Investor Diversification and International Equity Markets

It measures the expected-return differentials required to rationalise observed international equity holdings; it does not estimate expected returns, and it does not test a specific model of why those differentials arise (§II). The authors are explicit about the identification problem that forces this design: “With limited historical data on international equity returns, it is difficult to measure expected returns … or to infer the optimal portfolio weights … with any precision. We can, however, make reasonable estimates of the covariance matrix.” So they fix the covariance matrix and preferences and solve the optimality condition backwards for the expected-return vector consistent with actual holdings. Stated in their own words, the contribution is to “use a simple model of investor preferences and behavior to show that current portfolio patterns imply that investors in each nation expect returns in their domestic equity market to be several hundred basis points higher than returns in other markets.”

Very concentrated, on estimates the authors themselves describe as crude (§I, Table 1). At the market level, “[t]he domestic ownership shares of the world’s five largest stock markets are: United States, 92.2 percent; Japan, 95.7 percent; United Kingdom, 92 percent; Germany, 79 percent; and France, 89.4 percent.” At the investor level, and for December 1989: “Japanese investors had only 1.9 percent of their equity in foreign stocks, while U.S. investors held 6.2 percent of their equity portfolio overseas. The British, by comparison, held 18 percent of their portfolio abroad, divided almost equally among the United States, continental Europe, and Japan.” The construction method is stated in a footnote – the estimates “cumulate the net purchases of equity by investors in each country, with adjustments for both stock market and exchange rate movements,” using the US Treasury Bulletin and Howell and Cozzini (1990) – and Table 1 is introduced as “crude estimates of the equity portfolio allocation.” The authors also distinguish their approach from Cooper and Kaplanis (1986), whose “calculations are largely imputations that do not rely on country-by-country equity flows.”

No, and the authors treat it as partly a size effect and partly a recent and policy-driven change (§I). “Since the United Kingdom is a smaller share of the total world equity market than the United States or Japan, it is not surprising that its investors hold more equity outside their own borders.” But they stress the timing: “the diversification of U.K. portfolios is a recent phenomenon. At the end of 1979, U.K. pension funds, which today hold 21 percent of their assets in foreign equities, held only 6 percent of their portfolios abroad,” and “[t]he growth of international equity investments followed Prime Minister Thatcher’s relaxation of capital controls.” The observation cuts both ways in the paper’s own argument: it is evidence that controls did bind historically, which is why the authors confine their claim about non-binding constraints to the present, granting that “capital controls sustantially restricted equity flows in the 1970’s.”

Enough that the authors call the foregone diversification nontrivial, on real returns computed from a US investor’s perspective with quarterly currency hedging (§II). “We compute real returns from the perspective of a U.S. investor, assuming the investor uses three-month forward contracts to lock in an exchange rate for the amount of his initial investment each quarter. The average pairwise correlation between quarterly returns on the equity markets in the United States, Japan, the United Kingdom, France, Germany, and Canada for the 1975-89 period is .502. This suggests that nontrivial risk reduction is available from cross-border holdings.” Importantly for the robustness of the exercise, the authors report that “[t]he correlations are similar if the returns are measured in yen or pounds, and whether or not the exchange rate risk is hedged” – so the diversification gains they price are not an artefact of the currency or the hedging convention.

Several hundred basis points a year, with the largest figure for British investors and the mechanism visible in the risk differences across markets (§II, Table 2 Panel A). “In the most extreme case, British investors must expect annual returns in the U.K. market more than 500 basis points above those in the U.S. market to explain their 82 percent investment in domestic shares. This large implied differential reflects the substantially higher standard deviation of returns on the British market, relative to returns on the U.S. and Japanese markets.” For the other two: “For U.S. investors, the annual expected return on U.S. stocks must be 250 basis points above the expected return on Japanese stocks. In contrast, for Japanese investors, the expected return on Japanese stocks must be 350 basis points above the expected return on U.S. stocks.” The implied levels in Panel A are modest in absolute terms – for example 5.5 percent on US stocks for a US investor and 9.6 percent on UK stocks for a British investor – so the paper’s claim is about the wedge between home and foreign, not about implausible absolute return expectations.

Yes, and the authors single this out as “striking” (§II). “Our estimates suggest that Japanese investors, for example, expect returns from Japanese stocks which are more than 300 basis points greater than the returns U.S. investors expect. There are similar differences in the expectations of foreign and domestic investors in both the U.S. and U.K. equity markets.” This is the sharpest form of the paper’s result, because a common set of beliefs – however unusual – could not generate the observed holdings simultaneously in all three countries; the implied beliefs must be investor-specific, which is what makes the behavioural reading of §III the natural one.

Most of it for Japanese and British investors; much less of it for US investors (§II, Table 2 Panel B). The authors themselves raise the objection that “the implied alternative of equal expected returns across all markets may not be an appropriate benchmark,” and so “estimate the expected returns that would induce investors in each country to hold an international value-weighted stock portfolio.” Panel B reports the gap between the two implied vectors, and “[t]he results again suggest that investors expect domestic returns that are systematically higher than those implied by a diversified portfolio.” Quantitatively: “U.S. investors’ concentrated holdings of U.S. stocks can be explained by ‘optimistic’ expectations of roughly 90 basis points. A similar ‘pessimism’ of about 110 basis points is needed to justify U.S. investors’ underweighting of the Japanese market. Explaining the behavior of both Japanese and British investors requires more ‘optimism’ regarding their own markets: 250 basis points for the Japanese, and over 400 basis points in the United Kingdom.” One sentence in the published text does not sit easily with its own table: it says the differences “for U.S. and British investors … are rarely larger than 100 basis points,” whereas Panel B’s British column contains a 4.4-percentage-point own-market entry, matching the “over 400 basis points” figure two sentences later. Read with the table, the small deviations belong to the US column.

