? European Financial Management, 16: 449-479.

Berril, J. and Kearney, C. 2010. Firm-lever Internationalization and The Home Bias Puzzle.

Journal of Economics and Business, 62: 235-256.

Brooks, C. 2008. Introductory Econometrics for Finance. Cambridge University Press.

Brooks, R. and Del Negro, M. 2006. Firm-level Evidence on International Stock Market Comovement. Review of Finance, 10: 69-98.

Errunza, V.R., Hogan, K. and Hung, M.W. 1998. Can The Gains From International Diversification Be Achieved Without Trading Abroad? Journal of Finance, 54: 2075-2107.

Fatemi, A.M. 1984. Shareholder Benefits from Corporate International Diversification.

Journal of Finance, 39: 1325-1344.

Fellner, G. and Maciejovsky, B. 2003. The Equity Home Bias: Contrasting An Institutional With A Behavioral Explanation. Max Planck Institute for Research into Economic Systems,

Papers on Strategic Interaction.

Ferreira, M. and Gama, P. 2005. Have World, Country, And Industry Risks Changed Over Time? An Investigation of The Volatility of Developed Markets. Journal of Financial and

Quantitative Analysis, 40: 195-222.

French, K.R. and Poterba, J.M. 1991. Investor Diversification and International Equity Markets. American Economic Review Papers and Proceedings, 81: 222-226.

Heston, S.L. and Rouwenhorst, K.G. 1994. Does Industrial Structure Explain The Benefits of International Diversification? Journal of Financial Economics, 36: 3-27.

Huberman, G. Familiarity Breeds Investment. The Review of Financial Studies, 14: 659-680.

Huberman, G. and Kandel, S. 1987. Mean Variance Spanning. Journal of Finance, 42: 873-888.

Jacquillat, B. and Solnik, B. Multinationals Are Poor Tools For Diversification. The Journal

of Portfolio Management, 4: 8-12.

Levy, H. and Sarnat, M. 1970. International Diversification of Investment Portfolios. The

American Economic Review, 60: 668-675.

Markowitz, H. 1952. Portfolio Selection. The Journal of Finance, 7: 77-91.

Roll, R. 1992. Industrial Structure and The Comparative Behavior of International Stock Market Indices. Journal of Finance, 47: 3-41.

Root, F.R. 1994. International Trade and Investment. 7th Edition South Western.

Sharpe, W.F. 1964. Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk. The Journal of Finance, 19: 425-442.

Shuman, M. 2009. Invest Locally: Put Your Money Where Your Life Is. YES Magazine: The

New Economy.

Seasholes, M. S. and Zhu, N. 2010. Individual Investors and Local Bias. Journal of Finance, 65(5): 1987-2010.

APPENDIX I: Data Selection

Table 1: List of Industries in the Netherlands. Source: Industry Classification Benchmark.

# Industry

1. Oil and Gas

2. Basic Materials 3. Industrials 4. Consumer Goods 5. Healthcare 6. Consumer Services 7. Telecommunications 8. Financials 9. Technology

Table 2: List of Multinational Companies in the Netherlands. Source: Orbis.

APPENDIX II: Descriptive Statistics Independent Variables

AHOLD AKZO BAS_MAT CON_GDS CON_SVS DSM FIN FUGRO

Mean 0.002 0.002 0.002 0.001 0.001 0.003 0.001 0.005 Median 0.003 0.004 0.004 0.002 0.004 0.002 0.003 0.009 Maximum 0.276 0.198 0.195 0.080 0.189 0.205 0.254 0.205 Minimum -0.729 -0.150 -0.276 -0.102 -0.125 -0.191 -0.196 -0.359 Std. Dev. 0.056 0.042 0.051 0.023 0.025 0.040 0.050 0.049 Skewness -4.031 0.123 -0.181 -0.301 0.170 -0.154 -0.024 -0.951 Kurtosis 63.791 6.150 6.799 4.645 10.722 5.870 6.971 10.203 Jarque-Bera 76778.880 203.852 297.378 62.643 1219.655 170.110 322.024 1133.119 Probability 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Sum 1.118 1.008 0.955 0.718 0.714 1.354 0.258 2.364 Sum Sq. Dev. 1.535 0.875 1.257 0.264 0.312 0.786 1.215 1.174 Observations 490 490 490 490 490 490 490 490

HC INDUS KASBANK LANSCHOT LOCAL_NL OIL_GAS PHILLIPS SMALL