The S&P 500 Index was up 48.00 percent over the ... returns of each asset, and the correlation between ... percent; the Pacific countries, about 16 percent; and.
PERSPECTIVES
Foreign Stocks in Behavioral Portfolios Meir Statman
T
he foreign stocks in U.S. investors’ portfolios have not done well lately. To some, such as Lowenstein (1997), the lesson for U.S. investors is that global investing is bunk. “You can lead a happy investment life without leaving home,” he wrote. But the real lesson from the recent performance of foreign stocks is about the construction of behavioral portfolios and the role that investors assign to foreign stocks within behavioral portfolios. Foreign stocks have not done well lately, and some have been especially terrible. The Philippines Composite Index was down 30.55 percent for the year ending on March 31, 1998, the Thailand SET 50 Index was down 36.24 percent, and the Kuala Lumpur Composite Index was down 40.19 percent. In contrast, U.S. stocks have done fabulously well. The S&P 500 Index was up 48.00 percent over the same period. Even those who do not succumb to the isolationist impulse are shaken. Are we investing too much in foreign stocks? What is the right amount? One road follows Markowitz.
Mean–Variance-Optimized Portfolios Markowitz taught us to consider a portfolio as a whole. He taught us to identify risk with the variance of returns and to focus on the risk and the expected returns of the whole portfolio, not on the risk and expected returns of the individual assets within the portfolio. The optimal allocation to foreign stocks in a Markowitz framework is the allocation that takes us to the mean–variance-efficient frontier. The location of the mean–variance-efficient frontier depends on estimates of the mean–variance parameters. The parameters consist of the expected return of each asset, the standard deviation of returns of each asset, and the correlation between the returns of each pair of assets. We do not know these parameters with precision, but imagine that we use long-term, realized parameters as estimates of expected parameters.
Meir Statman is the Glenn Klimek Professor of Finance at the Leavey School of Business and Administration, Santa Clara University. 12
Consider five asset classes: large-capitalization U.S. stocks (S&P 500), small-capitalization U.S. stocks (CRSP 6–10, the index of the smallest five deciles in the CRSP universe), foreign stocks (the MSCI Europe/Australasia/Far East Index), bonds (five-year U.S. T-bonds), and cash (30-day U.S. T-bills). The mean annual returns, standard deviations, and correlations for these asset classes for the 1969–97 period are presented in Table 1. Note that the volatility of foreign stocks was indeed higher than the volatility of large-cap U.S. stocks. Foreign stock returns had a standard deviation of 21.94 percent, whereas large-cap U.S. stock returns had a standard deviation of only 16.53 percent. However, the realized returns of foreign stocks were also higher—14.13 percent for foreign stocks versus 13.40 percent for large-cap U.S. stocks. Note, in particular, the 0.479 correlation between foreign stocks and large-cap U.S. stocks, which is a correlation lower than, for example, the 0.818 correlation between small-cap U.S. stocks and large-cap U.S. stocks. The low correlation between foreign stocks and domestic stocks provides important diversification benefits. What do estimated mean–variance-efficient portfolios look like? Table 2 presents the allocation of assets in five portfolios ranging from a “conservative” portfolio with a standard deviation of 5.84 percent to an “aggressive” portfolio with a standard deviation of 16.18 percent. The proportion of foreign stocks among all stocks in the conservative portfolio exceeds 43 percent. The estimated mean– variance-efficient conservative portfolio has 17 percent in large-cap U.S. stocks, zero in small-cap U.S. stocks, and 13 percent in foreign stocks. The high allocation to foreign stocks in the estimated mean–variance-efficient conservative portfolio is not unique. As Table 2 shows, the proportion of foreign stocks among all stocks never falls below 35 percent in any estimated mean–variance-efficient portfolio.
The Global Market Portfolio Portfolio allocations based on mean–variance optimization are sensitive to the estimates of the parameters. Changes in expected returns, standard deviations, and correlations lead to changes in the
Association for Investment Management and Research
Foreign Stocks in Behavioral Portfolios Table 1.
