سرمایه انسانی سهامداران و پازل تنوع بخشی بین المللی

This paper evaluates the extent of the international diversification puzzle when human capital is considered part of the wealth of nations. The analysis examines whether (i) the inclusion of human capital in the wealth of portfolio of individuals, (ii) the different human capital assets held by stockholders and non-stockholders, and (iii) frictions in human capital markets, can help explain the puzzle. The methodology consists of comparing Hansen–Jagannathan bounds on the stochastic discount factor (IMRS) implied by human capital and financial returns across different countries. The results suggest that the information contained in the human capital of stockholders can greatly contribute towards explaining the international diversification puzzle.

The “international diversification puzzle” observes that investors hold too little of their financial wealth in foreign securities and that potential benefits from diversification exist. U.S. investors hold more than 90 percent of their financial assets in the form of domestic securities. In the United Kingdom, Germany, and Japan, for instance, the share of domestic financial assets in investors’ financial wealth portfolios exceeds 85 percent. Similar numbers are also observed for a variety of other countries. Using existing models of investment and portfolio diversification, the evidence reported in several studies indicates the existence of a significant “home bias” in international financial markets. Although these markets have experienced a very significant growth and a substantial increase in their levels of integration during the last few decades, the low existing levels of international diversification of financial wealth are still considered to be one of the most intriguing and elusive puzzles in international economics and finance. Different classes of potential explanations have been discussed in the literature. Among these are institutional barriers, transaction costs and explicit limits on cross-border investments.1 However, as French and Poterba (1991) discuss, all of these “are unlikely to explain the low levels of cross-border equity investment today” and the “apparent tendency for portfolio investors to overweight their own equity market” appears to be “the result of investors choices, rather than constraints” (italics added). Two recent papers by Baxter and Jermann (1997) and Bottazzi et al. (1996) suggest that one potentially crucial piece of this puzzle that has been consistently ignored in the literature is the role of human capital, an asset which is the largest component of the wealth portfolio of individuals and countries. 2 Given the dominant position of this asset in individual and aggregate wealth, it is important to evaluate how human capital affects the international diversification puzzle. The initial results that these authors obtain, however, are inconclusive and divergent for the role of human capital assets in this puzzle. 3 A common procedure in the literature is to examine the home bias puzzle at the aggregate level. Interestingly enough, most individuals hold few or no stocks and, as French and Poterba (1991) remark, the puzzle appears to be the result of investors’ choices. Given this heterogeneity in the population, this paper focuses on the analysis of the puzzle with human and financial assets at a basic disaggregated level. More precisely, the analysis evaluates whether the human capital of stockholders, rather than aggregate human capital, may help us understand the weak extent of international financial diversification of equity investments. The analysis initially examines the extent to which two aspects of the puzzle are important: the measurement of human capital returns and the asset pricing empirical methodology used. It then examines the role of two novel, additional aspects: the extent to which the human capital assets and returns of stockholders are different from those of non-stockholders, and the extent to which the consideration of human capital as a purely nontradable asset is also important. The paper, therefore, accounts for a fundamental source of heterogeneity in the population, and offers the first analysis of the international diversification puzzle with human capital assets at a basic disaggregated level. These four aspects are briefly discussed next. First, it is important to take into account whether the human capital assets and returns of stockholders differ from those of non-stockholders. From the analysis and data used, for instance, in Mankiw and Zeldes, 1991 and Blume and Zeldes, 1994 and Vissing-Jørgensen (1998), and from the Survey of Consumer Finances and other similar surveys, the differences in the demographic characteristics of stockholders and non-stockholders (e.g., age, education and experience), as well as in their consumption behavior, indicate that their human capital asset holdings are also very different. If human capital assets are to be brought to bear on the international diversification puzzle, then these differences, in principle, are potentially important and need to be accounted for. The differentiation between stockholders and non-stockholders has already been fruitful in explaining part of the equity premium puzzle and may also play an important role in the international diversification puzzle. Second, previous empirical analyses have approximated human capital returns by the growth rate in per capita labor income. This measure, however, ignores some of the fundamentals of three decades of work in human capital theory and labor economics, and assumes that labor income growth is unforecastable.4 In this paper, we examine other approximations of human capital returns that can account for some shortcomings of the growth rate in per capita labor income. Third, several specific models of investment under uncertainty and the valuation of risky assets do not seem to provide satisfactory explanations of the behavior of financial asset returns and the extent of efficient investments. For instance, the widely popular unconditional, static CAPM explains only an insignificant part of the cross-sectional variability of average financial returns.5 It would then seem necessary to avoid addressing issues of portfolio diversification using or imposing a specific model, if at all possible. Research in asset pricing and portfolio theory, however, has experienced tremendous progress in the last few years by following the methodology suggested by Hansen and Jagannathan (1991) and Gallant et al. (1990). These authors provide a means of addressing the issues of portfolio diversification and efficient investments in a general way. They derive boundary conditions (respectively, unconditional and conditional on the set of available information) for the moments of a generally defined stochastic discount factor. Different assumptions about the parametrization of this discount factor lead to standard models of asset pricing used in finance.6 A fundamental advantage of this methodology is that no specific parametrization needs to be assumed in order to address issues of portfolio diversification in a general, robust way. The only assumption that needs to be imposed is that portfolios with the same payoffs have the same price (the law of one price). In addition, empirical tests can be implemented using conditioning and unconditioning information, allowing the models to include different classes of frictions in human capital markets, and using the General Method of Moments, a method which requires very weak distributional assumptions about the observed data. Their methodological approach will be followed in this paper. Fourth, as Jagannathan and Wang (1996) point out, the opinion that human capital is not tradable is largely correct, but needs to be qualified. First, active insurance markets exist for hedging some of the risks in human capital investments. Examples include life insurance, unemployment insurance, medical insurance, marriage, and certain forms of partnerships. Second, the market value of mortgage loans, consumer credit and bank loans to the household sector represent 80 percent of GNP. Since these “can be viewed as borrowing against future income, it does not appear inappropriate to view human capital like any other form of capital, cash flows which are traded through issuance of financial assets” (Jagannathan and Wang, 1996, p. 13). Clearly, however, “frictions” in financial markets typically appear to be negligible relative to those in human capital markets (see Becker, 1993). The purpose of this paper is to take into account these four aspects simultaneously and examine the extent of the international diversification puzzle. The findings can be summarized as follows: (i) when human capital assets are ignored, there are indeed significant benefits from international financial diversification; (ii) if human capital assets are considered part of the wealth portfolio and returns are approximated by the growth rate in per capita labor income, the extent of the puzzle becomes smaller, not greater; (iii) the puzzle is further reduced if, instead of using the growth rate in per capita labor income, other alternative measures of human capital returns are used; (iv) for the U.S., the puzzle becomes close to insignificant, in most cases, when the human capital assets of U.S. stockholders alone are taken into account; (v) the previous results, from (i) to (iv), become stronger if, instead of considering human capital as a fully nontradable asset, it is assumed that it is subject to certain frictions (short-sale constraints and transaction costs) or, in the limit, fully tradable. The latter case, obviously, appears to be quite unrealistic. These results suggest that the empirical methodology, the specific approximation of human capital returns, the role of frictions and, most importantly, the differentiation between the human capital of stockholders and non-stockholders all appear to play a significant role in determining the extent of the puzzle. These features differentiate the analysis in this paper from previous work in the literature. An important ingredient of the analysis is the way human capital returns are measured. Unfortunately, while the availability of quality labor income data in the United States allows the calculation these measures of human capital, the data available for other countries do not. For this reason, these measures are calculated only for the United States. The period of analysis covers 1964–1996 and the countries considered are the ones in Baxter and Jermann (1997): the United States, the United Kingdom, Germany and Japan. These countries have actively traded stock markets, legal systems that enforce property rights, and few restrictions on movements of capital. The paper is organized as follows. Section 2 briefly reviews the methodology and its advantages. Section 3 derives individual labor-income-based measures of human capital returns and discusses other approximations of human capital returns. Section 4 describes the sources of the data. Section 5 tests and estimates the significance of the increase in the expected return per unit of risk that mean-variance investors could achieve by diversifying their wealth into the financial assets of other countries. The section concludes by discussing additional empirical evidence and the robustness of the findings. Finally, Section 6 presents some concluding remarks.

