A view on pension funds’ domestic equity holdings: Investment behavior, home bias, and determinants
Tjarco van Olphen S2356538
University of Groningen Faculty of Economics and Business
MSc Finance Thesis
Supervisor Prof. Dr. R.M. Salomons
January 2018
Abstract
Dutch pension funds are criticized for their lack of domestic investments, while studies show that pension funds prefer domestic equities. This paper tries to verify whether this criticism can be justified from an economic point of view. Using a panel data set between 2007 and 2016, this paper shows that Dutch pension funds hold too much domestic equities in their portfolios, but significant less than pension funds from other countries. Furthermore, it demonstrates that the size of the investments, the pension plan, and the euro influence the home bias of pension funds, all of which, can explain the low home bias for Dutch pension funds. Therefore, the criticism cannot be justified. Key words: pension funds, home bias, asset allocation, equity holdings, and
1. Introduction
The expression “do not put all your eggs into one basket” is a common one; nevertheless, investors have a tendency to put their eggs into one basket: the domestic equity market. Their strong preference for domestic equity in international markets, despite the well-documented benefits of international diversification, remains an important yet unresolved empirical puzzle in financial economics. This preference for domestic equities is called home bias (Coval and Moskowitz, 1999).
Pension funds are the main form of investment in these equity markets. They are especially important to the stock market, where large institutional investors dominate. In today’s world, there is increasing attention on pension funds, which are becoming larger and therefore more influential in the market. Furthermore, they are currently experiencing multiple problems, such as their slow recovery from the economic crisis, their expanding life expectancy, and the decrease in the labor force relative to the retirees. In addition, the investment behavior of pension funds is not only about generating the highest risk with the lowest return, but also about being accountable (Plender, 1983). Despite this increase in attention, research indicates that there is a decreasing but existing home bias in the investment decisions of Dutch pension funds (Rubbaniy, Lelyveld, and Verschoor, 2010). This is remarkable because Dutch pension funds have been at or near the top of the global pension rankings for years.
The Dutch minister of financial affairs argues that after the publication of the statistics of the institutional investments in the Netherlands (NLII), Dutch pension funds no longer invest enough in their own economy and mentions the need for pension funds to make more domestic investments. The statistics of the NLII reveal that Dutch pension funds invest 12.5% of their total investments in the Dutch market. It is not the first time that the government criticizes these funds for their lack of domestic investments.1 This criticism contradicts the existing literature, which suggests that equity home bias exists among pension funds (see, e.g., Rubbaniy et al., 2010; Lippi, 2016); however, quality studies on the equity allocation of pension funds are limited. This arises the following question: do Dutch pension funds invest too less in their domestic equities compared to pension funds from other countries, and if they do, why?
The answer to this question can verify whether the criticism of pension funds for their lack of domestic investment is justified from an economic point of view. To obtain valuable results, this research uses data created by De Nederlandse Bank (DNB), the Coordinated Portfolio Investment Survey (CPIS), and the annual reports of pension fund associations that contain the equity holdings of these funds and other investors. In this paper, I believe that I am the first to present facts about the home bias for pension funds at a macroeconomic level. In doing so, I have two goals: to present the equity investment behavior of pension funds and to provide an economic explanation for this behavior.
First, this paper examines the investment behavior of pension funds by exploring the equity home bias of Dutch pension funds and comparing it to those in other countries and to the home bias of the general investor. The results could help to distinguish pension funds from different countries as well as the general investor. Furthermore, since pension funds are one of the largest investors worldwide, these
1 The Dutch version of the financial times has two articles about the domestic investments of Pension
results reflect not only the national investment behavior but probably the general worldwide equity investment behavior as well.
Second, it shows the rational explanations for the home bias of general investors and pension funds, and it adds an extra determinant for the home bias of pension funds to the current theory: the pension plan. The results provide a macroeconomic explanation for the equity investment behaviors of pension funds, which in turn offers robust evidence for the explanation of the home bias at the pension fund level and contribute to the explanation of the unresolved empirical puzzle of home bias.
This paper is structured as follows. Section 2 discusses the background literature concerning home bias and the possible explanations for it. Section 3 describes the methods this study uses to answer the research question and elaborates on the explanatory variables used in this study. Section 4 describes an overview of the data descriptive statistics and the results are subsequently discussed in Section 5. The final section concludes the paper.
2. Literature Review
This section provides the theoretical background of this study. First, the existence of home bias is argued. Second, the possible economic explanations for home bias are discussed.
2.1. Home bias
Early on, the benefits of international diversification have been identified by Grubel (1968); Solnik (1974), Eldor, Pines, and Schwartz (1988); De Santis and Gerard (1997); and others. However, these benefits are becoming smaller. Research from You and Daigler (2010) demonstrate that correlations between international countries have begun to increase, with only Asian markets providing lower correlations relative to American and European markets. Research from Levy and Levy (2014) links these correlations to home bias. They argue that the easing of regulations on foreign investments makes them more attractive to investors. However, due to increases in foreign investments, correlations are increasing, and diversification benefits decrease. This, in turn, decreases the motivation to continue increasing foreign investments.
In spite of the proven benefits of diversification, many researchers report home bias in investment behavior. Home bias is first showed by French and Poterba (1991), who demonstrate that the degree of domestic equity holdings of Japan, the United States, the United Kingdom, France, and Germany are 95.7%, 92.2%, 92%, 89.4%, and 79% respectively for holdings at the end of 1989. More recent research by Sørensen, Wu, Yosha, and Zhu (2007) measures home bias by the share of the country in the total world equity market. They normalize equity home bias to be 1 if a country invests 100% domestically and -1 if they invest 100% in foreign equity. They find the average equity home bias among developed countries to be 0.83 in 1993 and 0.67 in 2003, while the home bias for the Netherlands tends to be the lowest at 0.37. Baele, Pungulescu, and Ter Horst (2007) estimate home bias using several estimation methods, and their results illustrate an average home bias between 0.42 and 0.8 for investors.
Second, as the costs of diversification decrease, the correlation with foreign markets will increase, making it less attractive to invest in them.
All these studies focus on home bias at a country level; in this sense, they are all theories of aggregate home bias. The paper by Hau and Rey (2008) is the first to present facts about home bias at a fund level. They find that a typical equity fund exhibits a much lower degree of home bias, as in the aggregate data. Chan, Covrig, and NG (2005) find robust evidence that mutual funds allocate a disproportionately larger fraction of investment to domestic stocks. More interestingly, according to them, different countries exhibit different levels of home bias.
