Patriotism in Your Portfolio

Patriotism in Your Portfolio Adair Morse and Sophie Shive∗ May 9, 2003

ABSTRACT More patriotic countries and more patriotic regions within the United States hold smaller foreign equity positions, in the time series and cross section of our data. The patriotism effect is robust to controls for transaction barriers, risk, and information, which constitute the standard explanations for home bias. Adding patriotism explains 12% in the variation of foreign holdings across countries. Economic, political, and social correlates of patriotism do not explain the effect of patriotism on the home bias. We estimate that patriotism is responsible for approximately 9-21% of the U.S. home bias and 7-17% of the Canadian home bias.

∗

Doctoral Students, University of Michigan Business School Department of Finance. Correspond-

ing author: Adair Morse, 701 Tappan Street, Ann Arbor, MI 48109-1234. Tel. 734-846-0057 Fax. 734-647-8133 E-mail [email protected]. We wish to thank Geert Bekaert, Sugato Bhattacharyya, Anusha Chari, Artyom Durnev, Herman Kamil, Han Kim, Linda Tesar, Anjan Thakor, Nejat Seyhun, Parker Shelby, Tyler Shumway, Lu Zheng, Luigi Zingales, and seminar participants at the University of Michigan’s Departments of Economics and Finance for their valuable comments.

Why do people invest close to home? The phenomenon of observed over-investment in home markets appears in a numerous incarnations: over-betting on the home sports teams, over-allocating 401(k) choices to the employer’s stock, and over-investing portfolios toward domestic assets.1 The over-investing in home markets is a puzzle for economic theory in that well-behaved mean-variance optimizers would normally allocate investment in proportion to each asset’s risk-return characteristics. Observed behavior is not consistent with diversification theories. Furthermore, an optimizer who has a personal interest in a home victory, in the success of her employer, or in the growth of her country should hedge possible under-performance by betting on the opposing team, shorting the employer assets, and allocating more assets abroad. Such hedged allocations are rarely observed. The existing literature introduces three principal explanations for home bias: transaction barriers, risk, and information asymmetries. A simple explanation has been largely omitted. Could it be that investor sentiment toward the home country explains part of the equity home bias? For many, the thought of betting against a home team or shorting one’s company’s stock and country’s market index may seem awkward. Such would be representations of sentiment-driven biases. Sentiment toward the home country is the defining characteristic of patriotism. With the purpose of exploring the role of sentiment on investment, we test the theory that patriotism can explain a portion of the observed portfolio selection bias by examining whether the equity home bias is greater for more patriotic nations. The economic significance of patriotism has not previously been examined, yet anecdotal evidence suggests that patriotism influences investment decisions. We were motivated to study this issue following the events of September 11, 2001. After the terrorist attacks on Washington and New York, the stock markets remained closed for a week. In the interim, much speculation ensued as to how far the market would drop; the media 1 The sum of the disproportionately large allocation of a country’s aggregate equity investment to domestic assets is known as the equity home bias.

1

advocated that patriots of the country should not sell their stocks upon re-opening of the market.2 In the subsequent months, patriotic calls shifted to encouraging a re-buying of stocks. In a November, 2001 Spectrem survey of affluent investors, 52% of respondents said they would show their patriotism by making investments in U.S. companies. Why are the patriotic actions of investors significant for the market? The implications of diversification biases are especially important, since undiversified investors take unsystematic risk. If a large enough group of investors holds an undiversified position, the bias may be priced by the market. The notion that investors might heed the call to patriotic behavior has a clear implication; it seems that patriotism may affect markets. The contribution of the paper is to explore the role of patriotism in explaining the equity home bias. Our hypothesis is that more patriotic investors will choose to invest more of their portfolio at home. In that the effect of patriotism is not in conflict with the existing explanations, our aim is to assess whether patriotism can significantly explain variation in foreign equity holdings after controlling for the effects of existing theories. To test the hypothesis, we use a patriotism measure from the World Values Survey, conducted by the University of Michigan. The survey asks individuals in 53 countries whether they are proud to be a resident of their country. Using the average scores to this question as a measure of patriotism, we show that patriotism is significant in explaining the weight of foreign equities in the residents’ total equity holdings. A one standard deviation change in patriotism decreases foreign equity holdings 4%. To control for endogeneity, patriotism instrumented by percentage of rural population is estimated. In instrumented results remain significant, and a one standard deviation in patriotism decreases foreign equity holdings 2%. The significance of patriotism remains after inclusion of controls for the other explanations for home bias. We find some support for risk and information explanations in our panel, and we find strong support for the transaction barriers explanation. After controlling for all of the standard explanations for 2 A November 1991 article in Money criticizes the call for a ‘patriot rally’ to prop up the market on the day it re-opened after the terrorist attacks of September 11, 2001 (Frederick, 2001).

2

home bias, both patriotism and instrumented patriotism explain 12% of the variation in foreign equity holdings. To test the robustness of this finding, we consider a within country sample and a first differences specification. For nine U.S. Census regions, higher patriotism scores are associated with lower foreign holdings. In the cross-country data, a first difference across survey years shows that greater increases in patriotism are accompanied by lower positive changes in foreign holdings.3 It is possible that correlates of patriotism could be driving its effect. We explore variables capturing economic, political and social correlations with patriotism. For upper income respondents of each countries, patriotism is most correlated with greater income disparity, more reported crime, higher trust of the government, less need for discussing politics, and higher charitable and religious inclinations. These correlates do not negate the significance of patriotism when included in the full model regression of foreign holdings on patriotism and the standard explanations for the home bias. Charitableness has explanatory power on foreign holdings beyond its role in patriotism. To test the robustness of our results, we investigate whether patriotic feelings might be endogenously determined by past country performance. We find no reduced significance of patriotism when controlling for past performance. Finally, we estimate that patriotic behavior is responsible for 9-21% of the U.S. equity home bias, and 7-17% of the Canadian home bias. We conclude that investor sentiment has a non-negligible role in an individual’s tendency to invest in the home market. The remainder of the paper is organized as follows. Section I discusses why investors’ patriotism can affect their equity holdings. Section II introduces our measures of patriotism and of the equity home bias. In Section III our main results are presented showing patriotism significant in explaining the home bias. Section IV contains a variety of robustness tests, including tests on the correlates, instrumented patriotism, alternative 3

Both of these tests are informal as the data are severely constrained.

3

measures of the explanations, and investor protection explorations. The last section concludes.

I. Why Might Patriotism Affect Foreign Holdings? Why might patriotism affect foreign holdings? Patriotic investing is an old and widespread phenomenon. In the United States, the promotion of war bonds using patriotic rhetoric dates back at least to the Civil War. To finance the Union effort, the U.S. government sold war bonds with names and distribution tactics that appealed to the patriotic call. Since then, the U.S. government has used war bonds to generate cash during times of country hostilities. For example, after the September 11th attacks, Series I and EE Treasury Bonds were dubbed ‘Patriot Bonds’, and in 2001, their sales rose to $ 6.6 billion, a 43% increase over the previous year (Sulon, 2001). While appeals to patriotism appear to encourage investment, no study has tested the economic impact of patriotism. French and Poterba (1991) and Tesar and Werner (1995) empirically document that investors’ domestic holdings are very high. French and Poterba calculate that U.S., Japanese and U.K. investors held 94, 98 and 82 percent of their portfolios in domestic assets. If one were to take the CAPM literally, each investor would hold the world market portfolio, thus holding all stocks in proportion to their market capitalization. Taking into account human capital, which follows the country’s fortunes, investors should invest even less in their home country (Baxter and Jermann, 1997). Pastor (2000) calculates that U.S. investors’ belief that their home market is as mean-variance efficient as the world market (it is not) must be extremely strong to justify the home bias in their equity holdings. It is not surprising that Obstfeld and Rogoff (2000) categorize the home bias as one of “the six major international macroeconomics puzzles”. Grinblatt and Keloharju (2001) find that Finnish investors prefer stocks of Finnish firms, especially those that communicate in their own language and are operated by Finnish executives.

