9 Home Bias in Japan

Alessandro Zanello, and Daniel Citrin
Published Date:
November 2008
  • ShareShare
Show Summary Details
W. Christopher Walker


An apparent preference for domestic over foreign assets, or “home bias,” has been a conspicuous feature of the behavior of many investors in Japan, both individual and institutional. This home bias has declined markedly in the past decade by some measures, but remains higher than average for mature market countries. To the extent that a strong revealed preference for domestic assets in Japan results in a suboptimal allocation of financial assets, it would represent an unexploited opportunity for higher returns on investment in a country that has recently suffered from low rates of return, and that faces the prospect of having to cover a large funding gap in social security programs. From a global perspective, an inefficient portfolio held by Japanese investors could have consequences for the prices of foreign and domestic assets.

This chapter provides an assessment, by type of investor and in the aggregate, of the recent declines in home bias, and of the remaining areas where portfolios are still concentrated in domestic assets. It considers regulatory and structural changes that have affected the degree of home bias. On the basis of a simple portfolio model, the likely effects of past home bias are also gauged. The chapter concludes with a discussion of the potential macroeconomic effects of continued reduction in home bias of Japanese investors, and of policies that might contribute to this reduction.

The chapter finds that deregulation over the past decade has eliminated many obstacles to outward investment. Changes in the government pension system and in government financial institutions have also favored greater holdings of foreign assets. As a result of these reforms, structural impediments to portfolio outflows no longer appear unusually high for a mature market country. Aversion to currency risk continues to be an important source of home bias, but some investors have recently become more accepting of this type of risk, as currency market volatility has decreased and domestic yields have remained low. Accordingly, reforms undertaken several years ago may have started to have a noticeable impact on portfolio outflows.

Previous research on home bias

Research on home bias frequently cites the international capital asset pricing model (ICAPM) as a point of reference. In most versions of the model, the mix and the relative proportions of assets held by an investor in the home country will be identical to those held by a foreign country investor. Accordingly, the share of own-country assets in each investor’s portfolio should be equal to the share of the domestic market in the world market—an allocation that is characterized as the absence of home bias. Empirical work in the 1980s and early 1990s found that this condition was far from obtaining (e.g. Tesar and Werner, 1995) even in the most internationalized markets. More recently, authors (e.g. IMF, 2005) have found a reduction in the degree of home bias in most mature asset markets (although some have argued that the decline of home bias in Japan has been slow relative to other developed economies). Nevertheless, investors still often show a marked preference for domestic assets.

A number of explanations have been suggested for the size and persistence of home bias, both in general and in Japan in particular. If some goods are nontradable, then under fairly general conditions, there will be some divergence from the ICAPM result on home bias.1 Another departure from the ICAPM predictions (believed to be particularly relevant for Japanese investors) relates to currency risk that cannot be fully hedged (Lewis, 1999). Less studied in the academic literature, but likely to be a practical problem, is the possibility that demographic or other differences among countries may lead to international differences in the liability structures of specific institutional investors, and thereby in efficient portfolio allocations.2 Other factors flagged in the literature as specific to Japan include the role of government financial institutions in absorbing domestic savings, the asset allocation of the National Pension system, the reporting requirements for capital outflows, and informal limits on the foreign asset holdings of institutional investors.

Measures of home bias in Japan

The standard measure of home bias in this study corresponds to the ICAPM concept tailored to accommodate differences in domestic market size. It is the share of foreign assets in the domestic portfolio, divided by the share of the foreign market(s) in the world market.3 So calculated, a foreign asset acceptance ratio (FAAR) of 100 percent for a given market and investor would indicate the absence of any home bias—the lower the FAAR, the greater the home bias. Because investor behavior may differ substantially between equity and bond markets, and because regulatory, tax, prudential, and other structural factors may affect the two types of portfolio assets in different ways, FAARs are computed (when the data are available) for both bonds and equities.