The literature shows that home bias changes over time and differs between investors as well as between countries. While there has been much research on home bias, the topic of home bias in pension funds is still largely unexplored. There are only two researches from a noteworthy journal that discuss the home bias in pension funds. First, Rubbaniy, Lelyveld, and Verschoor (2010) find a positive but rapidly decreasing home bias in Dutch pension funds in the period from 1992 to 2006. Their results demonstrate that equity home bias is decreasing, and that bond home bias is increasing. Second, Lippi (2016) has conducted research on Italian pension funds, and his obtained results show that home bias is a phenomenon that can that can be argued to exist in Italian occupational pension funds.
2.2. Economic explanations for home bias
Nevertheless, home bias seems to be grossly inefficient from a diversification point of view, academics have offered many explanations for this phenomenon. Researches by Stulz (1981) and Black (1984) link governmental restrictions to the overweighting of domestic assets in international portfolios. Even though many of these restrictions have decreased or disappeared, investors still have large amounts of their assets in domestic markets. This explanation could have been viable 30 years ago; however, it is not relevant in the globally integrated market of today (Tesar and Werner, 1995; Ahearne, Griever and Warnock, 2004). Therefore, other explanations for home bias have been examined.
An alternative explanation argues that investors have superior access to information about the domestic economy. Research by Low (1993), Coval (1996), and Brennan and Cao (1997) show this asymmetric information-based explanation of international capital market segmentation. A main critique on this information-based explanation is that investorsare free to choose to learn about foreign firms and therefore the initial information advantage can disappear; however, the research of Veldkamp and van Nieuwenburgh (2009) indicate that investors themselves choose to have different information sets. The investors believe they have superior access to information about their domestic markets, even though this is not the case. Therefore, their results demonstrate that the asymmetric information-based explanation is a behavioral explanation instead of an economic explanation for home bias.
particular as an explanation for the differences in home bias across financial assets. They use a Markowitz-type portfolio to measure home bias in which real exchange rate volatility causes a bias towards domestic financial assets. The findings show that this volatility is a driver for home bias and that when real exchange rate volatility is reduced to zero, the home bias decreases by 20 percentage points.
Common currencies, such as the euro, eliminate the real exchange rate risk with countries in the European Monetary Union (EMU). These countries have the same currency; therefore, exchange rates do not deviate from one another. After the introduction of the euro in 1999, European financial markets integrated considerably, and home bias decreased in European countries (Chan and Covrig, 2005; Balli, Basher and Ozer-Balli, 2012). However, this created a new bias called euro bias, a situation where euro investors tend to overweight their investments in European assets (Balli et al., 2010; Schoenmaker and Bosch, 2010). The EMU also creates unification benefits. Before the EMU, one paid dividend tax, for example, for an Italian company, and he or she could not claim it back; for the investor, this was simply a cost item. Since, dividend tax is deductible or recoverable in domestic markets the unification leads to fiscal benefits for the investor.
There are also possibilities of circumventing the real exchange rate risk. Cole and Obstfeld (1991) argue that it is possible to hedge against the real exchange rate risk by terms of trade. According to them, this is even possible in financial autarkies. Heathcote and Perri (2007) extend the argument to an open economy with only two goods of production, and they find that there are international diversification possibilities within the national economy. Nowadays, the most common way in which to implement these international diversification methods is by investing in multinational firms, which, to a large degree, depend on revenues from foreign activities. For example, even though Philips is a listed company on the AEX, it is a worldwide operating company; therefore, the Dutch economy is only accountable for a small part of the company’s revenues. This generates ‘foreign’ diversification benefits for investors while they invest in their domestic equity markets. Mathur and Hanagan (1983) find that even when investors do not trade in foreign assets, they do reap foreign diversification benefits by investing in MNOs. The conventional home bias measurement does not include these diversification probabilities. In the paper of Ghazalian and Furtans (2008), they demonstrate that disregarding the activities of multinational enterprises leads to an upward bias in the measurement of home bias. The benchmark results indicate that home bias is overstated by a factor of 1.7 when multinational activities are disregarded.
One of the arguments for investing in foreign equities is to hedge human capital, since returns on both labor and domestic equity are positively correlated with each other. Baxter and Hermann (1997) show that a substantial short position in domestic marketable assets is consequently required to hedge the human capital risk. This argument is originally defined by Brainard and Tobin (1992): “If the returns of the home country are more correlated with the human capital than the returns of foreign countries, then the investments should be more skewed toward foreign markets.” Therefore, in countries where the labor share is a larger part of the national income, international diversification offers more benefits.
holdings associated with portfolio size. Amadi (2004) finds a decrease in foreign investments when the market capitalization of the country increases because of an increase in investment availability. In addition, at a certain moment, investors are simply too large for their domestic economies; the total assets of some investors outgrow their domestic market capitalization, and they must invest in foreign equity markets. Furthermore, economies of scale have benefits because of relatively smaller transaction costs, more information, and stronger negotiation positions.
I add an extra possible explanation to the existing literature of determinants for home bias among pension funds: pension funds can have a defined contribution (DC) or a defined benefit (DB) pension plan2. In today’s world, pension plans are shifting from DBs towards DCs. The main differences between these two plans are the decision makers behind the investment choice. In a DC pension plan, the employee has (limited) decision making in the investment choice, whereas in a DB pension plan, in most cases, the pension fund or the employer makes the investment decisions. The John Hancock financial service (JFHS) clearly demonstrates a lack of even the most basic financial literacy among the general public. To exemplify, in its survey, the JHFS finds that the majority of the respondents think that their own company stocks are safer than a diversified portfolio. Benartzi and Thaler (2001) show that some employees spread their contributions evenly across the investment options (the 1/n diversification method), irrespective of the particular mix of options in the plan. While this does not contribute to the theory of home bias, it highlights that a naïve and less sophisticated diversification method is being utilized in the DC pension plan. Therefore, it is expected that internationally diversification is less considered in a DC pension plan and there subsequently is a higher level of home bias.
3. Methodology
This section discusses the way in which the research was performed. It consists of three parts. First, it presents the calculations for home bias. Next, it describes the used method for the economic explanation behind home bias. The last section demonstrates the explanatory variables used in the study and how they are derived.