4

While this is consistent with investors’ possessing additional information about firms that communicate in one’s own language, it is also consistent with patriotic investing. No previous study has explored the equity home bias in a behavioral light, although two closely related puzzles lend themselves well to behavioral theory. We briefly describe the role of ‘patriotism’ in the sports betting market and in the 401(k) asset allocation choice. Consider the (illegal) sports betting market and the case of a non-professional better. If the better is risk averse, expected utility theory predicts that she will hedge the possibility of low utility from a home team loss by betting on the opposing team. Contrary to this prediction, individuals tend to bet on their home team, even with unfavorable odds (Gray and Gray, 1997). Strumpf (2003) observes that illegal bookmakers charge extra for home team bets to compensate for the additional risk they incur due to the home bias of their clients. The same betters may be willing to bet on their own country’s stocks, even with unfavorable returns. A second related puzzle is the pattern of 401(k) portfolio allocations. People invest disproportionate amounts of their discretionary 401(k) retirement assets in their own company’s stocks. For example, Bernartzi (2001) finds that the retirement fund of Coca-Cola allocated 90% of its value to the firm stock, and employees themselves allocate 76% of their discretionary contributions to Coca-Cola shares. Although a typical large company is twice as risky as a diversified portfolio, John Hancock Financial Services (1999) reports that a majority of employees feel that their own company stock is safer than a diversified portfolio, and only 18% realize that their stock is riskier. In a similar study, the Vanguard Group (2001) reports that participants see ‘individual stocks’ as more risky than a diversified equity fund, and their own company stock as less risky. In a survey conducted by Bernartzi (2001),4 only 16.4% of respondents believed that their company stock was more risky than a diversified portfolio. This practice strongly contradicts mean-variance theory; the observed 401(k) allocations fail to diversify away 4

This can be found on Morningstar.com.

5

a firm’s unsystematic risk. Even worse, the employer stock is the asset most correlated with the employees’s human capital, and thus a rational agent should short employer stock, not hold large quantities of it. Both utility and behavioral theory can help us understand the home team bias, the 401(k) bias, and the equity home bias. First we consider traditional utility theory. If the future is not fully discounted and if capital flow is somewhat constrained across borders, then investing at home may bring future material benefits, such as jobs, improved infrastructure,5 and benefits from the taxes paid by corporations on the same soil. The anecdote below highlights the relevance of this point. A large central bank in an emerging market brought in a new director of the pension fund department who was to choose the portfolio allocations for the fund. After studying the portfolio, the new director approached the head of the central bank with an optimal international diversification strategy. The central bank head quickly rejected the plan, retorting that the fund investment was not to leave the country; the country needed the capital for growth. Investment in the home market might generate intangible utility such as the approval of others and the feeling of contributing to society. It may also reduce regret if the investment has low or negative returns. If monetary and non-monetary benefits outweigh diversification benefits, biased investment strategies are rational. Otherwise, they must be due to an irrational influence. In the case of home team betting, the benefits must be emotional because the money wagered cannot affect the outcome of the game. According to Kahneman and Lovallo’s (1993) ‘inside view’, insiders may strongly identify with an organization and may find it difficult to hold an independent view on the expected returns to company stock, and investors may have similar trouble evaluating their home market objectively. An inside view could either elicit overestimation of the mean return 5 Huberman (1997) shows that people invest a large portion of their portfolios in their local phone company.

6

(optimism)6 or underestimation of the associated risk (overconfidence)7 . Of course, the inside view allows only for such biases toward the entity with which an individual identifies.

II. Measuring Patriotism and the Equity Home Bias Patriotism: Love of country; devotion to the welfare of one’s country; the virtues and actions of a patriot; the passion which inspires one to serve one’s country.

- Websters’ Dictionary

Measuring patriotism is difficult because its definition is broad, and measures of individual’s emotions are scarce. The patriotism data come from the University of Michigan’s World Values Surveys conducted during two periods, 1990-1992 and 19951997.8 Appendix A highlights details of the survey methodology and questions posed. To measure patriotism, we focus on individuals’ responses to the question: How proud are you to be [substitute nationality]? Our measures of patriotism are mean country scores, averaged over individuals’ responses from 1 (not patriotic) to 4 (very patriotic). Since the more affluent account for 6

Tesar and Werner (1995) find that the lack of diversification due to the home bias can be explained by a 620 to 800 basis point over-estimation of domestic returns. Optimism can feed on itself; investors can be convinced that their home market is superior and selectively ignore contradictory evidence (cognitive dissonance). 7 Overconfidence is explored for example in Odean, (1999) and Barber and Odean, (2000). The “illusion of control” (Langer 1975), in which people treat chance events as controllable, may cause investors perceive lower risk due to their influence on the outcome. There is also evidence that overconfidence increases with familiarity (Heath and Tversky 1991). 8 The span of years reflects the time range in conducting the survey. Since the majority of countries in our sample for the first survey range were surveyed in 1990, we use 1990 as the base year for the first series. For the second survey, we use 1996 as the base year, reflecting a median and mode of the survey years.

7

the majority of stock holdings, we filter the responses to reflect only the upper half of income earners in each country.9 Table 1 lists the patriotism scores for the two survey time periods, from which we can draw some key insights. The table is arranged in increasing patriotism order for the countries in Survey 1 (1992-1993) and Survey 2 (1995-1997). The mean score increased marginally form 3.21 to 3.30.10 A systematic bias in the pattern of patriotism scores is not evident in the sample. The 1990-1992 survey does suggest that New World countries are more patriotic than Old World countries, but the larger 1995-1997 survey results are more diverse within income and location groupings. It is clear that Western European countries are less patriotic relative to the rest of the sample. Our objective is to use the patriotism score to analyze the home bias, specifically asking whether patriotism predict how much foreign equity each country’s nationals hold in their total portfolios. The foreign equity holding variable is a weight of foreign equities in a country’s aggregate equity portfolio. In other words, foreign equity holdings is the percentage of a residents’ total equity portfolio that is invested abroad.11 We calculate a weight of foreign holdings relative to total holdings (FEH) as follows:12 FEH =

Foreign equity holdings Market capitalization + Foreign equity holdings − Foreign equity liabilities

Data limitations in the survey and in the availability of foreign holdings restrict our sample to 29 countries, six of which have data for both 1991 and 1996, resulting in 33 data points all together. We were concerned that the availability of foreign holdings data was biased toward large countries. Thus, we perform a Heckman selection test to determine 9

In the United States, only 24% of wage earners under $30,000 held stock and 84 percent of Americans with income over $75,000 hold equities (Langer, 2001). Clearly, the divergence grows even larger for poorer countries in which disposable income for the poorer half of the population is limited. In the empirical results that follow, the use of the upper income mean of patriotism slightly improves our results relative to the full sample mean, but does not change the significance of the patriotism variable. 10 For countries participating in both surveys, the average patriotism score increased form 3.25 to 3.38. 11 Appendix B provides sources for all the variables used in the study. 12 Our results remain if we simply normalize foreign equity holdings by market capitalization.

8

if the selection of countries explained by GDP or market captialization is exogenous to the estimation of foreign holdings on patriotism. We fail to reject that country selection is independent of GDP and market capitalization with respective p-values of 0.94 and 0.13. This study is biased against finding evidence of patriotic investing, because we test the hypothesis that investors are only patriotic about their own country, not about a region or a common market of countries. Since we examine total foreign outward investment, and not toward which countries the outbound investment is aimed, we do not determine if citizens are simply investing in a country’s close allies. Two interesting possibilities arise if we find no relation between patriotism and foreign equity holdings. The first possibility is that people’s portfolio decisions are not swayed by emotion towards their country. Since patriotism affects other decisions such as careers and purchases, and studies in behavioral finance have shown that sentiment affects investment decisions, this would be an interesting finding. The second possibility is that patriotic people invest not only in their country but also in the group of countries that they consider ‘friendly’. Equity cross holdings data would allow us to test which of these hypotheses is correct. In the next section, we present the patriotism and foreign equity data and show our key result that patriotism is significant in explaining the panel of foreign equity holdings.

III. Empirical Results A. Panel Results The central finding of the paper is depicted in Figure 1, a plot of the foreign equity holdings against patriotism. This figure suggests a decreasing linear relation between pride-patriotism and foreign equity holdings. Patriotism declines as foreign equity hold9

ings increase. To support this claim, a simple regression including a dummy for the 1995-1997 survey period shows that patriotism is significant in explaining foreign investment, with an R-square of 0.22. The result appears as the first regression on Table II. Following the terrorist attacks of September 2001, the Boston Globe related a story of an individual who “wondered what would happen ‘if every red-blooded American... bought a few shares of their favorite stock on Monday.”13 We test the economic significance of our findings. According to our regression coefficient which is very stable across specifications, a one-unit rise in the patriotism score will cause the foreign holdings in the total equity position to rise by approximately 11 percentage points. To compare this to the equity home bias, note that the optimal CAPM portfolio consists of equity in all countries in proportion to their market capitalization. For example, in 1996 the United States comprised 33% of the world market capitalization. However, it held 93% of its equity at home. The difference, or home bias, is 60 percentage points. If, on a scale from 1 to 4, we can assume that 2.5 is a neutral answer, our regression results estimate that if the United States were to have a neutral patriotism score, the home bias would be reduced by 0.16, (a 27% reduction). We are cautious in drawing too much inference from this result before we explore other explanations. In the remainder of this section, we control for other variables which may be influencing the source of foreign holdings decisions.

