If you have already read the first article in Part 6 you will be better prepared for Karen Lewiss focus on the phenomenon of investor bias towards domestic stocks "" a good example of behaviour inconsistent with what standard models would predict. She believes part of the explanation lies in a perception of foreign asset risk different to that derived from the historical standard deviation of returns.
The Mexican crisis of 1994 and 1995 is a good example of an event more adverse than the historical variance would imply moreover, poor returns in a foreign portfolio may be correlated with each other, such as happened with the tequila effect on emerging markets, and foreign rules and regulation can make analysis more problematic.
The author suggest that there is more to it than this, that the market-timing strategies pursued by many international investors complicate the issue of optimal portfolio choice and that those rebalancing their portfolios in response to changing events may face greater risk.
lDespite the apparent attractiveness of foreign investments, financial market investors display a puzzling bias toward their own domestic stock.
Research carried out over the past two decades suggests the foreign asset allocation of domestic investors is not consistent with what standard finance models would predict.
Basic investment theory highlights the diversification potential from holding assets that do not move in lock-step: if one security in a portfolio generates particularly bad returns in a given period, the other securities are unlikely to perform as poorly. Therefore, the lower the co-movement between the returns on securities in a portfolio, the better for diversification.
Equity and other financial markets across countries display just this type of pattern. Table 1 shows the correlation between returns on stock market indices converted into dollar terms. For instance, the correlation between the returns on the US index, measured by the S&P 500, and the returns on other stock markets range from 0.7 for Canada to 0.22 for Italy.
These correlations suggest that holding foreign equities should reduce the variability on the overall equity portfolio of a US investor.
Risk and return trade-off
One way to demonstrate this is to examine the trade-off between risk and expected return for a US investor holding a portfolio of the S&P 500 and a successively higher portfolio weight in a mutual fund of G7 countries "" Canada, Japan, the UK, France, Italy, and Germany. This trade-off is depicted in Figure 1 for a US investor.
The returns are converted into dollars and then adjusted for inflation with the US price level. At point A, the investors portfolio is 100 per cent invested in the S&P 500. At point C, the investors portfolio is about 15 per cent S&P 500 and 85 per cent foreign. At point D, the foreign share is even greater. Finally, at point Z, the portfolio is completely invested in the foreign mutual fund.
As Figure 1 shows, placing a proportion of about 15 per cent of the portfolio into this foreign mutual fund will minimise the standard deviation of the overall portfolio. If the investor is willing to accept the same volatility as the US market, then he should place an even higher fraction of his wealth into foreign stocks as at point D and earn an expected real return of 5.75 per cent, instead of 4.40 per cent.
Foreign investments appear even more attractive when some of the assumptions behind this simple example are relaxed. For instance, the figure assumes that the US investor places all of his foreign assets into the particular foreign mutual fund of the G7 countries. If he allocates this portfolio efficiently among all possible combinations of domestic and foreign stock markets according to the so-called efficient frontier, then the gains from foreign investment are substantially higher.
The benefits of foreign investment are greater still when the set of foreign assets is extended to include stocks from non-G7 countries such as the emerging markets, as well as bonds.
Despite the apparent attractiveness of foreign investments, financial market research has found that domestic investors hold a much smaller share of their portfolio wealth in foreign assets than investment theory would predict. Based upon aggregated equity data, French and Poterba (1991) found that
US investors hold less than 5 per cent of their wealth in the equity of companies in Japan and in the UK.
This evidence would correspond to a point like B in Figure 1: a portfolio allocation that would imply a higher volatility and a lower return than even the most conservative efficient foreign investment allocation at point C. Furthermore, French and Poterba found this phenomenon was not unique to the US. Residents in Germany, Japan, the UK, France, and Canada all demonstrated a similar home bias towards domestic stocks as well.
The apparent lack of international diversification is puzzling when global capital markets are integrating and governmental restrictions on international capital flows are being reduced. Financial researchers have examined explanations for this investor home bias, but have yet to provide a definitive answer to it. However, examining some of the plausible explanations helps to demonstrate both the strengths and the weaknesses of intuition based upon the simple methodology described thus far.