3.1. Measure of home bias
Home bias is a situation in which an investor simply places too much weight on domestic assets, compared to the optimal weight of these assets. Equity home bias is the relative difference between the optimal and the actual domestic weight in equity. 𝐸𝐻𝐵𝑖 = 1 −()*+,-./ 12345,6+ 72/1689:;,63./ <2=4698 72/16895, (1)
The actual domestic holdings of a country are the ratio of its domestic equity holdings to its total equity holdings. The total equity holding is the sum of both the foreign and domestic equity holdings.
There are many methods for calculating the optimal foreign holdings. The data used in this paper only contain the asset allocation of pension funds and not the “exact location” of the investments, making certain methods somewhat difficult to implement. Therefore, this paper uses the model of Sharpe and Merton because of its simplicity and usefullness. This model, which relies on the CAPM, assumes that investors can trade with no transaction costs or taxes; however, as argued in Section 2.2, these play a small role in today’s equity markets. The optimal foreign holding in this method is estimated as the relative share of foreign equity to the total equity market.
𝑂𝑝𝑡𝑖𝑚𝑎𝑙 𝑓𝑜𝑟𝑒𝑖𝑔𝑛 ℎ𝑜𝑙𝑑𝑖𝑛𝑔 =ST2=/1T614)S12345,6+
ST2=/1T614 , (3)
However, since this estimation is a little less reliable than other measurements for home bias, the results do not differ much from other estimation methods for this bias (Mishra, 2015). Furthermore, other prestigious works have also used this method in their papers (see, e.g., Hau and Rey, 2008; Chan et al., 2005; Cooper and Kaplanis, 1986; Fidora et al., 2006).
3.2. Regression analysis
To show the possible explanations of equity home bias this study performs a regression analysis. The measure for home bias this study uses is viable for a regression analysis (Mishra, 2015). Since, limited data on pension funds is available; I run two different regressions regarding the equity home bias. One regression for the equity home bias of the general investor3 and one regression for the equity home bias of pension funds. From now on, the regression for the equity home bias of the general investor is called regression A, and the regression for the equity home bias of pension funds is regression B. The following equations give a representation of the specified regression model:
Regression A: 𝐼𝐻𝐵𝑖𝑡 = 𝛼 + 𝛽 ∗ 𝑋𝑖𝑡+ εit, (4) Regression B: 𝑃𝐻𝐵𝑖𝑡 = 𝛼 + 𝛽 ∗ 𝑋𝑖𝑡 + εit, (5)
Where 𝐼𝐻𝐵𝑖𝑡 and 𝑃𝐻𝐵𝑖𝑡 are the equity home bias for the general investor and the pension funds respectively for country i at time t, 𝛼 is the intercept, 𝑋𝑖𝑡 is the vector of explanatory variables, and εit is the error term. A Hausman test is performed to check whether a random or a fixed effect model is preferred. In both cases, the Hausman test rejects the null hypothesis that a random effect is preferred. Therefore, a fixed effect model is used.
To make use of these Ordinary Leasy Squares (OLS) regressions, the assumptions of OLS must hold. The full ideal conditions consist of a collection of assumptions about the linear regression model and the data generating process, and they can function as a description of an ideal data set. Regression A and B comply with the first assumption that the formula must be linear. Second, the dependent variables have to follow a normal distribution. This is tested by performing a Jarque-Bera test for both regression. The test does not reject the null hypothesis of normal distribution at the dependent variables. Third, a test for autocorrelation is conducted; the Breusch-Pagan and Pesaran test reject the null hypothesis of autocorrelation. To correct for heteroskedasticity, a robust regression is used to check the validity of the statistical
3 The general investor includes central banks, deposit-taking corporations other than the central bank,
tests. With this robust regression, parameter estimates do not change; however, the standard errors become slightly larger. The higher standard errors are included in the results. The tests for the OLS assumptions are presented in Appendix A.
3.3. Independent variables
Earlier studies provide explanations for the existence of home bias in the equity markets, for example, barriers to international investment (Stulz, 1981; Black, 1984), information asymmetries (van Nieuwbenborgh and Veldkamp, 2009), real exchange rate risk (Fidora et al., 2006), the hedging of human capital (Baxter and Jermann, 1997; Obstfeld and Rogoff, 1988; Stockman and Dellas, 1989;), indirect diversification possibilities (Mathur and Hanagan, 1983; Ghazalian and Furtans, 2008), and the size of the investments (Barron and Ni, 2008; Amadi, 2004). Many countries in the present market face no or low barriers to international investments and information accessibility. Information asymmetries and barriers to international investment consequently form no (economic) explanation for home bias and are thus excluded. The remaining explanations for home bias are the real exchange rate risk, the euro, indirect diversification possibilities, the correlation with human capital, the size of the investments, and the pension plan.
Fidora et al. (2006) link real exchange rate risk to real exchange rate volatility, and they show that the latter drives home bias. The real exchange rate volatility is measured over a 1- and 5-year period, the data for which were provided by the World Bank, which set the exchange rate of 2010 equal to 100 as a benchmark.
Investors in countries with a common currency face less real exchange rate risk, since there are no real exchange rate risk between countries with the same currency. Countries within the eurozone are marked with a 1 and 0 otherwise. The eurozone consists of Austria, Belgium, Cyprus, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Portugal, Slovakia, Slovenia, and Spain. The U.S. dollar is not considered as a common currency.
The analysis includes indirect diversification opportunities as a possible determinant of home bias. An indirect diversification is, for example, a domestic-based multinational firm. In most studies, domestic indirect diversification opportunities are measured by the relative size of a country’s exports to its GDP (see, e.g., Rowland and Thesar, 2004; Mathur and Hanagan, 1983; Ghazalian and Furtans, 2008). These data are obtained from the World Bank.
As discussed earlier, a reason to invest in foreign equities is to hedge human capital. Both investors and domestic equity are correlated to human capital, which is measured based on quality of education or labor income (Schumann, 2002). This research uses labor-income based measurements because the best available data for this correlation are the labor share income ratio provided by the organization for economic co-operation and development (OECD). This ratio is the share of labor income to the total income over all industries. The OECD data contain information up until 2013, and for 2013-2016, the data of Eurostat are used. These data provide the monthly average income of employees at a country level; these statistics are multiplied by the total number of working people and then multiplied by 12 to obtain the average annual labor income per country. The annual labor income is then divided by the total income in the country. For the missing values, an estimation is made.
account for this, size is measured by dividing the total equity portfolio of the investor by the country’s GDP. The total number of equities of pension funds is obtained from annual reports; the size of the general investors is obtained from the CPIS database; and the GDP data are obtained from the World Bank.