In 2003, the average FAAR in Japan across all types of investors was still relatively low by developed country standards, particularly for equities (Figure 9.1). Nevertheless, the 2003 figure represents an improvement with regard to the willingness to hold foreign equities and foreign portfolio assets in general.4

Figure 9.1Measures of Home Bias

a Balance-of-payments basis. Portfolio flows do not include acquisition of net international reserves by the authorities.

b A ratio of 100 percent corresponds to zero home bias.

Sources: IMF, World Economic Outlook (April 2005); Bank of Japan Ministry of Finance; Bank of Japan Flow of Funds; and IMF staff calculations.

Gross portfolio outflows reached record rates in 2003 and 2004, contributing to the increase in the FAAR (see Figure 9.1). One type of capital outflow not reflected in the FAAR measures is outward foreign direct investment (FDI), which peaked during Japan’s asset market boom in the late 1980s, exceeding portfolio outflows in 1990, before subsiding to an average of ¥2–4 trillion a year. Although FDI is potentially important as a means of increasing asset diversification and reducing home bias, the analysis in the present study focuses primarily on portfolio flows.

Institutional investors and the decline in structural sources of home bias

The past decade in Japan has been characterized by a decline in structural sources of home bias. Nevertheless, there remain some structural factors, most of which are common to most mature market economies, that incline domestic investors toward certain classes of domestic assets. This section discusses recent reforms, remaining structural sources of home bias, and recent portfolio flows trends, by class of investor.

Through 1998, most transactions on the capital/financial account were subject to the Foreign Exchange Control Law (FECL), which imposed fairly strict reporting requirements on capital outflows and prohibited certain types of foreign exchange-related derivatives transactions for all domestic investors. While the law did not prohibit most types of capital outflows, it was widely believed to have a dampening effect on several types of outward investment. The law was amended in Japan’s 1997–98 financial “Big Bang,” with the view of moving away from the notion of regulating foreign exchange transactions.

Although the FECL applied to all investors, many structural sources of home bias have been industry-specific, and many of those structural sources have been eliminated. Box 9.1 summarizes important regulatory and administrative changes in the past decade relevant to home bias in portfolio investment. As a consequence of these reforms, the regulatory environment faced by investors in Japan is now relatively open to acquisition of foreign assets, and, in general, does not appear to favor domestic over foreign assets. The following paragraphs detail the effects of specific reforms, and of recent economic developments, on holdings of foreign assets by type of investor.


Life insurers were among the first Japanese investors to purchase large amounts of foreign portfolio assets (see Figure 9.1). Japanese life insurers generally attempt to match the expected duration of their insurance liabilities with the purchase of longer duration bonds, as do life insurers elsewhere. Such bonds may be sovereign or corporate, foreign or domestic. Life insurers’ inclination to hold foreign bonds has remained fairly steady since 1990, even through the introduction of risk-based capital adequacy standards in 1996. The same institutions’ willingness to hold foreign equities vis-à-vis domestic equities has been lower than their willingness to hold bonds, even as the share of domestic equities in life insurance portfolios declined from over 40 percent in 1989 to less than 10 percent in 2004.

Box 9.1Regulatory and Structural Changes with Implications for Home Bias

1998Elimination of Foreign Exchange Control Law (replaced by Foreign Exchange Law).
1998Abandonment of informal asset allocation guidelines that indicated a limit of 30 percent of pension fund and life insurance funds in foreign assets.
1998Measure to allow the sale of investment trusts (including trusts concentrating on foreign assets) through bank windows.
2001Winding up of the Nempuku (public pension fund), which had invested primarily in domestic public works, and its replacement by the Government Pension Investment Fund, with an internationally diversified asset allocation target.
2001Introduction of Defined Contribution Pension Law, allowing the establishment of corporate defined contribution pension plans.
2004Measure to allow the sale of a wide range of securities through bank window (including sales of foreign shares).
2005End of unlimited deposit insurance on domestic (interest-bearing) demand deposits, increasing relative attractiveness of foreign currency deposits.
2005Revision of Financial Futures Exchange Law, reassuring domestic investors about the regulatory status of foreign exchange margin contracts.
2007Completion of the initial stage of privatization of Japan Post through conversion to a joint stock company.