Because withholding taxes on cross-border dividends are generally creditable at home, and the residual burden is an order of magnitude smaller than the differentials to be explained (§III). “Tax burdens that are higher on foreign than domestic equity income should lead investors toward holding domestic equity. There is little difference, however, between foreign and domestic tax burdens for most investors. Although all of the nations we examine impose a dividend withholding tax on payments to foreign shareholders, typically these payments can be credited against taxes in the investors’ home country.” The exception is handled quantitatively in a footnote: “Tax-exempt investors may face a burden from such taxes, since they have no tax liability against which to claim the credit. Even for these investors, however, the tax would only reduce expected after-tax returns in foreign markets by about 50 basis points” – well below the 250 to 500 basis points needed.

Because their sign is wrong and their scale is contradicted by gross flow volumes (§III). On sign: “The cost of trading may be lower in more liquid markets such as New York than elsewhere, but this should incline all investors toward the most liquid market, not toward their own domestic market.” The authors add a general-equilibrium point: “Since all shares must be held by someone, differences in transaction costs should be reflected in differences in expected returns.” On scale, they use turnover as the test: “The large gross equity flows across borders also suggest that transaction costs cannot explain why investors specialize in their home markets. For the United States in 1989, gross foreign equity purchases were fifty times net purchases.” Investors evidently trade across borders freely; they simply do not accumulate foreign positions.

They name concrete restrictions but argue observed holdings sit well inside them, with net selling by foreigners as the decisive evidence (§III). The examples given are that “[i]n France … a foreign investor may not hold more than 20 percent of any firm without authorization from the Ministry of Economy and Finance,” that “[i]n Japan, insurance companies cannot hold more than 30 percent of their assets in foreign securities,” and that “[m]any U.S. pension funds traditionally interpreted the ‘prudent man’ rule as limiting their degree of international exposure.” The argument that these are slack is behavioural rather than legal: “The current level of international portfolio investment seems to be well below any institutional constraint. In the mid-1980’s, for example, foreign investors were substantial net sellers of Japanese shares. Similarly, foreigners were net sellers of U.S. equities in 1988. Such reductions in international equity investments suggest that constraints on foreign holding are not binding, implying that incomplete diversification is the result of investor choices.” A ceiling cannot be what stops investors who are voluntarily moving away from it.

A contemporaneous survey of professional portfolio managers in Japan and the United States (§III). “Robert Shiller et al. (1990) report direct evidence on this question. In early 1990, they surveyed portfolio managers in Japan and the United States. The U.S. investors expected an average return of -0.3 percent on the Dow Jones Industrial Average over the next twelve months, compared with an expected return of -9.1 percent on the Nikkei. In contrast, Japanese investors expected an average return of 12.6 percent on the Dow, and 10.8 percent on the Nikkei.” The authors read the pattern carefully rather than overstating it: “While the Japanese investors were more optimistic than their U.S. counterparts with respect to both markets, they were relatively more optimistic about the Tokyo market” – a relative, not absolute, home-market optimism, which is exactly what their implied-return calculation requires.

Because the sampling uncertainty around mean equity returns is wider than the differentials in question (§III). “The statistical uncertainties associated with estimating expected returns in equity markets makes it difficult for investors to learn that expected returns in domestic markets are not systematically higher than those abroad. The standard error of the estimated mean annual return on the U.S. stock market, based on 60 years of data, is 200 basis points. Thus, the 95 percent confidence interval for the mean return spans 800 basis points. Because it is difficult to estimate ex ante returns, investors may follow their own idiosyncratic investment rules with impunity.” The authors pair this with a perception-of-risk channel: investors “may not evaluate the risk of different investments based solely on the historical standard deviation of returns,” and “may impute extra ‘risk’ to foreign investments because they know less about foreign markets, institutions, and firms,” citing Tversky and Heath’s finding that households treat unfamiliar gambles as riskier than familiar ones even when assigning both the same probability distribution. Country-specific closed-end funds, popular in the US in the late 1980s, “may overcome these fears.”

As one instance of a general pattern, which is part of why they favour a preference- or belief-based reading over an international-frictions one (§III). “The evidence of incomplete diversification presented here is consistent with evidence from many other markets. Ronald Lease et al. (1974) show that in the late 1960’s, many individuals held relatively few stocks. Both the mean and median in their sample of investors were close to eleven different securities. The rise of index mutual funds in the last two decades has improved the diversification of individual investors, but directly held equity still accounts for two and one-half times as much of household wealth as all mutual funds, of which index funds are only a small part.” Their closing example is domestic and non-financial: “Perhaps the most striking example of incomplete diversification is the tendency of most households to own residential real estate near where they work. The returns on their human and physical capital may consequently be highly correlated. This generates a much less diversified portfolio than holding, for example, a real estate investment trust with a national real estate portfolio.”

As an inference about which class of explanation must be doing the work, hedged on persistence (§III). The paper’s own summary claim is that “[t]he lack of diversification appears to be the result of investor choices, rather than institutional constraints” – “appears to be,” resting on the absence of an identified binding constraint rather than on a direct test of a behavioural model. The authors close by allowing that the phenomenon may be transitory: “Although the level of cross-border equity investment is low, it is growing and with time the international diversification puzzle may recede,” while maintaining that “[c]ross-border equity investment patterns may nevertheless provide important insights on how investors value risk and how they select portfolios.”