A potential determinant for the home bias in pension funds is the pension plan. Since the global world crisis in 2008, there has been a shift from DB to DC pension plans. In most cases, the employer offers two types of pension plans: old employees still receive a DB pension plan, and new ones receive a DC pension plan. Since this is a gradually changing process, the shift from DB to DC pension plans also changes gradually (Yermo and Severinson, 2010). The data are extracted from the global pension statistic by tower watson studies; however, some data are missing, due to the gradually changing process, the missing data can be estimated.
Table 1 provides an overview of all the independent variables included, how they are measured, and their expected signs. Their expected signs are based on the discussion in Section 2.2.
Table 1. Description of the explanatory variables
This table summarizes the explanatory variables for the regressions. The left column shows the explanatory variables, the middle column describes the variables and the right column represents the expected signs.
Variable Description Expected
sign
ReRR-1yr
The real exchange rate risk over 1 year is measured by the one year difference of the exchange rate divided by the previous year exchange rate.
+
ReRR-5yr
The real exchange rate risk over 5 years is measured by the 5-year difference of the exchange rate divided by the exchange rate 5 years before.
+
IDP The indirect diversification possibilities are defined by the export-to-GDP, which is the total exports of the country divided by their GDP.
+ Labor The labor share measures the fraction of national income accruing to
labor, which is the total labor income divided by the total income.
- Size The size is estimated by the total investments divided by the GDP. - Euro This dummy captures the euro. One if the country has the euro and
zero otherwise.
- DB The defined benefit is the amount of the pension funds’ assets that
belong to a defined benefit plan divided by its total assets.
4. Data
This section presents the sources and data used in this study. First, data for the home bias of pension funds are discussed. Thereafter, the data for home bias of the general investor are demonstrated and compared to the data of pension funds. Next, the sample characteristics are presented.
4.1. Data pension funds
4.1.1. Data Dutch pension funds
To construct the Dutch home bias, a data set of the DNB is employed. The DNB is responsible for prudential supervision of Dutch pension funds, and the data set provides an overview of the macroeconomic developments in the Dutch pension sector. The DNB acquires the information by a limited number of pension funds through the DRA reporting method. The data obtained is increased to national numbers through the use of statistical methods. The data set consists of more than 200 Dutch pension funds on a quarterly basis from 2008-2017 with their asset allocations—real estate, bonds, equities, cash, and derivatives—and their market-wise positions (the Netherlands, “other European countries,” and the “rest of the world”) for the asset classes of bonds and equities. It is an unbalanced panel, and non-sampled pension funds are missing; however, it covers 95% of all assets held by pension funds and is therefore considered to be a reliable sample for this research.
A large portion of the equity investment of Dutch pension funds is in other investment institutions. These institutions re-invest the (equity) investments of pension funds in bonds, equities, and real estate. The re-distribution of those investments is taken into account to obtain a clearer view of the distribution of the end-investment.
The available data from the DNB consist of both riskier assets, such as stocks, and less risky assets, such as fixed income. The question arises as to whether home bias exists in the different asset classes. The average home bias of domestic equities over the past eight years tends to be 5%, and the average home bias of bonds is 15.1%. There is a diminishing pattern observable for equity home bias until 2012, and from there on, the holdings in equities are quite stable, while in bonds, there is an increasing pattern up to 2011, at which point the holdings in domestic bonds are quite stable, as can be seen in Figure 1.
Figure 1 the evolution of the yearly average home bias of equities and bonds by Dutch pension funds
4.1.2 Data worldwide pension funds
To identify home bias in the equity investment of pension funds in other parts of the world, data from several reports are used. First, the pension fund indicators (PFI) report provides data on the domestic and foreign equity holdings as well as the size of the pension funds in Japan, the Netherlands, Switzerland, the UK, and the US in 2001 and
2016. These pension funds’ assets account for 67% of all the collective pension fund assets throughout the world. Furthermore, the PFI report contains data of the equity allocation of the UK’s pension funds from 1962 to 2016. Second, the Schweizer Pensionkassenstudie 2017 presents the domestic and foreign equity holdings of the Swiss pension funds in the period from 2007 to 2017. Third, a report by Callan associates is used for the holdings of equities by US pension funds from 2007 to 2017, and for the equity holdings of Japanese pension funds, the data from a report of the Japanese Pension Fund Association is used. Finally, the data of the equity holdings of Canadian pension funds are obtained from the annual reports of the Canadian pension plan investment board (CPPIB) from 2009 to 2016. While the CPPIB does not reflect all the equity investment by pension funds, it does cover more than 67% of the total investments of Canadian pension funds; therefore, it is considered a representative indicator for Canadian equity holdings. Furthermore, there is a clear pattern in the equity investments of the Canadian pension funds, and an approximation of investments for the years 2008 and 2007 can thus be made.
The data descriptive statistics in Figure 2 comply with the previous literature that all countries invest too much in domestic equities and that home bias decreases. The figure illustrates that Dutch pension funds have the lowest home bias. A possible explanation is that the Netherlands is in the EMU and therefore does not consider real exchange rate risk with other EMU countries. In addition, it is a member of the EU, which creates more stimulation to invest in other EU-countries.
Figure 2 The equity home bias of pension funds in Canada, Japan, the Netherlands, Switzerland, the UK, and the US in 2001 and 2016
4.1.3 Characteristics pension funds
Table 2 lists the old age dependency ratio in 2010 and the expected ratio in 2100 for the countries in the sample. This ratio is important because it presents the number of people who are in the working population relative to the number of retirees. In the future, the dependency ratio will be highest in Japan, the Netherlands, and Switzerland respectively. The value of the total pension assets over their GDP provides not only a scale of the pension funds’ activity but also the possibility to cover the future dependency ratio. Noticeably, in the Netherlands and Canada, pension funds play a large role in relation to their economy as a whole, with asset-to-GDP ratios of 1.72 and 1.36 respectively in 2016, whereas in Japan, the pension funds play a small part (0.27). The pension funds, with the exception of those in Japan, are also growing relative to their GDPs. The DB pension plans only account for 40% of U.S. pension funds; this is noticeably less than the other countries, where DB pension plans dominate.