The Japanese authorities have held discussions with international regulatory bodies (e.g. under the ongoing International Association of Insurance Supervisors (IAIS) and International Accounting Standards Board (IASB) initiatives) on solvency assessment and insurance accounting, which may be relevant for home bias. While no major changes in insurance regulations are anticipated in the near future, some convergence, for example to more demanding risk-based capital weightings, remains a possibility. That could have the effect of reducing the willingness of Japanese life insurers to hold foreign assets. Conversely, the difficulties that have been faced by Japanese insurers in meeting recurring liabilities promised through older savings vehicles may prompt some insurers to venture more into foreign markets in search of higher yields.

Market participants have reported that life insurers tend to respond to both the forward premium and the level of foreign exchange rates in determining what share of assets to allocate to foreign bonds—and how much of the allocation to leave unhedged. The forward premium rose in 2004 and the first half of 2005 with the increasing difference between short-term interest rates in Japan and the United States. As a result, the cost of hedging foreign currency exposure on investments in long-term bonds increased, and life insurers report that, while total holdings of foreign bonds have remained fairly constant, the share of unhedged bond holdings rose.

Pension funds

Public pension assets amounted to ¥147 trillion at the end of March 2004 (about 29 percent of GDP). Of the total, about half is currently administered by the Government Pension Investment Fund (GPIF), which began to take over management of public pension funds from the Trust Fund Bureau (TFB) in 2001. Previously, most of the public pension funds had been lent for public investment projects. Under the public pension reform of 2000, the funds were to be redirected from public investment to a conservative diversified target allocation (to be reached in 2008) that would include domestic equities and foreign bonds and equities, with professional fund managers engaged to do the detailed asset management. The target allocations for foreign bonds and foreign equities were raised by 1 percent each in March 2005 (Table 9.1).

Table 9.1Public Pension Portfolio Allocationa
Initial 2008 TargetRevised 2008 TargetActual 3/2005
Domestic bond share737275
Domestic equity share12118
Foreign bond share785
Foreign equity share896

All figures expressed in %. Actual 3/2005 allocation includes TFB component as domestic bonds.

Source: GPIF.

All figures expressed in %. Actual 3/2005 allocation includes TFB component as domestic bonds.

Source: GPIF.

About one-third of the total of ¥85 trillion in private pension funds (end of March 2004) are already devoted to foreign assets (see Figure 9.1). The share has risen steadily since 1998, partially in response to foreign exchange and pension reforms that included the abandonment of the informal “5-3-3-2” asset allocation guidelines.5 Although market participants do not perceive major changes now under way in the relative share of foreign assets held by defined benefit funds, the continuing transition from defined benefit to defined contribution plans may have the effect of continuing to increase the share of foreign assets held by private pension funds.6 As a greater share of pension money shifts to defined contribution, the need of fund managers to match the expected duration of liabilities diminishes, and their freedom to aim for higher risk-adjusted returns increases.7 This may also work to reduce home bias in private sector fund portfolios.

Fund managers of both defined benefit and defined contribution plans indicate that, while foreign currency assets can, in some cases, provide a useful natural hedge against inflation, Japan’s long experience with consumer price deflation has kept demand for inflation hedges at a minimum. Conversely, the tendency of the yen to remain in a relatively narrow trading range in recent years may have increased tolerance for currency exposure.