B. Standard explanations for the Home Bias Four main explanations have been proposed for the equity home bias. The first is that transactions barriers make it difficult to invest abroad. A second set of studies suggest that differences in the availability of information explains the home bias. Another line 13

The Boston Globe, September 18, 2001

10

of work considers additional risk that may come from investing abroad, potentially outweighing diversification benefits. We present our principal variables here, and perform robustness checks by including other variables in the following section.

B.1. Transactions Barriers Transaction barriers should negatively affect foreign investment, somewhat like a tax paid on the absolute value of the holdings of foreign stocks (Stulz, 1981). Unfortunately, measures of barriers to outward flows or inward repatriation of capital gains and dividends are limited. Hence, we seek measures which may be highly correlated with foreign equity holdings facilitation. Chari and Henry (2002) show that when a capital-poor country is liberalized, there is an infusion of capital from abroad, supporting the theory that liberalization allows investors to allocate capital optimally between countries. Since liberalizations allow capital to flow in both directions, the more liberalized countries’ nationals may be more likely to invest abroad. In studying the effects upon market growth, Bekaert and Harvey (2000) collect financial liberalization dates of emerging countries. Since all countries in our sample were ‘liberalized’ by the first survey’s date according to this paper, we use a dummy variable for capital account liberalization from the IMF Exchange Arrangements and Exchange Restrictions Annual Report.14 Additional details for these variables appear in Appendix A. Although our hypothesis is that foreign holdings should decrease with transactions barriers, it is well known that transactions barriers are likely not the entire source of the home bias (Cooper and Kaplanis 1994, Tesar and Werner 1995, Errunza, Hogan and Hung 2000, Glassman and Riddick 2001 and Fourth, Ahearne, Griever and Warnock 2001). 14

A measure of tariff and non-tariff barriers from the Economist Intelligence Unit (EIU) database yields similar results

11

Our regression results appear in Table IV. The coefficient on transactions barriers is positive and significant. When we add patriotism to the model, both patriotism and transaction barriers remain significant. We infer that patriotism is an effect separate from transactions barriers. The low R square on the single variable equations suggests that transaction barriers alone are not the complete source of home bias, but barriers appear somewhat economically significant. Table III shows the estimated change in foreign holdings for a one standard deviation change in other variables that are significant in explaining foreign holdings. According to this table, a one-standard deviation change in the IMF capital account restrictions would change the home bias by 5.5 percentage points. Adding transactions barriers to a model with patriotism and variables for information and risk model improves the R-square by .128.

B.2. Information Several studies find that differential information about securities contributes to the home bias.15 Brennan and Cao (1997) develop a model of international equity portfolio investment flows and show that differential information endowments among investors concerning equity knowledge can affect their investment decisions. Bhattacharya and Groznik (2002) find that U.S. investment in a particular foreign country is positively related to the income of the immigrant group from that country living in the U.S. Bhattacharya and Groznik infer that immigrants have more information about their home countries. Information effects are difficult to decipher; the authors find that the level of investment in foreign countries is unrelated to language or physical distance, which are the core components of distance gravity measures. It seems that information regarding investments across borders is not merely a matter of closeness, but rather a matter of specific country knowledge. Our information results are consis15

See Huberman (1999), Choe, Kho, and Stulz (2000), Froot, O’Connell and Seasholes (2000), Coval and Moskowitz (1999),(2001), Grinblatt and Keloharju (2001) Bhattacharya and Groznik (2002)

12

tent with this categorization of information as well. However it is worth pointing out that Bhattacharya and Groznik’s results are also consistent with our patriotism claims. Immigrants may invest in their countries of origin if that is where loyalties lie. There has been strong support for the information hypothesis for studies of U.S. and international investment. To name a few, Coval and Moskowitz (2001) show that mutual funds earn abnormal returns when they invest in nearby firms. This effect is more pronounced in small firms that operate out of remote locations, suggesting that the mutual fund managers have an informational advantage when they invest in these firms. Choe, Kho, and Stulz (2000) find that in Korean data from 1997 to 1998, domestic individual investors have a short-lived private information advantage for individual stocks over foreign investors, but almost no evidence that domestic institutional investors have such an advantage. Information about a stock may aid in an understanding of the parameters of the returns model and of the uncertainties inherent in the model. The comfort of feeling trust in domestic institutions closely resembles a set of behavioral findings that people dislike ambiguous situations.16 In particular, people underweight choices of which they have difficulty understanding the distribution of outcomes.17 We do not attempt to distinguish between the two explanations related to information: preference for reduced ambiguity (model risk), and the possession of true information (knowledge about the model parameters). To proxy for the information that may be conveyed about foreign firms through the media, we use the commonly used measure of newspaper circulation per capita

18

. We

choose this variable because newspaper circulation is likely to have a similar effect for all countries, contrary to televisions and internet which may have been scarce ten years ago in some countries. Table II shows that number of newspapers per capita is significantly 16

These findings are summarized by Barberis and Thaler (2002). Heath and Tversky (1991) 18 Number of televisions per capita yields the same results 17

13

and positively related to foreign equity holdings. When we add patriotism to the model, both remain significant and the coefficients do not change much, suggesting that both information and patriotism are important in explaining the home bias.

B.3. Risk An investor may choose to invest abroad to mitigate financial risk. An international portfolio may provide diversification, although the benefits are not clear because correlations between countries tend to increase in bear markets. Consistent with the CAPM, we predict that nationals of countries with high financial risk and low correlations with the world portfolio will choose to invest more abroad, other things equal. Contrary to this prediction, we find that correlation with the world index is positively related to foreign holdings. This is likely due to the fact that countries which are integrated tend to invest more abroad. Table VI shows that a one-standard deviation change in the correlation with the world would change foreign equity holdings by 5.5 percentage points. A one standard deviation change in diversification benefits changes our measure of foreign equity holdings by 3 percentage points. Other variables that proxy for risk are presented in the Robustness section.

C. Within-Country and Time Series Results To further investigate the role of patriotism on foreign equity holdings, we study the six countries that are present for both surveys in our sample. Our hypothesis is that when patriotism rises, foreign ivestment decreases more (or increases less) than instances when patriotism falls. Since our data constraints inhibit formal statistical tests, we simply plot changes in foreign equity holdings against changes in patriotism. Figure 2 shows that while there are only six countries in the sample that have data from both surveys, foreign investment falls when residents become more patriotic. The figure also

14

suggests that increasing globalism and increasing patriotism characterize the period between the 1990-1992 survey and the 1995-1997 survey. Germany is the exception in its decrease in patriotism, potentially because of the economic strains from unification during the period. The striking result is that even though the entire sample globalized their portfolios to some degree, Figure 2 shows clearly that countries who experience more growth in patriotism had less globalization of their portfolios. We have presented initial results across countries for points in time and across countries across time. Would patriotism explain portfolio variation within countries? Although we are again data constrained, we obtain patriotism data from the 1995-97 World Values Survey for nine regions in the United States and compare them to foreign equity holdings reported in the 1997 Survey of Consumer Finances. Since it is difficult to find household net worth data for these regions, we use household income as a proxy and normalize foreign holdings by this figure. The results are plotted in Figure 3. The figure shows that the more patriotic regions of the United States – West South Central, Mountain and West North Central – invest the least in foreign equities. Although one cannot draw statistical conclusions from nine data points, the results support the hypothesis that even within a single country, more patriotic investors are more home biased in their investments. To summarize this first analysis, we find that foreign holdings are inversely related to patriotism, and investigate whether the patriotism results holds in the presence of the the standard explanations for the home bias. In the following section, we consider some tests of the robustness of our results.