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
Canada Japan Netherlands Switzerland UK US
4.2. Data general investor
The data for the pension funds are quite limited. Therefore, data of the equity holdings of the general investor are used to provide more proof
4.2.1. Home bias general investor
A data set of the CPIS of the International Monetary Fund (IMF) is employed for the equity holdings of the general investor. The CPIS is a data collection method performed under the authority of the IMF that collects an economy’s data related to its holding of portfolio investment securities (equities and bonds). The data set provides annual investment data about the general investors. The IMF data set contains the equity and debt holdings for almost all the countries in the world between 2001 and 2016. For this research, the data of 13 EU countries, Australia, Canada, Japan, Switzerland, the UK, and the US are used. The 13 EU countries are selected to show the effect of the euro on home bias. Canada, Japan, Switzerland, the UK, and the US are selected based on the data availability of the pension funds in these countries. The data for 2001-2016 are used because this period is assumed to be able to reflect the long-term behavior of investors.
Table 3 presents the home bias level of general investors and that of pension funds in Canada, Japan, the Netherlands, Switzerland, the UK, and the US in 2001 and 2016. The table shows that general Dutch investors have the lowest equity home bias in 2001 (0.52) and in 2016 (0.30), while Japan has the highest home bias (0.89) in 2001. In 2016, Japan and the US both has the highest equity home bias of 0.71. A decreasing home bias is observed over time for the general investor, with the exception of Canada, which has an increasing home bias of 0.62 in 2001 and 0.69 in 2016. The table also shows that in all countries, the home bias of pension funds is lower than that of general investors. It can be assumed that pension funds, as professional asset managers, are subject to a lower home bias than non-financial corporations or households. This is known as the so-called ‘professionalism effect.’ General investors also include these non-financial corporations. The home bias of Dutch pension funds demonstrates a low home bias (0.17 in 2001 and 0.04 in 2016), compared to the other countries. In 2001, the pension funds of Japan and the UK have the highest home bias (0.62), and those of the US has the highest home bias in 2016 (0.45). The home bias of Canadian pension funds is the closest to the home bias level of the Netherlands; however, it is still more than twice as much: 9% and 4% respectively.
Table 2. Characteristics of pension funds
This table shows the characteristics of pension funds in Canada, Japan, the Netherlands, Switzerland, the UK, and the US in 2016. Total assets are the assets of pension funds in billions of USD, total pension assets over GDP are for 2007 (2016 in parentheses), DB is the
percentage of DB pension plans in each country, and the dependency ratio is the ratio of the population aged 65+ per 100 population in % for the years 2010 and 2100.
Table 3. Equity home bias
The equity home bias level in percentage (%) for 2001 and 2016 for general investors and pension funds in Canada, Japan, the Netherlands, Switzerland, the UK, and the US. Pension
funds equity home bias for Canada in 2001 is missing. General investors’ equity
home bias
Pension funds’ equity home bias
2001 2016 2001 2016 Canada 0.62 0.69 - 0.09 Japan 0.89 0.71 0.62 0.44 Netherlands 0.52 0.30 0.17 0.04 Switzerland 0.53 0.42 0.53 0.41 UK 0.72 0.58 0.62 0.31 US 0.76 0.71 0.58 0.45
4.2.2. Regional bias towards Europe
As previously stated, the EMU has created a new form of home bias: euro bias, and the CPIS data set also provides data to demonstrate this bias. Euro bias can be estimated in a similar way to the analysis for domestic home bias. However, in this case, the domestic equities are seen as the equities inside eurozone countries.
Figure 3 illustrates euro bias in European countries and demonstrates that it is positive but decreasing in eurozone countries. In the non-eurozone countries, there is currently a negative euro bias, while a positive euro bias exists in the eurozone. The UK is the only non-eurozone country that shows a positive home bias towards Europe, whereas for Finland, it is the opposite; since the introduction of the euro in 1999, the bias towards other eurozone countries decreased and even became negative.
Figure 3, Regional home bias of EMU and non-EMU countries towards Europe in 2001, 2006, 2011, and 2016
A first glance at the data descriptive statistics shows that there is a positive but decreasing trend of home bias among general investors and pension funds. Furthermore, the statistics indicate that there are substantial differences in the home bias between investors as well as countries. These results serve as further motivation to investigate whether there is an economic explanation for home bias.
4.3 Descriptive statistics
The summary statistics of the variables for regression A and B are presented in Table 5. The left column contains the variables of home bias of general investors for 13 EU countries, Australia, Canada, Japan, Switzerland, the UK, and the US from 2001 to 2016. In the right column are the variables for Canada, Japan, the Netherlands, Switzerland, the UK, and the US from 2007 to 2016. The statistics demonstrate no extreme values. The number of observations in the sample for pension funds is small; therefore, there is a possibility that the results cannot fully reflect the population. To overcome this problem, a test is performed on general investors to create more observations. In much research, no distinction is made between the determinants of the investors. Therefore, I believe that a test for the determinants of home bias among general investors can support the economic explanation for home bias among pension funds.
The small number of observations for pension funds creates another drawback. Two variables have to be dropped from this equation because a panel regression has the requirement that the number of regressors is smaller than the number of cross-sections. The variables ReRR-1yr and RReR-5yr are dropped, because these variables have the least fit with the regression. In addition, an omitted variable test does not reject the null hypothesis that these variables are irrelevant. The regressions with ReRR-1yr and RReR-5yr together with the omitted variables test are presented in Appendix B. Furthermore, the variable euro is dropped because there was only one country (the Netherlands) with the euro currency, this creates a near singular matrix and therefore the fixed effect model cannot be tested. However, it is assumed in this paper that regression A, the test for euro bias and the theory will provide enough evidence for the influence of the euro.
Because there are two different regressions, there are also two separate correlation tables. The correlation tables are presented in appendix C. Since the correlation matrices demonstrate a low correlation between the variables as well as with the residuals it is expected that this forms no problem for this. Furthermore, a Jarque-Bera test is performed to check for non-normality in the residuals. It does not reject the null hypothesis that the residuals of regression A and B are normal distributed. If there was non-normality in the residuals, outliers had to be deleted.
Table 3. Summary statistics
The descriptive statistics of variables used to measure the determinants of equity home bias among general investors (left) and pension funds (right). Real exchange rate volatility over 1
year (ReRV-1yr), Real exchange rate volatility over 5 years (ReRV-5yr), share labor income/total income (labor), indirect diversification possibilities export/GDP (IDP), size of portfolio total assets/GDP (Size), countries with the euro (1) or not (0), and the percentage of
pension funds with a DB in a country.