Mutual funds (investment trusts) and individuals

The amount of money in investment trusts is still somewhat low by mature market standards, at about 11 percent of GDP or ¥58 trillion at the end of 2004, but it has risen by 48 percent since 1998. Market participants consider this to be related to the “Big Bang” reforms of 1997–98, notably a provision in the reforms allowing the sale of investment trusts through banks. Investment trusts are not subject to any limits on foreign asset holdings, and do not have any specific liability structure driving asset allocation. However, they do have a need to provide risk-adjusted returns that conform to their customers’ retirement or other financial objectives.

Investment trusts’ exposure to foreign assets has recently increased, particularly in foreign bonds (see Figure 9.1). Market participants report that the share of foreign bonds held by investment trusts continues to rise as households seek out savings vehicles that provide steady income flows at higher yields. Other factors cited as supporting this trend include Japan’s persistently low money market rates, the maturing of higher-yielding domestic bonds, the end of blanket guarantees on domestic bank deposits, and the development of individual foreign asset-based mutual funds tailored to the preference of many older households for a steady stream of yen income.

Individual investors in Japan are also able to hold foreign assets directly and to hold foreign currency bank deposits, although for many types of foreign assets it may still be easier to gain exposure through investment trusts. Some market participants note resistance on the part of individual investors to direct foreign currency exposure. This inclination has been supported by the yen’s notable real appreciation trend over a 50-year span. Nevertheless, households’ acceptance of foreign exchange risk may be rising slowly, spurred in part by the yen’s relative stability in the past half-decade, the low level of domestic yields, and the end of blanket guarantees on savings deposits in April 2002 (the last factor may have reduced the relative appeal of yen deposits).8 A decline in the attractiveness of savings products available through government financial institutions (such as Postal Savings at Japan Post) may also have added to the relative appeal of foreign assets. Foreign currency deposits held by Japanese households rose from ¥1.2 trillion at the end of 1997 to ¥5.9 trillion by the end of 2004 (about 0.4 percent of gross household financial assets).

Estimated costs of past home bias in risk-adjusted returns

As first step toward assessing the extent to which the reluctance to hold foreign assets may have limited investment performance, this section provides an order-of-magnitude estimate of the gains that would have been available to domestic investors in shifting to the optimal portfolio for a given level of risk (the “risk reward” or “investment possibilities” frontier). The frontier is estimated empirically using historical returns and covariances among the major foreign and domestic asset classes available to domestic investors in Japan. This approach does not presuppose whether international diversification would have benefited Japanese investors.9 Actual portfolios are then located relative to the estimated frontier, to determine what changes in foreign asset holdings would have made the portfolios more efficient. The result provides a useful reference point for estimating the scale and cost of past home bias.

In contrast with the ICAPM structure, in which the “safe” asset is not necessarily assumed to be a domestic asset, this approach acknowledges the role of currency risk in portfolio selection in Japan by assuming that the safe asset is a one-year domestic bond with a known yield (this has been very close to zero in recent years). The returns on the other three asset classes (domestic equities, foreign bonds, and foreign equities) are determined from average one-year yen returns over the period 1981–2004. Foreign bond returns are based on one-year Treasury yields or one-year U.S. dollar LIBOR rates at the beginning of each year, adjusted for ex post currency movements. Foreign and domestic equity returns are based on historical one-year changes in the yen value of the S&P 500 and the TOPIX, respectively. Covariances among asset classes are estimated from historical data. The mean returns and covariances (represented as correlations) are provided in Tables 9.2 and 9.3 and Figure 9.2. These are used to generate the estimated risk-return frontier for domestic investors in Japan, with the straight line in Figure 9.2 showing the various combinations of expected return and volatility that would have been available to domestic investors.10

Table 9.2Correlation Matrix, 1981–2004a
Domestic EquityForeign EquityDomestic BondForeign Bond
Domestic equity1-0.150-0.59
Foreign equity-0.15100.61
Domestic bond0010
Foreign bond-0.590.6101

Domestic bond variance assumed to be zero.

Domestic bond variance assumed to be zero.