Mean Annual Returns, Standard Deviations, and Correlations for Five Asset Classes, 1969–97 Large-Cap U.S. Stocks
Small-Cap U.S. Stocks
Foreign Stocks
T-Bonds
Cash
Mean annual return Standard deviation
13.40% 16.53
13.99% 23.86
14.13% 21.94
8.85% 6.98
6.85% 2.63
Correlation Large-cap U.S. stocks Small-cap U.S. stocks Foreign stocks Bonds Cash
1.000 0.818 0.479 0.397 –0.067
1.000 0.425 0.255 –0.046
1.000 0.085 –0.189
1.000 0.233
1.000
optimal portfolio allocations, including the allocation to foreign stocks. Perhaps the allocation to foreign stocks in these estimated mean– variance-efficient portfolios is so high because realized returns were used in place of the true, but unknown, expected returns? One way to judge is to consider allocations that stem from another benchmark. That benchmark is the proportion of foreign stocks in a “global market portfolio,” a portfolio of all stocks, foreign and domestic, in the world. According to Morgan Stanley Capital International, the market value of U.S. stocks as of December 31, 1997, was only about 49.8 percent of the global market value of stocks. Foreign stocks made up the other 50.2 percent (Europe, about 32 percent; the Pacific countries, about 16 percent; and Canada, about 3 percent), which is more than the allocations to foreign stocks in our estimated mean–variance-efficient portfolios.
Behavioral Portfolios Markowitz taught us to build a mean– variance-efficient portfolio as a whole—that is, by taking proper account of the correlations between assets. But most investors fail to learn this lesson. Instead, investors build behavioral portfolios as pyramids of assets, in which layers of assets are associated with particular goals and particular
Table 2.
attitudes toward risk, and in which correlations between assets are overlooked. So, for example, some investors who insist on U.S. Treasury securities or certificates of deposit (CDs) for the “downside protection” layer of their portfolios also insist on foreign stock funds, initial public offerings (IPOs), and lottery tickets in their “upside potential” layer. The structural difference between a mean–variance-efficient portfolio and a behavioral portfolio is shown in Figure 1. Shefrin and Statman (1997) described behavioral portfolios, and Fisher and Statman (1997) argued that the model portfolios that mutual fund companies prescribe are behavioral portfolios. Investors who construct their portfolios as layered pyramids are overlooking the correlations between assets. In particular, they are overlooking the risk-reduction benefits that foreign stocks provide to the overall portfolio. Instead, they are focusing on the risk of each asset in isolation from all other portfolio assets. For example, here is how the risk of foreign stocks is described in the brochure that accompanies the model portfolios of Fidelity Investments: Foreign investments involve risks that are in addition to those of U.S. investments, including political and economic risks, as well as the risk of currency fluctuations. These risks may be magnified in emerging markets.
Estimated Mean–Variance-Efficient Portfolios, 1969–97
Asset Class/Statistic Large-cap U.S. stocks Small-cap U.S. stocks Foreign stocks Bonds Cash Total Mean annual portfolio return Standard deviation of portfolio return Allocation to foreign stocks among all stocks
March/April 1999
Conservative 17% 0 13 31 39 100% 9.50% 5.84 43.33
Moderately Conservative 25% 0 18 45 12 100% 10.68% 8.37 41.86
Moderate 39% 0 23 38 0 100% 11.82% 10.97 37.10
Moderately Aggressive 52% 0 28 20 0 100% 12.70% 13.24 35.00
Aggressive 57% 2 41 0 0 100% 13.71% 16.18 41.00
13
Financial Analysts Journal Figure 1. Structures of Mean–Variance Portfolios and Behavioral Portfolios Mean–Variance Portfolio
Behavioral Portfolio
Mean–variance portfolios are constructed as a whole, and only the expected return and the variance of the entire portfolio matter. Covariance between assets is crucial in determination of the variance of the portfolio.
Behavioral portfolios are constructed not as a whole but layer by layer, where each layer is associated with a goal and is filled with securities that correspond to that goal. Covariance between assets is overlooked.