Extreme events
First, the risk on foreign assets perceived by investors may differ from that measured by the investment analysis. The investment analysis treats risk as the historical standard deviation of returns. The risk perceived by investors may differ from this for reasons ranging from irrational fears to rational, but alternative, characterisations of the risk.
If irrational fears are behind the investor home bias, then financial research will be unable to explain the bias. On the other hand, some of the perceptions of greater risk other than those measured by historical standard deviations may have some basis in observable facts. There are at least three reasons why these perceptions may be justifiable:
lThe probability of adverse events occurring may be greater than the historical variance implies.
lPoor returns in the foreign portfolio may be correlated with each other, especially during the periods of adverse events.
lRisk of foreign government expropriation.
These three possibilities are discussed below.
zThe assumption lurking behind the investment theory outlined above is that the distribution of returns can be fully characterised by their means and variances, as in the normal distribution. However, the distributions of both stock returns and the exchange rates used to convert the stock returns into domestic currency have been found to have fatter tails than a normal distribution. In other words, the likelihood that extreme events can occur is higher than a normal distribution would imply.
The likelihood of extreme negative returns is particularly high for emerging economy stock markets. For example, in the Mexican stock market decline of December 1994 through March 1995, the market declined by 30 per cent in Mexican peso terms. However, the return to dollar-based investors fell even more.
On December 20 1994, the peso was devalued by 15 per cent. Two days later, it was allowed to float and the peso subsequently depreciated in value by about 50 per cent against the dollar through the end of March. Thus, a US investor holding Mexican stocks would have lost both because of the stock market collapse and the peso depreciation. The cumulative loss from both currency and equity from the beginning of December 1994 to the end of March 1995 was 65 per cent. Domestic investors thus have two components to their foreign asset risk: a stock market risk and a currency risk.
The events following Mexico point to another reason why domestic investors may consider the perceived risks to exceed the historical variances: bad news in one market can generate capital losses in other markets. Uncertainty about the wisdom of investing in Mexico following the crisis induced a re-balancing of portfolios by many US investors out of other emerging markets.
This so-called tequila effect was felt through stock market declines not only in Latin American countries such as Brazil, Chile, Argentina and Columbia, but also as far away as the Philippines, Thailand, Indonesia and Malaysia. Thus, a domestic US investor holding foreign equities in emerging markets following the crisis would not only have suffered losses in his Mexican investment, but also in his entire emerging market portfolio.
Domestic investors leery of foreign investments often cite concerns about rules and regulations in foreign countries. Variations in accounting rules, required disclosures, and reporting styles from country to country make analysis that much more problematic.
These three justifications for perceived risk appear unlikely to explain entirely an investors home bias. Even if the probability of a calamitous event, such as a market crash, exceeds the probability implied by the historical variance, these possibilities seem more likely to explain a bias against emerging market funds than foreign funds as a whole. If emerging market funds are correlated during crisis periods, then domestic investors would tend to treat these stocks as a group within their portfolios.
But the home bias observation by domestic investors is that they skew their portfolio holdings away from foreign assets in general, including securities in industrialised countries such as Japan and the UK.
Moreover, even in the case of Mexico, the market decline was followed by a market rebound. Faster rebounds were experienced by the other emerging markets hit by the tequila effect.
Therefore, an investor with a buy and hold strategy would have weathered even this rather severe storm. Finally, instances of foreign expropriation of investments are the exception rather than the rule, particularly as more governments relax restrictions on foreign investment.
But is there still a way to understand the common perception that the risk is greater than the risk-return trade-off suggests? Yes, if the uncertainty surrounding this trade-off itself is considered.
Indeed, the international risk-return trade-off itself varies over time and may be rather unpredictable. For instance, Figure 1 is drawn assuming that the expected returns are the historical means of the inflation-adjusted stock returns while the volatility is given by the historical standard deviation of these returns. However, both stock returns and exchange rates are notoriously variable. Therefore, re-estimating the same relationship over a different time period can give quite different results. Clearly, the set of shares of the foreign asset that would be optimal differs considerably depending upon what is the appropriate trade-off.