Table 9.3Actual and Alternative Portfolio Allocationsa



(incl. NIR)






Domestic bond share6965613146
Domestic equity share2221112016
Foreign bond share712244333
Foreign equity share22365
Expected return1.791.891.793.182.49
Standard deviation5.885.263.305.884.59

All figures expressed in %.

Sources: Coordinated Portfolio Investment Survey, Bank of Japan flow of funds; and IMF staff calculations.

All figures expressed in %.

Sources: Coordinated Portfolio Investment Survey, Bank of Japan flow of funds; and IMF staff calculations.

Figure 9.2Risk-Return Frontier for Japanese Investors

Source: IMF staff estimates.

The calculation suggests that there would have been gains in risk-adjusted returns in moving from the actual portfolio to the frontier. The gains could have been as great as a reduction of volatility by half, an increase in expected returns of more than 1 percent, or some combination of those outcomes.11 As shown in Table 9.3, shifting either in the direction of less volatility or in the direction of greater returns would have entailed substantial increases in holdings of foreign assets. Selecting the mid-point representing equal gains in reduced volatility and increased expected returns (last column of Table 9.3) implies that the foreign bond holdings would have increased from 7 percent to 33 percent and foreign equity holdings from 2 percent to 5 percent of their portfolios.12 From the perspective of individual institutional investors, this calculation is independent of assumptions about liabilities. Consequently, it does not take into account the incentives facing, for example, a life insurance portfolio manager seeking a good match between the duration of the firm’s assets and that of its expected liabilities.

Macroeconomic consequences of reduction in home bias

For Japanese investors, a reduction in home bias would entail higher portfolio outflows, and, based on historical data, higher average returns on investment. To gauge the likely impact on the balance of payments, one useful framework is the asset market model of Blanchard and others (2005), which allows for exogenous changes in the elasticity of substitution between foreign and domestic assets. In this framework, an increase in Japanese investors’ willingness to hold foreign assets (due, perhaps, to financial reform) would cause a shift in the relative price of foreign and domestic assets, in the form of an initial yen weakening. Japan’s current account balance would improve, due both to higher income inflows and to the trade effect of the yen depreciation.

Over time, however, the yen would appreciate gradually after the initial depreciation, to compensate foreign investors for continuing to purchase more Japanese portfolio assets than they had before the change in home bias.13,14 The value of the yen in the steady state would be higher than in the initial state, as higher net investment income from Japan’s improved net and gross foreign assets positions would allow the trade surplus to be lower than in the initial state. The effect on national income would be positive throughout. Such a sequence of adjustments is consistent with the idea that reduction in home bias may be an important channel for closing the funding gap for social security programs associated with an aging society.

While this framework is internally consistent, there are other conceivable scenarios, including an initial decline in the current account surplus if Japanese individuals increase domestic absorption in response to the opportunity to earn higher returns. Even in this case, however, the shift in investors’ relative preferences towards foreign assets would probably result in an initial exchange rate depreciation in response to the heightened outflows, in order to induce foreigners to hold more Japanese assets. The effect on national income would still be positive.

Policy considerations and conclusion

Policies that have the effect of continuing to reduce home bias appear desirable, both from a global efficiency perspective and in light of the need to raise average returns on assets held by Japanese investors. On the basis of Japan’s experience since the financial Big Bang, structural reforms that expanded the range of financial products available to the public, and that increased their ease of purchase, appear to have contributed to a fall in home bias. Further policy changes, for example in continuing to open bank channels to include sales of a broader range of insurance products than is currently permitted, promise similar benefits. Continued shifts from defined benefit to defined contribution pension plans may also tend to reduce home bias.15 Even among defined benefit plans, an increase in targeted levels of foreign asset holdings may be justified from both liability hedging and risk-return perspectives.