Home Bias General Investors Home Bias Pension Funds Variable Mean Max Min Std.
Dev.
5. Empirical Results
This chapter presents the results from the panel regressions. The results of regression A are presented first, followed by the results of regression B.
5.1. Results for the general investor
First, the results of regression A in table 4 demonstrate that the real exchange rate is insignificant, which indicates that real exchange rate volatility does not have an effect on the home bias of the general investor. The progress towards a global monetary ability made in recent years could be an important factor for the insignificant relation. International markets are increasingly integrating, which results in the real exchange rate risk having a smaller influence because there is simply less volatility. Furthermore, the sample consists of developed countries, where there is less real exchange rate volatility. The real exchange rate risk is consequently not considered as a barrier for international equities. In addition, this risk can be hedged by holding other foreign equity, thereby reducing home bias.
Second, the real exchange rate risk as a barrier is also negligible because indirect diversification possibilities, such as MNOs, can act as a hedging channel for real exchange rate fluctuations. However, the results from regression A show that the indirect diversification possibilities are insignificant, which indicates that indirect diversification possibilities do not have an effect on the home bias of the general
Table 4. OLS output
The dependent variable is the ratio of home bias for Australia, Austria, Canada, Denmark, Finland, France, Germany, Greece, Japan, Italy, the Netherlands, Portugal, Spain, Switzerland, the UK and the US. The variables are real exchange rate risk over 1 year (ReRR-1yr), real exchange rate risk over 5 years (ReRR-5yr), share labor income (labor),
indirect diversification possibilities (IDP), relative size of investment to the GDP (Size), dummy variable for the euro and the error term. *, **, *** indicates statistical significance
at the 10%, 5% and 1% level respectively.
investor. Nowadays, the barriers for investing in foreign equities are low. Therefore, investors do not perceive the barriers as high enough to consider indirect diversification possibilities.
Third, the results show an insignificant relation between the labor share and home bias. The theory behind this is that investors who are more aligned with their own human capital need to diversify their equity to foreign countries. However, domestic equity today is less influenced by a domestic economy. For instance, in the Netherlands, the shares of a company such as Philips has a low correlation with the Dutch economy; therefore, investing in domestic equities is also a way to diversify its correlation with domestic labor income and is thus not a reason to invest in foreign equity.
Fourth, the results correspond to the expectations that size is a determinant of home bias. The sign of size is negative and significant, which indicates that larger investors have a lower home bias. The results support my prior statement that economies of scale, in general, provide investors with more opportunities to diversify their assets across markets, thereby, reducing their domestic bias. Similarly, it complies with the expectation that larger investors are too substantial to diversify their risks domestically and thus need to diversify abroad.
Last, eurozone countries demonstrate a significantly lower home bias than non-eurozone countries. This result corresponds with the findings of the euro bias and can be explained by union-wide inflation targets and more stable exchange rates.
5.2. Results for pension funds
Table 5 reveals that indirect diversification possibilities have a significant and negative effect on the home bias of pension funds, which indicates that more indirect diversification possibilities in a country persuades pension funds to invest more in foreign equities. This contradicts the expectation that it would have a positive effect. A possible explanation can be that the estimation method for indirect diversification also can be used as a measurement for the openness of the economy. Countries with a higher export-to-GDP ratio are more open towards foreign investments. If companies within these countries are likely to do business abroad and diversify their business to international markets, investors could act in the same way.
The sign of labor in regression B is insignificant, which means it does not influence the home bias of pension funds. An explanation can be that the equity market already has a low correlation with human capital; therefore, this does not have to be considered as a reason to invest in foreign equities.
Last, DB has a positive and significant sign, which means that countries with a higher percentage of DB pension plans have a higher level of home bias. This contradicts my expectation that a DC pension plan has a higher home bias. The expectation was based on the naïve diversification of the decision maker in the DC pension plan. A possible explanation for this result can be that naïve diversification methods can lead to an overweighting in foreign equity (foreign bias). Naïve diversification does not necessarily mean investing more in domestic equities; nevertheless, this was expected because the decision maker in a DC plan, the individual, was believed to have a higher preference for domestic assets than pension funds. However, these results indicate that the opposite is true.
6. Conclusion
This research presents a comprehensive and thorough analysis of the equity holdings of pension funds from 2007 to 2017. It is the first to seek evidence of the home bias of pension funds in countries from around the world. In contrast to earlier studies, it uses macroeconomic data from annual reports to examine the economic factors that can possibly capture pension funds’ equity home bias.
The data provide robust evidence of the ongoing but declining home bias in developed markets and the existence of regional bias in Eurozone countries. The data also demonstrate a lower level of home bias among pension funds than for the general investor, and it indicates that Dutch pension funds have the lowest home bias, compared to other countries in the sample. These findings are a cause for further investigations into the economic explanation for the discrepancies between pension funds and general investors.
Table 5. OLS output
The dependent variable is the ratio of home bias for pension funds in Canada, Japan, the Netherlands, Switzerland, the UK, and the US. The variables are the share of labor income
(Labor), indirect diversification possibilities (IDP), the relative size of investment to their GDP (Size) and the percentage of DB pension plans in the country. *, **, *** indicates
The previous literature provides six rational explanations for equity home bias. This research adds an extra explanation, namely the type of pension plan, for the equity home bias of pension funds. In the end, only three variables show a highly significant effect on the home bias of pension funds: common currency, size, and the type of pension plan. Larger pension funds with a common currency and a DC pension plan are likely to have a lower home bias. The Dutch pension funds are the biggest and the only European investor in the sample. This, together with the shifting from a DB to DC pension plan causes the low level of Dutch home bias. The outlook is that pension funds will become even bigger and the shifting from DB to DC pension plans will continue. The results indicate that this has an effect on the equity home bias of pension funds and can therefore cause it to decline further.
From an economic point of view, the call from the Dutch minister for Dutch pension funds to invest more in their domestic equity markets is a cumbersome mission; however, in today’s world, pension funds also have the social obligation to invest in their domestic markets. There is a trade-off between the social (home bias) and economic (no home bias) aspects, where pension funds in this trade-off are increasingly moving to the economic side of the scale because they are becoming larger and shifting from DB to DC pension plans. My expectation is that home bias in pension funds is likely to remain because of social obligation, but pension funds have to be reminded to the social obligation. The call from the Dutch mission is from an economic point of view not justifiable, but it can serve as a reminder for pension funds to keep investing in the domestic market. These results bring the literature a step closer to the solution of the unresolved academic puzzle of home bias.