IV. Robustness Tests This section considers some correlates of patriotism that may be behind its effect. Next we explore whether a common variable explains both patriotism and foreign invest15

ment holdings. We also test for the robustness of the patriotism variable against other measures of transactions barriers, risk, and information. Last, we consider the role of corporate governance. We find that these tests do not negate the effect of patriotism on the home bias.

A. Correlates of Patriotism We investigate what factors relate to a country’s patriotism score. Then, we consider whether any of these factors drives the patriotism-foreign equity holdings result displayed in Figure 1. We distinguish three types of possible correlates of patriotism - economic, political, and social - and examine which ones are significantly related to our patriotism score. First we explore factors that may affect residents’ patriotism for economic reasons. We expect patriotism to decrease with income disparity; high inequality, especially if the inequality resulted in poverty at the low end, may be viewed as the ineffectiveness of the state. We use income disparity as measured by the Theil Index from the University of Texas. We also consider GDP per capita. Partial correlations are reported in Table IV. Controlling for the survey year, both income disparity and inequality are significantly correlated with patriotism. Regression results are presented in Table V. Since our data points are limited and since the variables are all collinear, we only run patriotism against each variable alone. We find that more income disparity is indicative of more patriotism. This result is surprising; the ex post intuition is that the upper half of populations are more patriotic for the populations who have advantaged the most. In other words, since our measure of patriotism is confined to the upper income half of every population, it may be that where disparity is high, the upper income groups are benefitting most from the government and thus are more patriotic.

16

Next, we explore four possible political correlates to patriotism. All of the political correlates have significant explanatory power. The first two political correlates to patriotism are the levels of crime and corruption. Presumably a country with low crime and corruption levels will have more patriotic residents. Our measure of crime,the number of murders, robberies or other violent crimes per 100,000 households, is taken from the 1995 World Competitiveness Yearbook. The corruption index is from the Internet Center for Corruption Research, University of Goettingen and Transparency International. Two other possible political correlates to patriotism are from the World Values Survey. Responses to ‘How much do you trust your national government?’ and ‘How often do you discuss politics at home?’ capture faith and interest in national politics, respectively. In the latter case, however, a converse argument could be made that political discussions may be enhanced by a dissatisfaction with the government in power. Table IV reports that more crime and more corruption are indicative of higher patriotism. The positive association between crime and patriotism remains even after controlling for country per capital GDP. Corruption is negative and no longer significant after including GDP in the regression. We posit that this result may be due to increased reporting of crime resulting from better law enforcement, or it may be a sign of economic freedom. Trust of government is positively correlated to patriotism (those who trust their government tend to be more patriotic); however, the unreported partial correlation is not significant. Discussing politics is significantly negatively related to patriotism, meaning that those who frequently discuss politics at home tend to be less patriotic. The inference is that more discussions of politics might be related to discontentment with the governing powers. Three possible social correlates to patriotism are the involvement of citizens in charitable organizations, the importance of religion, and ethnolinguistic fractionalization within the population. The first two variables are from the World Values Survey, and the third is from Easterly and Kraay (1999). Ethnolinguistic fractionalization increases

17

with the number of ethnic groups and languages within a country. We expect more charitable and religious populations to be more patriotic, and we expect ethnically diverse populations to exhibit more loyalty to communities and less national patriotism. The social section of Table V presents the regression results of patriotism on social correlates. All of the social correlates are highly significant. More charitable and more religious people are also more patriotic. Contrary to our prediction, higher ethnolinguistic fractionalization is correlated with higher patriotism. In our sample, ethnolinguistic fractionalization is correlated with New World Countries. It is unclear whether entholinguistic fractionalization is itself the attribute driving this association. Given that several variables are correlated with patriotism, we are interested in knowing whether any one of them is driving the significant explanatory power of patriotism on foreign holdings. We regress patriotism individually on each of the economic, political and social correlates, including a survey year dummy variable. Next, we include patriotism and variables to control for the three other possible causes of the equity home bias in the model to check if the correlate is a driving force behind patriotism’s significant explanatory power of foreign equity holdings. These results are shown in Table V. Of the variables listed above, only Charity has explanatory power for foreign equity holdings, and it does not negate the effect of the patriotism variable. We conclude from the correlate analysis that patriotism seems to capture a number of fundamental characteristics of a country’s residents. More religious people are more patriotic and invest less abroad. We do not draw inference as to what the primal cause might be. Instead, we show that patriotism is a bundle of properties which may be influencing portfolio positions.

18

B. Endogeneity We perform two further robustness checks to determine whether our significant patriotism results are proxying for some omitted variable. First, we test whether patriotism is endogenous to a country’s equity market performance. We construct the one, three and ten-year lagged excess market returns for each country in our sample and find that they do not explain the proportion of foreign equity holdings, either by themselves or as an offset to patriotism. The lagged returns also do not explain the patriotism score. Lagged market returns may be a noisy measure of how nationals perceive their home market’s performance. For this reason, we use the residuals from this regression and find that the patriotism cross section that is unexplained by lagged returns is still significant in our analysis. In that rural populations are insulated more from the influences from the rest of the world, choose to instrument patriotism with the percentage of the country’s population that lives in rural areas. The number of people in rural setting will should not be correlated with any omitted variable that is endogenous to patriotism and foreign holdings. As a first quick test of the appropriateness of our instrument, the correlation between rural population and the residuals of the full model of foreign holdings on patriotism and the three standard home bias explanations is not significant at 0.14. In the first stage of a 2SLS procedure, our instrument significantly covaries with patriotism with an R-Square of 0.12. Although this R-Square is rather low, the instrumented patriotism is significant at the 1% level in explaining the Home Bias. The results are presented in columns 2 and 10 of Table II. In a model of foreign holdings solely on patriotism and the survey dummy, a Hausman test of the equivalence of the instrumented and the OLS patriotism coefficients cannot fails to reject. In the full model of foreign holdings on patriotism, transaction barriers, information and risk, the Hausman test rejects equivalence of the instrumented patriotism coefficient and the OLS coefficient. For this reason, we calibrate our economic significance results to the use of the instrument.

19

With the OLS patriotism, a one standard deviation change in patriotism would be associated with a 4% change in foreign holdings percentage. In the instrumented model, the same change in patriotism is associated with a 2% change in foreign holdings. Although the economic significance is lowered, a 2% change in foreign holdings relative to the total equity position is a large impact. For the U.S. and Canada, a shift of patriotism down to the neutral patriotism score of 2 would explain 9% and 7% of the home bias. The OLS estimate of the patriotism coefficient explains 21% of the U.S. home bias and 17% of Canada’s home bias. Finally, we note that inclusion of the instrumented patriotism in the full model of the standard explanations for the home bias increases the variation explained of foreign holdings by 11%. This is almost identical to the 12% increase in R-Squared seen when using the OLS patriotism. The instrumental analysis concludes that the association of patriotism with foreign equity holdings is very robust to endogeneity controls. The second robustness test we undertake is to search for an instrument of patriotism that is not correlated with economic activity.

C. Alternative Variables We consider whether alternative measures for transaction barriers, information, and risk affect the home bias and our patriotism variable. As a further robustness check, we control for the importance of banks, the extent of development and GDP per capita. First we consider whether other measures of the existing explanations for the home bias also explain the effect of patriotism. As alternative control for transactions barriers, we use a measure of financial restrictions from the EIU Database. Like the IMF variable, it is positively and significantly related to the extent of equity holdings, but the effect of patriotism remains.

