There is much evidence of the existence of home bias and local bias. The degree of home bias can be measured by various approaches for numerous countries. There is an observable variation in the extent of home bias in the course of time. Besides, induced by some factors the degree varies as well.
4.1 Empirical evidence and measures on the existence of home bias and local bias
Evidence on the existence of the home bias can be observed in most countries worldwide. French and Poterba (
1991), Cooper and Kaplanis (
1994), Tesar and Werner (
1995) report evidence for OECD countries and Stockman and Dellas (
1989) and Dziuda and Mondria (
2012) for additional countries. Oehler et al. (
2008) confirm a significant home bias of German mutual fund investors and even a European home bias. German mutual funds not only hold a more-than-optimal share of German assets but also hold higher-than-optimal weights of other European countries’ assets compared to the world market portfolio. Lütje and Menkhoff (
2007) also prove the existence of the home bias specifically for German investors, Dahlquist (
2001) for investments in Sweden. Lippi (
2016) confirms home bias for Italian professional occupational pension fund managers investing in government securities, corporate bonds and equities.
Relating to different asset types, the existence of a home bias has been underlined for equities (Bradshaw et al.
2004; Diyarbakirlioglu
2011; Tesar and Werner
1995), bonds (Fidora et al.
2007; Solnik and Zuo
2016; Tse
1999; Ferreira and Miguel
2011; Tesar and Werner
1995), real estate (Imazeki and Gallimore
2009; Eichholtz et al.
2001) and mutual funds (Coval and Moskowitz
1999; Giannetti and Laeven
2016; Lütje and Menkhoff
2007; Oehler et al.
2008).
There is also much evidence of the existence of the intra-national local bias. Coval and Moskowitz (
1999,
2001) document a local bias for mutual funds. Amongst others, Ivkovic and Weisbenner (
2005), Seasholes and Zhu (
2010), Huberman (
2001) as well as Hong et al. (
2005) confirm the existence of a local bias of individual U.S. investors. Grinblatt and Keloharju (
2001) prove a significant local bias of Finish investors, Kang and Stulz (
1997) for Japanese. For German individual investors, Baltzer et al. (
2013) also document a local bias. Parwada (
2008) examines the location and portfolio choice of investment start-ups. The degree of start-ups’ local bias is three times higher than the local bias extent reported by Coval and Moskowitz (
2001). Pool et al. (
2012) show that mutual fund managers in the U.S. overweight their home states where they come from. The degree of local bias is hardly measurable and measurements are even worse to compare to each other because local (or regional) is not a clearly defined area, in particular compared to home bias for which the borders (of a country) are clearly determined.
4 Additionally, it is hard to determine a comparing, well-diversified portfolio which is crucial to measuring the degree of the bias.
In contrary, the home bias can be measured much better. In their basic work on the home bias, French and Poterba (
1991) show the degree of home bias for three countries based on data from the end of 1989. U.S. investors invest 93.8% domestically, Japanese 98.11% and UK investors just 82% of their equity portfolios. The lower level of British domestic investments was due to “Prime Minister Thatcher’s relaxation of capital control” (French and Poterba
1991, p. 223). These figures underline that there is a home bias, but the figures do not take into account the optimal weight of every country in a well-diversified portfolio. Also with data from the end of the 1980s, Cooper and Kaplanis (
1994, p. 46) support the results of French and Poterba (
1991) but calculate a better comparable measure of the home bias by calculating the “domestic equities relative to the proportion of domestic equities in the world market portfolio”: US 98%, UK 78.5%, Japan 86.7%, Germany 75.4%, France 64.4% and Sweden even 100%. Still, the home bias remains significant and strong.