In the past decade, Japan has reduced or eliminated many of the administrative and regulatory sources of home bias, to the extent that the remaining structural impediments to acquiring or holding foreign assets do not appear out of the norm for a mature market country. Over the same period, aversion to currency risk appears to have lessened, although it remains an important factor in asset allocation. Due in part to these changes, gross portfolio outflows have accelerated since 2002 and have continued at that higher pace. Home bias has declined, in the aggregate and for investors such as mutual funds. Where home bias is still significant, it may in part reflect a natural lag in adjusting asset stocks, pointing to the possibility of further reductions in home bias in the future.


See, for example, Obstfeld and Rogoff (1996), pp. 319-25.

See, for example, IMF (2004a, 2004b).

This is measured as [(foreign assets held by domestic residents)/(domestic market capitalization + foreign assets held by domestic residents − domestic assets held by foreign residents)]/[(world market capitalization − domestic market capitalization)/(world market capitalization)]. This measure is also used in Bertaut and Griever (2004).

In the April 2005 WEO, this result is described as a relative lack of change in Japanese home bias, as the measure of home bias is implicitly computed as the difference between actual and benchmark holdings of foreign assets (the numerator and denominator of the FAAR), rather than as the ratio of the two. Because Japan’s equity market capitalization declined substantially as a share of the world market capitalization in the 1990s, the FAAR and the WEO measures yield divergent conclusions about the home bias trend in Japan.

In practice, a large proportion of asset allocations appeared to be guided by the following limits: at least 50 percent in safe assets (government bonds or bank deposits), and no more than 30 percent in equities, no more than 30 percent in foreign assets, and no more than 20 percent in real estate.

The Defined Contribution Pension Law of 2001 opened the way for the establishment of such plans.

Defined contribution plans still accounted for only about 1 percent of private pensions at the end of March 2004, but appear to have grown rapidly since then. As noted in Chapter III of the IMF’s September 2004 Global Financial Stability Report (IMF, 2004b), there is a sense in which defined contribution fund managers do not have a liability matching problem, since they are only committed to paying beneficiaries according to the eventual value of the fund’s investments. However, the fund manager’s fiduciary responsibility, and the goal of most fund beneficiaries, is to provide an adequate or desired level of income in retirement.

Chapter 10 of this volume provides an update on capital outflows and the yen carry-trade.

The theoretical conditions underlying the ICAPM may not obtain in practice. In addition, the theoretical model does not indicate the scale of gains from moving to a more efficient portfolio, or whether the gains would outweigh the associated transactions costs.

Because the frontier is depicted in risk-return (expected return, standard deviation) space, it is not possible to see directly from the location of a portfolio in that space what the composition of the portfolio is. For points on the frontier, however, there is a one-to-one correspondence with a specific portfolio mix of the four assets. Two of these frontier points are shown relative to the “actual” portfolio, with the composition of the portfolios identified in boxes in Figure 9.2.

Any shift to the north-west in Figure 9.2 represents a gain for a risk-averse investor, whatever the investor’s relative risk aversion. Holding expected returns steady and reducing volatility corresponds to a shift from point A to point B; holding volatility steady and raising expected returns corresponds to a shift to point C.

The actual portfolio in the first column of Table 9.3 does not include the government’s holdings of $673 billion in net international reserves (NIR) (end of 2003), most of which is held in foreign bonds. The second column shows portfolio expected return and variance computed to include the NIR.

If Japanese investors shift their portfolios from domestic to foreign assets, relative to the benchmark case, then foreigners must be induced through relative price changes to shift their portfolios in the direction of holding more Japanese assets—otherwise, asset markets will not clear. This requires offering a higher return to foreigners from holding Japanese assets than in the benchmark case, which requires that the yen be expected to appreciate over time. This requirement is met in the model with a one-time initial depreciation, followed by gradual appreciation.

Chapter 16 of this volume discusses the impact of capital outflow on the transitional dynamics of the yen-dollar exchange rate.

Chapter 6 of this volume touches on the state of play with the diffusion of defined contribution pension schemes.

    Other Resources Citing This Publication