20

Next, we consider some alternative measures of information available to a country’s investors. Investors may be more informed if they are ‘close’ to a firm. For example, it may be more difficult for Japanese investors to learn about foreign investment opportunities than for German investors, because Germany is physically close to several other large capital markets. Hypothesizing that information is negatively affected by distance, we use three variables to test for the effect of distance. The first is number of airline departures normalized by population. Air travel is the leading means to travel abroad, and frequent travel brings people effectively closer to foreign countries where they can become informed about foreign stocks. Our measure also includes domestic flights, but the normalization by population (and thus country) size is an attempt to control for this factor. Airline departures per person are significantly positively related to foreign investment, suggesting that individuals who travel more often by air invest more abroad. A one-standard deviation change in airline departures may change foreign equity holdings by 4.7 percentage points, which is an economic significance similar to that of television news or transactions barriers. We also consider two physical distance measures.19 We define proximity over a domain of all countries B with listed market capitalization as follows: 
MktcapB ProxA = Average (MaxDist-DistAB ) Mktcap−A

(1)

Where MaxDist is the maximum distance in between two countries in our sample (we want to construct a proximity, not a distance variable), DistAB is the distance between countries A and B, MktcapB is the market capitalization of country B, and Mktcap−A is the sum of the market capitalizations of all countries excluding A. Table V shows that this proximity variable is of the right sign - countries that are ‘closer’ using this variable have higher foreign investment measures - but is not significant. We also employ a natural measure of distance between a country and the average of all other countries, 19

Data for this ‘gravity’ measure is from http://www.macalester.edu/research/economics

21

and it is not significant. We hypothesize that distance may only be important when comparing local firms to other domestic firms. Since our study is not refined to the level of Coval and Moskowitz (2001), who have data on individual firms, we are not able to pick up this distance effect. We test several financial risk measures for robustness. The first is from Campbell Harvey’s (1995) Country Risk analysis tables. The study contains several country risk variables: Credit, Economic, Financial, and Political, but they are pairwise highly correlated (correlation coefficients above .9), so we use only the financial variable. Our second variable is the Sharpe ratio of the country’s returns, minus the Sharpe ratio of the world portfolio translated into local currency, to capture exchange rate movements. Sharpe ratios are constructed with daily returns from the previous year from Datastream and the one-year LIBOR rate on the last day of the year. If this measure is high, investors are less likely to gain less from diversifying abroad. Our third variable is the difference between the variance of stock returns on the local exchange relative to the covariance of the local exchange with the world market. This is also a measure of diversification benefits according to the following argument. If markets are segmented, the difference between the local exchange variance and the covariance between the local and world markets will be high, and there will be gains from diversification. If this difference is low, the local market behaves much like the world market, and systematic risk (and the benefit from diversification) is low. According to Table VI, both the own country financial risk index, and the measure of diversification benefits are significant in explaining foreign equity investment. None of these affects the patriotism variable, and the significance of financial risk disappears when patriotism is added to the model. Other variables may be important in explaining the home bias. In this section we test these variables Following Demirguc-Kunt and Levine (1999) and Zingales (1998), we include an indicator of whether the economy is bank (relationship) based or market

22

based. Our purpose is to capture investor portfolio choices and not portfolio allocations by banks. There are competing hypotheses as to the sign of the bank variable. If bankers’ investment choices are less affected by behavioral biases, then the bank coefficient should be positive and the patriotism coefficient should increase its negativity. Conversely, if relationship-based systems imply a connection between debt and equity markets, then the bank variable should be negative. Next, we include an indicator for developing countries from Demirguc-Kunt and Levine (1999). The holders of foreign equity may not be representative of domestic investors in developing countries because the state and large institutions may play a larger role in developing country equity markets. The developing variable could also capture differences in education, risk, or other factors. Table ? shows that development is not significant, possibly because of multiple effects and the low power of an indicator. Finally, we control for GDP per capita. In higher income countries, a larger proportion of the population invests abroad. We include the GDP per capita to control for any effects that reflect income in the selection of who makes the investment decision.

D. Investor Protection As a final set of robustness checks, we consider the corporate governance environment of the investing country. Motivation for this section stems from Dahlquist, Pinkowitz, Stulz and Williamson (2001) (DPSW). DPSW consider the environment of the receiving country of U.S. outward investment. In particular, they focus on the amount of closely held shares in a country and the investor protection ratings. The amount of closely held shares captures the role of the large shareholders in the country in controlling the market portfolio. Their hypothesis is that closely held percentage measure provides an indication of the amount of a market available for outside investment. DPSW find that indeed the closely held measure is consistently positively significant in explaining the

23

U.S. outward investment to that country. Additionally expropriation risk explains a portion of the variation in investment positions. In our study, we are focused on the cross-section of investors’ countries rather than the cross-section of host countries for investment. In DPSW, the governance affects the amount of inflows of investment. In our study, the overall outflows may be affected by the governance at home. We construct an average measure of closely held shares for each country by using the average closely held percentage for a country for all firms in Worldscope. Then we take the difference between the closely held measure for a country and the rest of the world average. Table VII presents the result that the closely held percentage neither increases nor decreases the foreign holdings. Closely held may have two offsetting effects. On the one hand, the relative percentage of closely held shares may reflect poorer governance, thus encouraging more foreign investment. Conversely, the closely measure is a reflection of the investor group for a country. Thus, a higher closely held percentage may reflect more inward-focused investment if the largest shareholders are extracting private benefits of control. To isolate if these governance effects are at work, we follow DPSW and include first investor protection and then expropriation risk from EIU. Table VII shows that a country with better investor protection scores and less expropriation risk invest more abroad. This is inconsistent with the concept that investors flee their own poor governance countries and is consistent with the hypothesis that a country’s overall investment position is impacted by inward biases resulting from holding private benefits of control.

V. Conclusion We test the relation between patriotism and the equity home bias, and find that investors in more patriotic countries hold smaller foreign equity positions. Similarly, investors in more patriotic U.S. regions hold less foreign equities. Changes in patriotic behavior 24

are negatively related to changes in foreign equity holdings. This result is robust to controls for transactions barriers, risk and information, and economic, political and social correlates of patriotism do not negate the significance of patriotism. We find strong support for the transaction barriers explanation for the home bias in addition to patriotism. In our sample, patriotism explains 9-21% of the US home bias and 7-17% of the Canadian home bias. Additionally, inclusion of patriotism accounts for 12% in variation of foreign holdings, and a one standard deviation change in patriotism decreases foreign holdings 2-4%. Our results are robust to tests of endogeneity. We find that neither instrumenting patriotism with the percentage of rural population nor including past country index returns negates the statistical or economic significance of our patriotism result. We also test for the robustness of the patriotism variable against other measures of transactions barriers, risk, and information. Last, we consider the role of corporate governance. Higher investor protection and lower expropriation risk are increasing with foreign equity holdings. This result is consistent with the hypothesis that domestic investors in poor governance countries are able to take advantage of private benefits of control. None of the robustness tests of endogeneity, alternative measures and corporate governance negates the effect of patriotism on the home bias. Three implications can be drawn from our study. The first is that irrational behavior appears to have a large part in explaining the phenomenon that manifests itself in the home team bias, the 401(k) investment allocations, and the equity home bias. The second is that non-mean variance optimizing behavior may be a large determinant of investment decisions. The third implication of this study is that policies aimed at reducing the home bias may be less successful in countries where investors prefer not to invest abroad because they are patriotic. Investor education may have an important role in reducing the home bias. Patriotism results in a winner’s curse in the sense that the person valuing a stock most highly will ultimately be the highest bidder in any auction. The same theory

25

applies to markets as a whole. The citizens of a country will likely be the highest bidder for their own country’s assets, thus possibly driving up the price (and capitalization) of their own market. In a more general equilibrium setting, people invest disproportionately at home, but a portion of this over-investment is offset by the home biased choices of other countries. This may push prices upward in more patriotic countries, and downward in less patriotic countries. Price distortion, however, does not necessarily point to a ”free lunch” (Barberis, Nicholas and Richard Thaler, (2002). The holding of domestic equities may enter directly into the utility function. Thus, if one were to hold the mean-variance portfolio in lieu of the home bias portfolio, one’s utility may be lower. Also, prices may still be unaffected if there exists a group of deep pocketed arbitrageurs who are not affected by patriotism and understand the market capitalization biases caused by countries’ home biases. They can take short (long) positions in a country if domestic home bias dominates (is subsumed by) the bias of foreign investors choosing not to invest in that country. This general equilibrium framework leaves avenues for further exploration of how home bias in general influences the cross section of returns across countries. The effect of patriotism may also explain the home bias in consumer products. Lewis (1999) finds that the equity home bias and the home bias in consumption are linked. Concurrently, Bennett and Young’s (1999) theoretical model suggests that optimal portfolios should be biased towards equities in commodities that attract a large share in its consumption expenditure.20 A future avenue of research will explore whether the effect of patriotism on the equity home bias comes via consumption choices, and whether patriotism has the potential to affect prices.

20

This is controversial: Uppal’s (1993) theoretical model suggests that it is unlikely.