Building upon recent insights in the asset pricing and the international finance literatures the analysis in this paper finds that if human capital is considered part of the wealth of nations, as it must be, gains from international financial diversification for a mean-variance investor appear to be smaller than previously reported and, in some cases, close to negligible. We approached the question of international asset diversification in a novel way by using the methodology proposed by Hansen and Jagannathan (1991). While the use of equity returns, instead of fundamentals-based measures of capital returns, does contribute towards explaining the international diversification puzzle, the key features of the analysis are the use of measures of human capital returns that account for skill premiums and, most importantly, the extent to which the human capital of stockholders is different from that of non-stockholders. The results show that all these aspects are important and indicate that the information contained in the human capital returns of stockholders can greatly contribute towards explaining the international diversification puzzle. The vastly different results obtained for stockholders and non-stockholders also suggest that access to international investments through 401K and similar plans will mostly benefit less educated and less wealthy individuals. The analysis in this paper may be extended in various directions. First, it may be possible to implement it with data on wages and asset holdings for other countries (e.g., the Family Expenditure Survey in the United Kingdom), and other conditioning methodology (e.g., Beckaert and Liu, 1999). Second, as mentioned in the introduction, some human capital risks may also be hedged by life insurance, unemployment insurance, medical insurance, marriage or other forms of partnerships. Knowing how individuals hedge some human capital risks in these or other ways should prove useful in examining the extent of the puzzle in further detail. Similarly, a close examination of the limitations on the extent to which human capital portfolios may be adjusted both downward (because of irreversibility) and upward at different frequencies deserves careful consideration. Lastly, further disaggregated data of the demographic characteristics of stockholders (e.g., occupation and industry), and of the mix between stocks and other financial assets that they actually hold, will provide valuable information about the covariance structure of risk-return differentials of human and financial assets for this and other asset pricing puzzles. Data currently available, however, may not be detailed enough to implement this idea anytime soon. In conclusion, the roles of human capital, market segmentation, and heterogeneity in the international diversification puzzle examined in this paper along with the robustness of the patterns of the findings are perhaps best interpreted as one more glimpse of the promise of multidisciplinary inquiry (human capital, labor economics and asset pricing) to meet the challenges and puzzles of modern capital theories.