The findings in this paper are limited because the sample has only a small number of observations (60). This paper provides additional proof by performing a second regression for the general investor with a larger number of observations (288). This regression presents the same significant determinants for home bias as for the home bias of pension funds; however, whether this is enough to represent the economic explanation for the behavior of pension funds is still questionable. Furthermore, this paper neglects the effects of informational asymmetry. Investors could have an informational advantage with regard to domestic equities, and they could therefore create excess returns on domestic equities. Since the data did not have the exact location of the investments, it was not possible to test whether investors generated excess returns on domestic equities. However, if this would be available it is likely that it does not have a big impact on the results, because information asymmetries between mature markets are believed to be small.
References
Ahearne, G., Griever, W., Warnock, E., 2004. Information costs and home bias: an analysis of US holdings of foreign equities. Journal of International Economics 62, 313–336.
Amadi, A., 2004. Equity home bias: a disappearing phenomenon? Working paper. University of California, Berkeley.
Baele, L., Pungulescu, C., Ter Horst, J., 2007. Model uncertainty, financial market integration and the home bias puzzle. Journal of International Money and Finance 26, 606-630.
Balli, F., Basher, S., Ozer-Balli, H., 2010. From home bias to euro bias: disentangling the effects of monetary union on the European financial markets. Journal of
Economics and Business 62, 347-366.
Barron, M., Ni, J., 2008. Endogenous asymmetric information and international equity home bias: The effects of portfolio size and information costs. Journal of International Money and Finance 27, 617-635.
Baxter, M., Jermann, J., 1997. The international diversification puzzle is worse than you think. The American Review 87, 170-180.
Bekaert G, Harvey C., Lundblad, C., Siegel, S., 2007. Global Growth Opportunities and Market Integration. Journal of Finance 62, 1081–1137.
Benartzi, S., Thaler, R., 2001. Naïve diversification strategies in defined contribution saving plans. American Economic Review 110, 73–92.
Benartzi, S., Thaler, R., 2007. Heuristics and biases in retirement savings behavior. Journal of Economic Perspectives 21, 81–104.
Bernheim Douglas, B., 1996. Financial illiteracy, education, and retirement saving. Working paper. University of Pennsylvania, Pennsylvania.
Bikker, J., de Dreu. J., 2006. Pension fund efficiency: the impact of scale, governance and plan design. Working paper. De Nederlandse bank, Amsterdam.
Black, F., 1974. International capital market equilibrium with investment barriers. Journal of financial economics 1, 337-352.
Brennan, M., Cao, H., 1997. International portfolio investment flows. Journal of finance 52, 1855-1880.
Chan, K., Covrig, V., 2012. What determines mutual fund trading in foreign stocks? Journal of International Money and Finance 31, 793-817.
Cole, H., Obstfeld M., 1991. Commodity trade and international risk sharing: how much do financial markets matter? Journal of Monetary Economics 28, 3–24.
Cooper, I., Kaplanis, E., 1986. Recent developments in corporate finance. Cambridge University Press, Cambridge.
Coval, J., 1996. International capital flows when investors have local information. Working paper. University of Michigan, Michigan.
Coval, J. D., Moskowitz, T. J., 1999. Home bias at home: local equity preference in domestic portfolios. The Journal of Finance, 54, 2045–2073.
De Santis, G., 1993. Globalization of equity markets: a test using time-varying
conditional moments. Unpublished working paper. University of Southern California, Los Angeles.
De Santis, G., Gerard, B., 1997. International asset pricing and portfolio diversification with time-varying risk. The Journal of Finance 52, 1881-1912.
Dreu, J., Bikker, J., 2012. Investor sophistication and risk taking. Journal of Banking and Finance 36, 2145-2156.
Eldor, R., Pines, D., Schwartz, A., 1988. Home asset preference and productivity shocks. Journal of International Economics 25, 165-176.
Fidora, M., Fratzscher, M., Thimann, C. 2006. Home bias in global bond and equity markets: the role of real exchange rate volatility. Working paper. European Central Bank, Frankfurt.
Frankel, J., 1982. A search of the exchange risk premium: a six-currency rest
assuming mean-variance optimization. Journal of International Money and Finance 1, 255–274.
French, K., Poterba, J., 1991. Investor diversification and international equity markets. American Economic Review 81, 222-226.
Ghazalian, P., Furtan, W., 2008. The effects of multinational activities on the
measurement of home bias. Journal of the Japanese and International Economies 22, 401- 416
Grauer, R., Hakansson N., 1987. Gains from international diversification: returns on portfolios of stocks and bonds. Journal of Finance 42, 721-741.
Grubel, H., 1968. Internationally diversified portfolios. American Economic Review 58, 1299-14.
Heathcote J., Perri, F., 2013. The international diversification puzzle is not as bad as you think. Journal of Political Economy 121, 1108–1159.
John Hancock Financial Services, 1995. Gallup Survey of Defined Contribution Plan Participants.
Levy, H., Levy, M., 2014. The home bias is here to stay. Journal of Banking and Finance 47, 29–40.
Lippi, A., 2016. (Country) Home bias in Italian occupational pension funds asset allocation choices. The Quarterly Review of Economics and Finance 59, 78–82. Low, A., 1993. Essays on asymmetric information international finance. Unpublished working paper. University of California, Los Angeles.
Mathur, I., Hanagan, K., 1983. Are multinational corporations’ superior investment vehicles for achieving international diversification? Journal of International Business Studies 14, 135-146.
McQueen, G., Stenkrona, A., 2012. The home-institution bias. Journal of Banking and Finance 36, 1627-1638.
Mishra, A., 2015. Measures of equity home bias puzzle. Journal of Empirical Finance 34, 293–312.
Plender, J., 1983. Pension funds and accountability, The Political Quarterly 54, 221-231.
Rowland, P., and Tesar, L., 1998. Multinationals and the gains from international diversification. Review of Economic Dynamics 7, 789-826.
Rubbaniy, G., van Lelyveld, I., Verschoor, W., 2010. Home bias and Dutch pension funds’ investment behaviour. European journal of finance 20, 78–993.
Schoenmaker, D., and Bosch, T., 2008. Is the home bias in equities and bonds declining in Europe? Investment Management and Financial Innovations 5, 90-102. Schumann, C., 2002. Measuring human capital – old and new approaches and their suitability for growth analysis in transition countries. Working paper. University of Tartu, Estonia.