26

References Ahearne, Allan, William Griever, and Francis Warnock, 2001, Information costs and home bias: An analysis of U.S. holdings of foreign equities, Federal Reserve Board, International Finance Division, Working Paper 691, Washington, D.C. Barberis, Nicolas, and Richard Thaler, 2002, Handbook of Economics of Finance. (edited by George Constantinides, Milt Harris and Ren´e Stulz, North Holland). Baxter, Marianne, and Urban Jermann, 1997, The international diversification puzzle is worse than you think, American Economic Review. Bekaert, Gert, and Cambell Harvey, 2000, Foreign Speculators and Emerging Equity Markets, Journal of finance 45, 565–613. Bennett, James, and Leslie Young, 1999, International Stock Market Equilibrium with Heterogenous Tastes, The American Economic Review 89, 639–648. Bernartzi, Shlomo, 2001, Excessive extrapolation and the allocation of 401(k) accounts to company stock, Journal of Finance 56, 1747–1764. Bhattacharya, Uptal, and Peter Groznik, 2002, Melting Pot or Salad Bowl: Some Evidence from U.S. Investments Abroad, Working paper, University of Indiana. Bohn, Henning, and Linda Tesar, 1996, U.S. Equity Investment in Foreign Markets: Portfolio Rebalancing or Return Chasing?, The American Economic Review 86, 77– 81. Brennan, Michael, and Henry Cao, 1997, International Portfolio Investment Flows, Journal of Finance 52, 1851–1880. Chari, Anusha, and Peter Blair Henry, 2001, Stock Market Liberalization and the Repricing of Systematic Risk, NBER Working paper 8265. Chari, Anusha, and Peter Blair Henry, 2002, Capital Account Liberalization: Allocative Efficiency or Animal Spirits?, NBER Working paper 8908. Choe, Hyuk, Bong-Chan Kho, and Ren´e Stulz, January 2001, Do Domestic investors have more valuable information about individual stocks than foreign investors?, National Bureau of Economic Research Working Paper. Cooper, Ian, and Evi Kaplanis, 1994, Home bias in equity portfolios, inflation hedging and international capital market equilibrium,, Review of Financial Studies 7, 45–60. Coval, Josh, and Tobias Moskowitz, 1999, Home Bias at Home: Local Equity Preference in Domestic Portfolios, Journal of Finance 54, 2045. Coval, Josh, and Tobias Moskowitz, 2001, The geography of investment: Informed trading and asset prices, Journal of Political Economy 109, 811–841. Dahlquist, Magnus, Lee Pinkowitz, Ren´e Stulz, and Rohan Williamson, 2003, Corporate Governance and the Home Bias, Journal of Financial and Quantitative Analysis, forthcoming. 27

Demirguc-Kunt, Asli, and Ross Levine, 2001, Financial Structure and Economic Growth: A Cross-Country Comparison of Banks, Markets, and Development,Cambridge, MA: MIT Press, chap. Bank Based and Market Based financial systems: cross country comparisons. Errunza, Vihang, Ked Hogan, and Mao-Wei Hung, 2000, Can the gains from international diversification be achieved without trading abroad?, Journal of Finance 54, 2075–2107. Feldstein, Martin, and Charles Horioka, 1980, Domestic Saving and International Capital Flows, Economic Journal 90, 314–329. Frederick, Jim, 2001, National Securities, Money 30, 31. French, Kenneth, and James Poterba, 1991, Investor Diversification and international equity markets, American Economic Review 81, 222–226. Froot, Kenneth, Paul O’Connell, and Mark Seasholes, 2001, The Portfolio flows of international investors, Journal of Financial Economics 59, 151–193. Gordon, Rupert, 2000, Modernity, Freedom and the State: Hegel’s Concept of Patriotism, Review of Politics 62. Griffin, John, Federico Nardari, and Ren´e Stulz, 2002, Daily Cross-border equity flows; pushed or pulled?, . Grinblatt, Mark, and Matti Keloharju, 2001, Distance, Language, and Culture Bias: The Role of Investor Sophistication, Journal of Finance 56, 1053–1073. Guiso, Luigi, Paola Sapienza, and Luigi Zingales, 2003, People’s Opium? Religion and Economic Attitudes, forthcoming in the Journal of Monetary Economics. Harvey, Cambell, 1995, The Risk Exposure of Emerging Equity Markets, World Bank Economic Review pp. 19–50. Heath, Chip, and Amos Tversky, 1991, Preference and Belief: Ambiguity and Competence in Choice Under Uncertainty, Journal of Risk and Uncertainty 4. Huberman, Gur, 2001, Familiarity Breeds Investment, Review of Financial Studies 14, 659–680. Kahneman, Daniel, and Dan Lovallo, 1993, Timid Choices and Bold Forecasts: A Cognitive Perspective on Risk Taking, Management Science 39. Karolyi, Andrew, and Ren´e Stulz, 2002, Are financial assets priced locally or globally?, Working paper. Langer, Ellen, 1975, The illusion of control, Journal of Personality and Social Psychology pp. 311–328. Langer, Gary, July 17 2001, Market Gyration Brings Jitters, ABCnews.com. Lewis, Karen, 1999, Trying to Explain Home Bias in Equities and Consumption, Journal of Economic Literature 37, 571–608. 28

Obstfeld, Maurice, and Kenneth Rogoff, 2000, The Six Major Puzzles in International Macroeconomics: Is there a Common Cause?, NBER Working Paper No. 7777 Cambridge (MA): NBER. Pastor, Lubos, 2000, Portfolio Selection and Asset Pricing Models, Journal of Finance pp. 179–219. Rowland, Patrick, and Linda Tesar, 2000, Multinationals and the gains from international diversification, Working paper. Serrat, Angel, 2001, A Dynamic Equilibrium Model of International Portfolio Holdings, Econometrica 69, 1467–1490. Strumpf, Koleman, 2003, Illegal Sports Bookmakers, University of North Carolina Chapel Hill Working Paper. Stulz, Ren´e, 1981, A model of international asset pricing, Journal of financial economics 9, 383–406. Suh, J.H., 2000, Home bias among institutional investors: a study of the economist quaterly portfolio poll, Working paper, SK Research Institute. Sulon, Bill, 2001, Savings Bonds Appeal to Patriotic Pennsylvania Investors, The Patriot-News November 20. Tesar, Linda, and Ingrid Werner, 1995, Home Bias and the Globalization of Securities Markets, Journal of International Money and Finance 14, 467–492. Uppal, Raman, 1993, A General Equilibrium Model Of International Portfolio Choice, Journal of Finance 48, 529–553. Warnock, Francis, 2001, Home Bias and High Turnover Reconsiered, Federal Reserve working paper number 702.

29

Appendix A: World Values Survey The survey is conducted and held by the Inter University Consortium for Political and Social Research at the University of Michigan. The surveys are designed to enable a cross-national comparison of values and norms on many topics. Adults 18 and over were interviewed face-to-face in 53 countries. We choose only individuals in the upper half of the income distribution of their country. The number of remaining respondents in our sample ranges between 96 and 2562 per country. The raw scores are originally 1 for high patriotism and 4 for low patriotism, but we subtract them from 5 in order for higher scores to denote higher patriotism. Other survey variables are subtracted from 5 as well to ease interpretation.

Variable Patriotism

Question “How proud are you to be [insert nationality]?”

Religion

“Please say, for each of the following, how important it is in your life. […]Would you say Religion is…”

Trust in Government

“I am going to name a number of organizations. For each one, could you tell me how much confidence you have in them: is it a great deal of confidence, quite a lot of confidence, not very much confidence or none at all? […] The government in [capital city]” “Now I am going to read off a list of voluntary organizations; for each one, could you tell me whether you are an active member, an inactive member or not a member of that type of organization? [….] Charitable organization.” “When you get together with your friends, would you say you discuss political matters frequently, occasionally or never?” “Here is a scale of incomes. We would like to know in what group your household is, counting all wages, salaries, pensions and other incomes that come in. Just give the letter of the group your household falls into, before taxes and other deductions.” “Of course, we all hope that there will not be another war, but if it were to come to that, would you be willing to fight for your country?”

Charity

Discuss politics

Income

Fight

30

Scale 1: Very proud 2: Quite proud 3: Not very proud 4: Not at all proud 4: Very important, 3: Rather important, 2: Not very important 1: Not important at all. 4: A great deal, 3: Quite a lot, 2: Not very much, 1: None at all.

3: Active member, 2: Inactive member, 1: Not a member.

3: Frequently 2: Occasionally 1: Never Income categories coded by decile for each society, with 1=lowest decile, 10=highest decile.

1: No 2: Yes

Appendix B: Data Sources Variable

Notation

Definition and Source

Flights per 1000 People

AIR

World Bank World Development Indicators Database. Count of number of air departures per person.