Efficient frontier
We must ask then, how much confidence domestic investors can have in the international efficient frontier? One way to answer this question is to estimate the standard deviations around this trade-off itself. A recent study by Gorman and Jorgensen (1996) suggests the international efficient frontier has been extremely poorly estimated.
In fact, there is so much uncertainty about the position of the efficient frontier that it may not be possible to know whether observations of the actual portfolio, such as point B, are actually on this efficient risk return trade-off. What this means is that investors who seek to find the truly optimal allocation of their portfolio into foreign stocks may have a hard time determining this allocation. Uncertainty about the position of the efficient frontier implies the position on this trade-off is difficult to pin down.
On the other hand, the large movements of capital into and out of emerging markets during the early 1990s suggests many investors do not follow buy and hold strategies. Table 2 shows that over the 1990s portfolio investment in developing countries and countries in transition (such as the former Communist countries) has been quite volatile.
During the period from 1990 to 1993, flows into these markets from developed countries grew from $18.3 billion to $92.3 billion. Following the increase in US interest rates beginning in early 1994, this pattern reversed, with a decline to a $42.9 billion net increase in 1995.
The pattern is even more pronounced when we focus on the Western hemisphere. Capital inflows grew from $17.4 billion in 1990 to peak at $51.6 billion in 1993, only to drop to $10 billion in 1995. The table shows similar variability in portfolio investments in other parts of the world.
These movements are substantially more variable when viewed as net movements into US-based mutual funds invested in different parts of the world. In late 1994 and early 1995, some international funds experienced significant net outflows.
These different flows "" in response to such variables as changes in US interest rates and individual country events such as the Mexican peso crisis of December 1994 "" suggest that domestic investors may be trying to follow market-timing strategies.
Market-timing
If domestic investors are following market-timing strategies, then the distribution of the stock returns themselves suggest complicated issues for optimal portfolio choice. For instance, the standard mean-variance analysis treats these means and variances of international stock returns as constant, as they are in the long run. However, over short-run periods, the distributions of stock returns and currencies are known to display time-varying variances and means. Therefore, investors considering re-balancing in response to changing events may face significantly greater risk than that implied by looking at long-run means and variances.
While these qualifications imply that it may not be possible to pin-point where domestic investors should be on the efficient frontier, the relatively low correlation of international stock returns clearly suggests foreign stocks are important for diversification. As international capital markets become more integrated, this simple intuition would suggest that domestic investors should be holding more foreign assets. Transactions costs and perceptions of risk greater than measured risk may plausibly seem large to small individual investors, but these factors should be less important for institutional investors.
Indeed, if one examines the trend in foreign asset holdings instead of the level, domestic investors appear to be catching on to the benefits of international diversification. This trend is particularly striking for some institutional investors.
Table 3 and Figure 2 show the proportion of foreign securities held by these types of investors. In 1980, US pension fund managers held only 0.7 per cent of their portfolio in foreign securities. Just 13 years later, the fraction had increased to 5.7 per cent. This pattern is even more impressive for some of the other industrialised countries. The fractions of foreign securities held by these investors in the UK are large by comparison. In 1993, UK pension funds held almost 20 per cent, life insurance companies about 12 per cent, and mutual funds 36 per cent in foreign assets. Even the relatively conservative US mutual funds allocated 10 per cent to foreign securities.
While domestic investors continue to hold only a small fraction of their portfolio wealth in foreign assets, the overall trends suggest that they are becoming more aware of these securities as investment opportunities. As many small investors have found domestic mutual funds to be relatively low cost ways of diversifying their portfolio, these same investors have discovered the potential for investing in their foreign counterparts.
Whether the vehicle is a country fund comprised only of securities from an individual country or a global fund that invests only a fraction of its portfolio in foreign securities, growth in the mutual fund industry in the international capital market has significantly reduced the costs of foreign investing.
These investment patterns suggest that perhaps the home bias will disappear over time. Domestic investors may not be biased towards home markets they may just need time to learn about the costs and advantages of foreign asset allocation.