Foreign Stocks
LargeCap U.S. Stocks
SmallCap U.S. Stocks
Upside-Potential Layer (contains, for example, foreign stocks, aggressive growth funds, IPOs, lottery tickets)
Bonds
Cash
Downside-Potential Layer (contains, for example, T-bills, CDs, money market funds)
Foreign stocks are not alone in the upside-potential layer. They compete for their share with the other risky assets in that layer. Indeed, foreign stocks are at a disadvantage in the competition for a share in the upside-potential layer because foreign stocks are less familiar than domestic stocks. Investors, like all people, prefer familiar bets over unfamiliar ones. Heath and Tversky (1991) reported that people who consider themselves experts on sports prefer to bet on sports and people who consider themselves experts in politics prefer to bet on politics. This preference remains strong even when both groups consider the odds in sports events to be identical to the odds in political events. The tendency of U.S. investors to overweight U.S. stocks in their portfolios is matched by the tendency of Japanese investors to overweight Japanese stocks and the tendency of British investors to overweight British stocks. The tendency has been dubbed the “home bias.” It is as if investors prefer to stay at home. Investors underweight all unfamiliar stocks, not only foreign stocks. Huberman (1997) reported that residents of New York State concentrate their telephone stock holdings in NYNEX, the local telephone company, whereas residents of California concentrate their holdings in Pacific Bell. Estimated mean–variance-efficient portfolios call for an allocation of more than one-third of portfolio stocks to foreign stocks. And the global stock market portfolio calls for an allocation of more than one-half the stock portfolio to foreign stocks. But lack of familiarity hampers foreign stocks in the competition for portfolio allocation. Allocations to foreign stocks in model portfolios of mutual fund companies are much lower than 14
one-half or even one-third. For example, the allocation to foreign stocks in the model portfolios prescribed by the Vanguard Group does not exceed 15 percent of all stocks. The model portfolios of Charles Schwab & Company prescribe a greater allocation to foreign stocks than the model portfolios of any other mutual fund company. Still, as Table 3 shows, even Charles Schwab prescribes a much lower allocation to foreign stocks than the allocation prescribed by mean–variance analysis. Financial advisors who advocate global diversification in the name of mean–variance optimization must be puzzled by the demands of many investors to drop foreign stocks from their portfolios. After all, the feature that makes foreign stocks attractive in mean–variance portfolios is the low correlation of foreign stock returns with returns of other portfolio assets. Surely, the returns experienced in the last few years have demonstrated that the correlations between foreign stocks and domestic stocks are indeed low: When U.S. stocks were flying high, Japanese stocks were burrowing under ground. So, why are investors so apprehensive about global investing? The answer, apparently, is that few U.S. investors bought foreign stocks for their mean–variance benefits. Investors nodded their heads as advisors explained the mean–variance benefits of foreign stocks, but they bought them because they thought that foreign stocks would be winners in the upside-potential layers of their behavioral portfolios. Investors thought foreign stocks would be a good upside-potential asset because they were extrapolating the high foreign stock returns of the 1980s into the future. Today, the same investors extrapolate that more recent dismal performance of
Association for Investment Management and Research
March/April 1999
15% 0 5 55 25 100%
Large-cap U.S. stocks Small-cap U.S. stocks Foreign stocks Bonds Cash Total
17% 0 13 31 39 100%
Mean–Variance Efficient 20% 10 10 45 15 100%
25% 0 18 45 12 100%
Mean–Variance Efficient
Moderately Conservative Portfolio (8.37%)a Schwab
Percentages in parentheses are portfolio standard deviations.
a
Schwab
Conservative Portfolio (5.84%)a
30% 15 15 30 10 100%
Schwab 39% 0 23 38 0 100%
Mean–Variance Efficient
Moderate Portfolio (10.97%)a
35% 20 20 20 5 100%
Schwab
52% 0 28 20 0 100%
Mean–Variance Efficient
Moderately Aggressive Portfolio (13.24%)a
Allocations in Charles Schwab Model Portfolios and Estimated Mean–Variance-Efficient Portfolios, 1969–97
Asset
Table 3.