Bank vs. Market

BANK

Demirguc-Kunt and Levine (1999). Rankings are an average of ratios and range in our sample from negative 2 (market based) to 1 (bank based).

Correlation with the World

CORRWORLD

Datastream. Calculated using one lagged year of daily data.

Corruption

CORRUPT

Diversification Risk Reduction

DIV_BEN

Ethnoliguistic Fractionalization

ETHNOLING

Financial Risk

FINRISK

Foreign Equity Holdings

FEH

Foreign Equity Holdings – U.S.

FEH-US

Imports

IMPORTS

U.N. Comtrade Database. Imports divided by GDP.

Natural Distance

NDISTANCE

Great Circle. Average distance from a country to 141 other countries.

Newspaper Circulation

NEWSPAPER

Market Weighted Proximity Capital Account Liberalization

University of Goettingen and Transparency International: Internet Center for Corruption Research. Ranking range from 1 (not corrupt) to 10 (very corrupt). Datastream. Calculated as variance of country index minus covariance of the country index with the world market index. Easterly and Kraay (1999). Measure of linguistic and ethnic group dispersion ranging from 0 (not very diverse) to 100 (very diverse). Harvey: http://www.duke.edu/~charvey/ Index of financial sector risk ranging from 0 (low risk) to 100 (high risk). For 1995 only. IFS Database. Foreign equity holdings divided by total equity held by country residents, defined as country market capitalization plus residents’ foreign equity holdings minus foreigners’ equity holdings in the domestic market. Survey of Consumer Finance (1997). Foreign equity holdings for 9 U.S. Census regions divided by average region income.

World Bank World Development Indicators Database. Circulation of daily newspapers per 1000 people. Great Circle (distance) and Euromonitor (market capitalization). Variable PROXIMITY creating by averaging the difference between the furthest country distance and all other countries, each weighted by market capitalization. IMF Capital Account Restrictions from the Exchange Arrangements and RESTRICTIONS Exchange Restrictions Annual Reports.

Percentage of Rural Residents

RURAL

World Development Indicators. Percentage of population that live in rural areas.

Tariff and NonTariff Barriers

T_NTB

EIU Database. Barrier index ranging from 1 (few barriers) to 5 (many barriers).

Televisions

TV

EIU Database. Televisions per capita.

Violent Crime

CRIME

World Competitiveness Yearbook (1995). Murders, violent crimes and robberies per 100,000 people in 1993.

31

o o o

Switz_2 Belgium_1

NL_1

0.30

Foreign equity investment

o

Germany_2

o

UK_1

0.20 o

o o Germany_1

o

France_1

Italy_1 o

o

Sweden_2

Denmark_1

Australia_2 Sweden_1

0.10

o

USA_1 o

Latvia_2 o o Japan_2

Moldova_2 Lithuania_2 o o o Russia_2

0.00

2.50

o USA_2

o Canada_1

Finland_2 o o Peru_2 Slovenia_2 o Spain_2 o Chile_2 Poland_2 o o o o S. Africa_1 o o S. Africa_2 o Argentina_2 Turkey_2 Venezuela_2

Spain_1 o Finland_1 o

3.00

3.50

Patriotism score

Figure 1. Negative Relation Between Foreign Equity Holdings and Patriotism. The dashed line is the fitted regression line. Countries are shown with labels representing survey 1 (1992-1993) or survey 2 (1995-1997).

32

o

Germany

Change in Foreign Equity Holdings

0.100

o

Sweden

0.075

0.050

o

USA

o

Finland

0.025

o

o

0.000

-0.10

0.00

0.10

South Africa

Spain

0.20

Change in Patriotism

Figure 2. Negative Relation Between the Change in Foreign Equity Holdings and the Change in Patriotism. The changes are measured from the 1990-1992 survey to the 1995-1997 survey.

33

Α

Midwest: East North Central

0.08

Foreign Equity Holdings

Α

Northeast: New England

Α

0.06

Α

Northeast: Middle Atlantic

West: Pacific Division Southeast: South Central Α

Α

Α

West: Mountain Division

South: South Atlantic

0.04

Southwest: South Central

Α

0.02 Α

3.70

Midwest: North Central

3.75

3.80

3.85

Patriotism score

Figure 3. Negative Relation Between U.S. Regional Foreign Equity Holdings and Regional Patriotism Score The 9-level Census division codes are as follows: 1. 2. 3. 4. 5. 6. 7. 8. 9.

Northeast: New England Division (CT, ME, MA, NH, RI, VT) Northeast: Middle Atlantic Division (NY, NJ, PA) South: South Atlantic Division: (DE, DC, FL, GA, MD, NC, SC, VA, WV) South: East South Central Division: (AL, KY, MS, TN) South: West South Central Division: (AR, LA, OK TX) Midwest: East North Central Division (IL, IN, MI OH WI) Midwest: West North Central Division (IA, KS, MN, MO,NE, ND, SD) West: Mountain Division (AZ, CO, ID, MT, NV, UT, WY, NM) West: Pacific Division (AK, CA, HI, OR, WA)

34

Table I Patriotism Scores Across Countries The Patriotism Score refers to the average answer for high income residents of a country to the question: "Are you proud to be [insert nationality]?".

First Wave: 1990-1992 Patriotism High Income Score Observations Country Germany 2.70 441 Netherlands 2.87 394 2.95 699 Belgium France 3.06 287 3.10 441 Italy Sweden 3.17 401 Spain 3.18 994 Finland 3.19 481 3.23 440 Denmark Canada 3.51 895 Slovenia 3.51 186 South Africa 3.52 1096 3.72 524 USA Average 3.21 560 Second Wave: 1995-1997 Patriotism High Income Score Observations Country 2.53 527 Germany Latvia 2.77 457 Lithuania 2.80 646 Japan 2.82 349 Russia 2.89 844 2.97 168 Moldova Switzerland 2.98 708 Finland 3.39 284 Argentina 3.33 316 Sweden 3.33 465 Spain 3.41 474 Chile 3.46 491 3.46 488 Turkey Poland 3.64 25 Australia 3.69 803 Peru 3.79 134 3.79 858 USA South Africa 3.80 2562 Venezuela 3.89 96 Average 3.30 563

35

36

Survey Dum Obs R² Hausman p-value

CorrWorld

Newspapers

Restrictions

Patriotism

IV-rural 2

Y 33 0.22 0.17

Y 33 0.29

-0.109 ** -0.048 *** (2.23) (2.84)

OLS 1

Y 33 0.37

Y 33 0.45

-0.115 *** -0.105 *** (3.64) (3.48)

-0.087 ** (2.06)

Y 33 0.19

0.216 * (1.85)

Y 33 0.27

0.161 (1.38)

-0.091 * (1.84)

-0.107 ** (2.48)

OLS 8

Y 28 0.37

Y 28 0.50

0.273 *** 0.252 *** (2.97) (3.00)

Dependent Variable: Foreign Equity Holdings OLS OLS OLS OLS OLS 3 4 5 6 7

IV-rural 10

Y 28 0.63

0.124 (1.38)

0.013 (0.15)

0.03

Y 28 0.62

0.070 0.08

0.043 (0.50)

-0.084 ** -0.062 ** (2.73) (1.91)

-0.106 ** -0.051 ** (2.58) (2.47)

OLS 9

OLS and IV estimates are reported for regressions of foreign equity holdings on the patriotism score and the standard explanations for home bias. Foreign holdings is the percentage of a country's total equity holdings that is foreign, calculated from IFS data. Patriotism is from the World Values Survey and is an increasing index from 1 to 4. Restrictions is an IMF indicator variable of Capital Account Restrictions. Newspapers is the amount of daily circulation per 1,000 people from World Development Indicators. CorrWorld is a calculated variable of the correlation of a country's market index with the rest of the world; CorrWorld is calculated from daily returns for 1 year prior to the survey date with Datastream data. In columns 2 and 10, Patriotism is instrumented by the percentage of rural population. The Hausman p-values result from the test that the OLS estimator is a consistent and efficient estimator of the true parameter. The null hypothesis is that the IV coefficient on patriotism is equal to the coefficient on the instrumented patriotism varaible. Significance at the 10\%, 5\% and 1\% levels are indicated by *, ** and *** respectively. T-statistics are reported in parentheses.