Nowadays,
5 home bias is measured with numerous approaches (some just slightly different to others). Table
1 gives an overview about the results of home bias measurements of selected countries which are explained briefly in the following section. One way to measure it, is to set the share of domestic assets in relation to the ‘optimal’ CAPM share of domestic assets. The difference of weights of domestic assets (in the actual and ‘optimal’ portfolio) is then considered as a measure of home bias. The CAPM home bias is defined according to Morse and Shive (
2011, p. 418):
$$CAPM\ Home\ Bias\; \% = domestic\ holdings\; \%-\frac{home\ capitalization}{world\ capitalization}$$
Table 1
Different measurements of home bias for selected countries
U.S. | 93.8 | 98.0 | 0.61 | 0.695 | 0.730 | 0.3881 | 75.1 | 92.1 | 42.21 | 0.82 | 0.6118 |
UK | 82.00 | 78.5 | 1.67 | 1.714 | 0.380 | 0.3493 | 67.1 | 65.4 | 20.75 | 0.73 | 0.5629 |
Japan | 98.11 | 86.7 | 1.86 | 2.363 | 0.682 | 0.6053 | 89.3 | – | 14.83 | 0.91 | 0.7916 |
Germany | – | 75.4 | 2.12 | 2.171 | 0.307 | 0.295 | 61.6 | 55.4 | 16.40 | 0.7 | 0.4609 |
France | – | 64.4 | 2.55 | 2.651 | 0.533 | 0.5095 | 72.4 | 55.4 | 32.42 | 0.82 | 0.6345 |
Italy | – | 91.0 | 2.77 | 3.028 | 0.339 | 0.3315 | 57.3 | 55.4 | 16.63 | 0.83 | 0.3902 |
Sweden | – | 100.0 | 3.81 | 3.927 | 0.472 | 0.4571 | – | – | 41.43 | 0.77 | 0.5035 |
Hau and Rey (
2008, p. 335) “estimate total investment in the domestic market by domestic agents, … then simply divide it by total domestic market capitalization”. The data is not normalized by the relation of the domestic capitalization to the world capitalization. Fidora et al. (
2007) give comprehensive data on the degree of home bias which in contrast to Hau and Rey (
2008) is related to the share of the world capitalization based on the formula:
$$Home\ Bias\ of\ Country\ i=1-\frac{w_i}{w_i^{\ast }}$$
Fidora et al. (
2007, p. 635) define
wi as the “share of international assets in the country’s portfolio” and
\({w}_i^{\ast }\) as the “market weight of the rest of the world seen from the viewpoint of a given country
i”. Mature economies (e. g. the U.S., the UK, Germany, Japan etc.) exhibit, on average, a home bias of 67.6%. Emerging economies (e. g. in Asia and Latin America) show a significant higher degree of around 95%.
Chan et al. (
2005) apply a resembling measure, but express the home bias as a natural logarithm. Lau et al. (
2010) calculate their home bias measure exactly the same way as Chan et al. (
2005). Since the measurement of Lau et al. (
2010) is based on data over a longer period of time (from 1998 to 2007), they obtain slightly different results. Anderson et al. (
2011) approach the calculation similar to the general definition of Chan et al. (
2005), but distinct in two important aspects: First, Anderson et al. (
2011) perform a subtraction and second they do not express the results logarithmically. Both factors cause the very different and not comparable measures of Chan et al. (
2005) and Anderson et al. (
2011). Mondria and Wu (
2010) use the same definition of home bias as Ahearne et al. (
2004). Mondria and Wu (
2010) define home bias as ‘one minus the ratio share of “foreign equities in country i’s portfolio” and “the share of foreign equities in the world portfolio” from perspective of country I’. Going more into detail regarding the measurement approaches, there are different ways on how to build up the optimal weight of a country of a portfolio. Mishra (
2015) shows different approaches and measures of home bias based on different optimal portfolios. A comprehensive measure of both home bias and foreign bias can be found in Cooper et al. (
2018) who integrate home and foreign bias in one model and then measure so-called pure home bias relative to the model. Pure home bias is just the part of home bias which cannot be explained by foreign bias and distance effects. Thus, it is not a measure of home bias as considered by the large part of authors (and in this review) and the results are therefore not included in the comparison of home bias measurements in this review. However, the model is very compelling and seems a promising approach different to the large part of existing studies. Cooper et al. (
2018) find that pure home bias can just be observed in emerging markets. For developed countries foreign bias can explain the large part of total home bias variation, i. e. “the home country is very much like a foreign country with zero distance. Investors do not appear to exhibit a pure fear of foreign investment separate from their general dislike of distance” (Cooper et al.
2018).
As summarized in Table
1, the measures are quite different. All papers confirm the existence of home bias for all countries although no consistent and standardized measure is applied. That is why the results sometimes show inverse directions. For example, Chan et al. (
2005) measure a larger home bias of the UK compared to the U.S., whereas the findings of Fidora et al. (
2007) show the contrary, even when neglecting the logarithmic presentation and even though the data is about a similar period of time. The consistency and accuracy of data and measurements is only guaranteed within one specific study and within one specific method of measurement. There are some papers stating that mismeasurement of the home bias leads to its existence. For example, Lewis (
1999) states that the used mathematical models may be the only reason that there is a home bias. But, since all of the presented studies provide evidence for the existence with different models, the effect of mismeasurement when dealing with the pure existence seems to be marginal.