Sørensen, B., Wu, Y., Yosha, O., Zhu, Y., 2007. Home bias and international risk sharing: twin puzzles separated at birth. Journal of International Money and Finance 26, 587–605.
Solnik, B., 1974. Why not diversify internationally rather than domestically? Financial Analysts Journal 30, 91-135.
Tesar, L., Werner, I., 1995. Home bias and high turnover. Journal of International Money and Finance 14, 467-492.
Tobin, J., Brainard, W., 1992. Financial intermediaries and the effectiveness of monetary controls. The American Economic Review 53, 383-400.
Van Nieuwenburgh, S., Veldkamp, L., 2009. Information immobility and the home bias puzzle. The Journal of Finance 64, 1187–1215.
Yermo, J., Severinson, C., 2010. The impact of the. financial crisis on defined benefit plans and the need for counter-cyclical funding regulations. Working Paper. The organisation for economic co-operation and development, Paris.
Appendices
Appendix A.1: Robustness checks
Table A.1. Jarque-Bera test results for dependent variables Dependent Variable Skewness Kurtosis Probability General Investor -0.388 3.238 0.373 Pension Funds -0.151 2.208 0.407
Table A.2. Correlation and redundancy test results
Tests for correlation and redundancy test for two regressions with the dependent variables general investor and pension funds, respectively. *, **, *** indicates statistical significance at the 10%, 5% and 1% level respectively.
Breusch-Pagan LM Pesaran scaled LM Bias-corrected scaled LM Pesaran CD Redundant Fixed Effects Tests General Investor 694.6*** 30.9*** 30.3*** 20.7*** 5.2*** Pension Funds 40.2*** 4.6*** 4.3*** 4.2*** 74.27*** Appendix B: Regressions
Table B.1 OLS output – Test with variables ReRR-1yr, ReRR-5yr, Size, and DB The dependent variable is the ratio of home bias for pension funds in Canada, Japan, the Netherlands, Switzerland, the UK, and the US. The variables are real exchange rate risk over 1 year (ReRR-1yr), real exchange rate risk over 5 years (ReRR-5yr), the relative size of investment to their GDP (Size) and the percentage of DB pension plans in the country. *, **, *** indicates statistical significance at the 10%, 5% and 1% level respectively.
Table B.2 OLS output – Test with variables ReRR-5yr, Labor, Size, and DB The dependent variable is the ratio of home bias for pension funds in Canada, Japan, the Netherlands, Switzerland, the UK, and the US. The variables are real exchange rate risk over 5 years (ReRR-5yr), the share of labor income (Labor), the size of investment to their GDP (Size) and the percentage of DB pension plans in the country. *, **, *** indicates statistical significance at the 10%, 5% and 1% level respectively.
Coefficient t-statistic ReRR-5yr -0.065 -0.623 labor 0.773* 1.877 Size -0.168*** -8.091 DB 1.898*** 7.668 Constant -1.518*** -5.349 R2 0.957 Adjusted R2 0.949 Observations 60
Table B.3 OLS output – Sample with ReRR-1yr, Labor, Size, and DB The dependent variable is the ratio of home bias for pension funds in Canada, Japan, the Netherlands, Switzerland, the UK, and the US. The variables are real exchange rate risk over 1 year (ReRR-1yr), the share of labor income (Labor), indirect diversification
possibilities (IDP), the relative size of investment to their GDP (Size) and the percentage of DB pension plans in the country. *, **, *** indicates statistical significance at the 10%, 5% and 1% level respectively.
Table B.6. Omitted variable test Dependent Variable Value Probability
ReRR-1yr 0.532 0.597
ReRR-5yr 1.318 0.194
Table B.4 OLS output – Test with variables ReRR-1yr, IDP, Size, and DB The dependent variable is the ratio of home bias for pension funds in Canada, Japan, the Netherlands, Switzerland, the UK, and the US. The variables are the real exchange rate risk over 1 year (ReRR-1yr), indirect diversification possibilities (IDP), the relative size of investment to their GDP (Size) and the percentage of DB pension plans in the country. *, **, *** indicates statistical significance at the 10%, 5% and 1% level respectively.
Coefficient t-statistic ReRR-1yr -0.069 -0.347 IDP -0.184** -4.340 Size -0.119*** -5.086 DB 1.951*** 8.778 Constant -1.060*** -5.529 R2 0.955 Adjusted R2 0.948 Observations 60
Table B.5 OLS output – Test with variables ReRR-5yr, IDP, Size, and DB The dependent variable is the ratio of home bias for pension funds in Canada, Japan, the Netherlands, Switzerland, the UK, and the US. The variables are real exchange rate risk over 5 years (ReRR-5yr), indirect diversification possibilities (IDP), the relative size of investment to their GDP (Size) and the percentage of DB pension plans in the country. *, **, *** indicates statistical significance at the 10%, 5% and 1% level respectively.
Appendix C: Correlation matrices
Table C.1. Correlation matrix – Sample with the general investor Correlation matrix containing the correlation of the independent variables used to investigate the determinants of the home bias general investors. Real exchange rate risk over 1 year (ReRR-1yr), Real exchange rate risk over 5 years (ReRR-5yr), share labor income (labor), indirect diversification possibilities (IDP), Size of investment (Size), dummy variable for the euro and the error term.
ReRR-1yr
ReRR-5yr
IDP labor Size Euro Error Term ReRR-1yr 1 ReRR-5yr 0.310 1 IDP -0.249 -0.323 1 Labor -0.211 -0.279 0.024 1 Size -0.019 -0.134 0.653 0.050 1 Euro -0.421 -0.420 0.282 0.458 -0.061 1 Error Term 0 0 0 0 0 0 1
Table C.2. Correlation matrix – sample with pension funds
Table 4: Correlation matrix containing the correlation of the independent variables used to investigate the determinants of the home bias general investors. share labor income (labor), indirect diversification possibilities (IDP), Size of investment (Size), the ratio of a DB pension plan and the error term.
Labor IDP Size DB Error Term Labor 1 IDP -0.286 1 Size -0.525 -0.506 1 DB 0.504 -0.127 -0.206 1 Error Term 0 0 0 0 1
Table C.3. Jarque-Bera test results for residuals
Dependent Variable Skewness Kurtosis Probability General Investor -1.333 1.152 0.139