40% 25 30 0 5 100%
Schwab
57% 2 41 0 0 100%
Mean–Variance Efficient
Aggressive Portfolio (16.18%)a
Foreign Stocks in Behavioral Portfolios
15
Financial Analysts Journal foreign stocks and conclude that upside-potential money belongs in Internet IPOs. Advisors respond to the complaints about foreign stocks in a number of ways. Some try more mean–variance education; they point out that foreign stocks were not always losers and are not likely to remain losers forever. Others reduce portfolio allocations to all foreign markets or to the hardest hit Asian markets. Still others try to make foreign stocks more familiar by using American Depositary Receipts in place of foreign stocks or by hedging the foreign currencies that correspond to foreign stock holdings.
Conclusion Investors like the idea of diversification. They like the idea of getting the highest possible expected return for the risk they are willing to assume. But risk has meaning only when looking forward into the future. Risk fades away when looking at the past. When investors look at the past, they do not want diversification and they do not want a reduction in risk. They want to have been in the best performing asset class. The best performing asset class over the last few years was the class of U.S. stocks. The last few years have been painful to investment advisors who advocated diversified global portfolios. Investment xenophobes like Lowenstein are triumphant as they bash advisors who encourage investors to venture out of the home market: Thus, the “sound” investor is defined as one who has moved a goodly chunk of his money out of the society he knows to countries with which he is unfamiliar, each according to their market weights. Indeed, not to put assets in such terra incognita is deemed to be “unsound.” More than one so-called expert has recoiled with horror at the news that my own family is geographically undiversified. We have no money in the former Yugoslavia, none in the present Ar-
gentina, none in the future Republic of Antarctica, none in Zambia, Belgium, or Kazakhstan.
Lack of familiarity and recent dismal returns make foreign stocks unattractive to U.S. investors. And mean–variance arguments emphasizing the low correlations between foreign stocks and domestic ones are not likely to change either fact. Typical investors who bought foreign stocks in the 1990s did not buy them for their mean–variance benefits. They bought them for the upside-potential layer of their behavioral portfolios; they expected foreign stocks to make them rich. The bad news for those who ventured into foreign stocks in the 1990s is that the returns were dismal. The good news for advisors is that investors are now familiar with foreign stocks. Familiarity makes foreign stocks more acceptable. Therefore, all that might be needed for an increase in the foreign stock allocations in U.S. investors’ portfolios is a sustained increase in the returns of foreign stocks. The Philippines Composite Index that was down 30.55 percent for the year ending March 31, 1998, was up 19.75 percent for the quarter ending March 31, 1998. The Kuala Lumpur Composite Index that was down 40.19 for the year was up 21.04 percent for the same quarter. McGough (1998) wrote: Some analysts have begun questioning the wisdom of putting money in foreign-stock funds. “But the strong showing of international funds in the quarter restores your faith in international investing,” said Susan Dziubinski, editor of Morningstar Investor, a newsletter. “There are opportunities in other countries.”
The author thanks Rosemary Macedo, Robert Murdock, Jennifer Clayton, Jonathan Scheid, and Weston Wellington for their help. He thanks the Dean Witter Foundation for financial support.
References Fisher, Kenneth, and Meir Statman. 1997. “Investment Advice from Mutual Fund Companies.” Journal of Portfolio Management, vol. 24, no. 1 (Fall):9–25. Huberman, Gur. 1997. “Familiarity Breeds Investment.” Working paper. Columbia University. Heath, Chip, and Amos Tversky. 1991. “Preferences and Beliefs: Ambiguity and Competence in Choice under Uncertainty.” Journal of Risk and Uncertainty, vol. 4, no. 1 (January):5–28.
16
Lowenstein, Roger. 1997. “‘97 Moral: Drop Global-Investing Bunk.” Wall Street Journal (December 18):C1. McGough, Robert. 1998. “Mutual Funds Are on a Roll, So Far.” Wall Street Journal (April 2):C1. Shefrin, Hersh, and Meir Statman. 1997. “Behavioral Portfolio Theory.” Working paper. Santa Clara University.
Association for Investment Management and Research