4.2 Empirical measuring the varying degree of home bias and local bias
The degree of home bias varies by two points of view, a general decline in home bias in the course of time and relating to particular factors which impact the extent of home bias. Since most of the available data is provided to U.S. investors (Karolyi
2016), the degree of home bias for U.S. investors is best analysed in empirical research (Eichler
2012). For local bias, there is no disposable data.
Since the percentage of foreign ownership at the Japanese stock market from the 1970s to the 1990s has increased, it can be considered as an indication of a general decline of home bias on course of time (Kang and Stulz
1997). Explicitly measured decline by Levy and Levy (
2014) shows a decrease of U.S. home bias from 1988 until the 2000s. This finding is confirmed by Ahearne et al. (
2004). After the early 2000s, home bias first slightly increased but has fallen again until 2012 and remains on a significant level around 40%. Support of the decline of home bias for other countries comes from Fidora et al. (
2007). Both equity and bond home bias in mature markets have decreased during 1997 to 2003. Unfortunately, more recent data on the degree of home bias for other countries than the U.S. could not be found in any considered paper.
The home bias also varies due to particular factors/variables. The economic respectively financial development of a country is one factor, however most studies show that there is no statistically significant correlation and impact. Bae et al. (
2008) exclude economic development of a country as a driving force for the equity home bias. Also, Dahlquist et al. (
2003) challenge the influence and importance of the financial development on the equity home bias, as differences in financial development will be reflected in stock prices. Chan et al. (
2005) support these findings and do not find a significant impact of economic development. They detect that merely the stock market development and familiarity have a statistically significant influence on the extent of home bias. Imazeki and Gallimore (
2009) use the same approach as Chan et al. (
2005) but examine real estate mutual funds and report similar results. The only significant factor which influences the degree of home bias in real estate is a combination of two variables: real estate market capitalization size and real estate market transparency. According to Imazeki and Gallimore (
2009) general economic development also seems to not be important when studying real estate home bias. However, there is evidence for an influence of the variable ‘economic development’ with respect to home bias in bonds (Ferreira and Miguel
2011).
In accordance with Pool et al. (
2012) resource-constraints of managers influence the degree of home bias. Managers with more limited resources exhibit more home bias. Investors with a small amount of invested money are more inclined to exhibit home bias (Karlsson and Nordén
2007). The effect of the size can be also transferred to the countries’ size, i.e. the size has a positive impact on home bias as in a big country an investor has more opportunities to diversify his portfolio and is not dependent on diversifying internationally (Mishra
2015).
Anderson et al. (
2011) examine the influence of culture on home bias and show that high values of the variables long-term orientation and masculinity lead to a relative decrease in the level of home bias, whereas uncertainty avoidance as a cultural characteristic increases home bias. The influence of gender is also proven by Karlsson and Nordén (
2007) who show that overconfident investors (mostly men) are more probable to show home bias. Lütje and Menkhoff (
2007) also underline the influence of overconfidence on home bias.
6 The impact of cultural variables on investment decisions is confirmed by Beugelsdijk and Frijns (
2010), though only for foreign bias.
Employees in the public sector (having a high job security) acting as investors and investors with a low education/sophistication are more inclined of being home-biased (Karlsson and Nordén
2007). According to Mondria and Wu (
2010) home bias decreases with financial openness but remains in the long run due to interaction between “portfolio and information choices” (Mondria and Wu
2010). Banks just like institutional ‘investors’ also exhibit an information-based home bias when they give loans to enterprises (Presbitero et al.
2014). Banks even exhibit home bias when allocating their own bank assets and do not diversify internationally with the help of international subsidiaries (García-Herrero and Vázquez
2013).
Shapira and Venezia (
2001) examine differences in (behavioural) patterns of institutional and individual investors and show that professional investors have a better diversified portfolio (less home bias) than individual investors. Sometimes the individual investors influence the institutional one. That means that the institutional investor is the one who actually invests but has to consider the ‘wishes’ of the individuals, e. g. considering mutual fund investing (Oehler et al.
2008). Lütje and Menkhoff (
2007) for home bias and Ivkovic and Weisbenner (
2005) for local bias provide similar evidence in favour of a higher bias of individual investors.