Table II Estimations Showing Patriotism's Significance in Explaining the Home Bias

37

Variable Patriotism Restrictions Newspapers CorrWorld

Change in Foreign Equity Holdings for 1 Change in R² when Standard Deviation Variable Added to Full Estimated Coefficient Change in Variable Model -0.116 -0.039 0.124 -0.086 -0.041 0.128 0.122 0.020 0.008 0.194 0.039 0.037

Measures of economic significance are explored in columns two and three. Column two reports the change in the percentage of equity holdings that are foreign for a one standard deviation change in the row variable. The final column shows the change in R-Squared that occurs when the row variable is added to the full model of foreign equity holdings on the survey year dummy, patiotism, transaction restrictions, newspapers per capita, the correlation of the country market with the world, and GDP per capita. The full model is defined to be all of the variables except the one that is added for calculating the change in R-Squared.

Table III Economic Significance of Patriotism and Standard Explanations

38

Patriotism GDP/Cap Inequality Patriotism 1 GDP/Cap -0.25 ** 1 Inequality 0.40 *** -0.25 * 1 Crime 0.41 *** -0.19 -0.03 Corruption 0.26 * -0.83 *** 0.51 *** Discuss -0.32 *** -0.01 -0.36 *** Trust Gov 0.18 0.11 0.07 Ethnoling 0.43 *** -0.35 ** 0.39 *** Charity 0.23 * 0.31 ** 0.25 * Religion 0.67 *** -0.52 *** 0.61 *** 1 0.05 -0.02 0.24 0.63 0.41 ** 0.44 ***

Crime

1 -0.16 -0.20 0.33 ** -0.19 0.59 ***

Corruption

1 -0.17 0.03 -0.02 -0.37 ***

Discuss

1 0.23 -0.12 0.18

Trust Gov

1 0.10 0.56 ***

Ethnoling

1 0.18

Charity

Partial correlations, controlling for survey year, are presented for economic, political, and social correlates with patriotism. Economic correlates are income and inequality. GDP/Cap is from the World Development Indicators. Inequality is the Theil Index from the Univerity of Texas at Austin. Political correlates are crime, corruption, discuss politics, and trust of government. Crime is violent crimes per 100,000 people in 1993 from the World Competitiveness yearbook. Corruption is the Tranparency International Index increasing in corruption form 1 to 10. Discuss refers to the World Values Survey question of how often an individual discusses politics. Trust of government is the response to the world values survey of whether an individual trusts the government. Social correlates are entholinguistic dispersion, charity and religion. Ethnolinguistic dispersion is an index from Easterly and Kraay (1999). Charity is the response to the World Values Survey question of charitable activity. Religion is the response to the World Values Survey of how religious an individual rates herself. Significance in the partial correlations is denoted by *, **, and *** for the 10%, 5% and 1% levels respectively.

Table IV Partial Correlations, Controlling for Survey Year, of Patriotism and Economic, Political and Social Variables

Table V Regressions Showing the Patriotism Result is Not Driven Solely by Correlates OLS estimates are reported for regressions of foreign equity holdings on the patriotism score and significant patriotism correlates. Foreign holdings is the percentage of a country's total equity holdings that is foreign, calculated from IFS data. Patriotism is from the World Values Survey and is an increasing index from 1 to 4. Restrictions is an IMF indicator variable of Capital Account Restrictions. Newspapers is the amount of daily circulation per 1,000 people from World Development Indicators. CorrWorld is a calculated variable of the correlation of a country's market index with the rest of the world; CorrWolrd is calculated from daily returns for 1 year prior to the survey date with Datastream data. GDP/Cap is GDP (in millions) per capita from World Development Indicators. Crime isviolent crimes per 100,000 people in 1993 from the World Competitiveness yearbook. Corruption is the Tranparency International Index increasing in corruption form 1 to 10. Discuss refers to the World Values Survey question of how often an individual discusses politics. Enthnoling is the enthnoliguistic dispersion from Easterly and Kraay (1999). Charity and Religion are World Values Survey questions asking whether the individual participates in charitable and religious activity. Significance at the 10\%, 5\% and 1\% levels are indicated by *, ** and *** respectively. T-statistics are reported in parentheses. 1

2

3

4

5

6

7

8

Patriotism

-0.115 ** (2.67)

-0.092 * (1.79)

-0.122 ** (2.41)

-0.116 ** (2.75)

-0.112 ** (2.66)

-0.134 ** (2.39)

-0.165 *** (3.62)

-0.133 ** (2.15)

Restrictions

-0.087 ** (2.77)

-0.079 ** (2.45)

-0.083 ** (2.71)

-0.073 ** (2.19)

-0.097 ** (2.77)

-0.088 ** (2.69)

-0.079 ** (2.71)

-0.087 ** (2.75)

Newspaper

0.1284 (0.73)

-0.045 (0.43)

-0.083 (0.90)

-0.102 (0.85)

0.001 (0.02)

0.042 (0.38)

0.031 (0.36)

0.060 (0.50)

CorrWorld

0.1924 (1.50)

-0.126 (1.36)

-0.050 (0.47)

-0.090 (0.97)

0.103 (1.09)

0.103 (0.99)

0.008 0.09

0.127 (1.40)

GDP/Cap

-2.730 (0.76) 0.002 (0.47)

Inequality

0.277 (1.05)

Crime

-0.015 (1.42)

Corruption

-0.075 (0.78)

Discuss

0.483 (0.59)

Enthnoling

0.274 ** (2.39)

Charity Religion Survey Dum Obs R2

Y 28

Y 27

Y 25

Y 27

Y 28

Y 26

Y 26

0.64

0.60

0.67

0.66

0.64

0.62

0.70

39

0.033 (0.59) Y 28 0.63

Table VI Robustness Regressions: Patriotism's Significance with Other Measures of Home Bias Explanations OLS estimates are reported for regressions of foreign equity holdings on patriotism and alternative measures of transaction barriers, information and risk. Foreign holdings is the percentage of a country's total equity holdings that is foreign, calculated from IFS data. Patriotism is from the World Values Survey and is an increasing index from 1 to 4. T_NTB is an index of tariff and nontariff barriers from the EIU. Televisions per capita is also from EIU. AirDepart is the number of flight departures per person from World Development Indicators. Proximity is a average market weighted distance to all other countries. The natural distance measure is from Great Circle. DiveBene is diversification benefits calculated as varaince of a country index minus the covaraince with the world market index. Sharpe is the country index Sharpe ratio calculated from Datastream data. Finrisk is Cambel Harvey's index of financial sector risk. Significance at the 10\%, 5\% and 1\% levels are indicated by *, ** and *** respectively. T-statistics are reported in parentheses.

1 Patriotism Barriers T_NTB Information TV/Cap

2

3

4

5

6

7

-0.148 *** -0.097 * -0.136 *** -0.101 * -0.106 * -0.125 ** -0.108 ** -0.144 ** (2.89) (1.96) (3.05) (1.74) (1.88) (2.61) (2.17) (2.57) 0.051 (2.04) * 0.206 * (1.79) 0.005 *** (2.96)

AirDepart

0.168 (0.29)

Proximity

0.001 (0.12)

NDistance Risk DivBene

-5.770 (1.56) -0.168 (0.17)

Sharpe

-0.005 * (1.75)

FinRisk Survey Dum R2 Obs

8

Y 0.39 29

Y 0.31 32

Y 0.40 33

Y 0.22 33

40

Y 0.22 33

Y 0.38 28

Y 0.22 33

Y 0.42 33

Table VII Patriotism and the Role of Corporate Governance for Foreign Holdings OLS estimates are reported for regressions of foreign equity holdings on the patriotism score and corporate governance measures. Foreign holdings is the percentage of a country's total equity holdings that is foreign, calculated from IFS data. Patriotism is from the World Values Survey and is an increasing index from 1 to 4. Closely is the average percentage of closely held shares for all Worldscope country firms minus the average closely held percentage from the rest of the world. Investor protection and expropriation risk are from EIU; both of these measures are increasing in better protection.

Dependent Variable: Foreign Equity Holdings 1 2 3 4 5 -0.228 *** (3.22)

Patriotism Closely

0.0899 (0.04)

-0.141 ** (2.74)

0.0393 ** (2.22)

0.0341 ** (2.14)

Expropriate Y 22 0.06

-0.155 *** (2.99)

-1.811 (0.97)

InvestProt

Survey Dum Obs R²

6

Y 22 0.40

Y 29 0.21

41

Y 29 0.39

0.072 (1.61)

0.074 * (1.88)

Y 29 0.15

Y 29 0.37