11.4 RISK OF FOREIGN SECURITIES. Exhibit 11.3 presents the correlation between
the equity markets of several countries for the period 1991–2000. These correlation
coefficients have been computed using monthly returns on market indexes. The in-
dexes are computed by Morgan Stanley Capital International. They are market-
weighted indexes with each stock’s proportion in the index determined by its market
value divided by the aggregate market value of all stocks in that market. The indexes
include securities representing approximately 60% of the aggregate market value of
each country. All returns were converted to U.S. dollars at prevailing exchange rates
before correlations were calculated. Thus, Exhibit 11.3 presents the correlation from
the viewpoint of a U.S. investor. These are very low correlation coefficients relative
to those found within a domestic market. The average correlation coefficient between
a pair of U.S. common stocks is about 0.40, and the correlation between U.S. indexes
is much higher. For example, the correlation between the S&P index of 425 large
stocks and the rest of the stocks on the New York Stock Exchange is about 0.97. The
correlation between a market-weighted portfolio of the 1,000 largest stocks in the
U.S. market and a market-weighted portfolio of the next 2,000 largest stocks is ap-
proximately 0.92. Finally, the correlation coefficient between two 100-security port-
folios drawn at random from the New York Stock Exchange is on the order of 0.95.
The numbers in the table are much smaller than this, with the average correlation
being 0.48.
The correlations between international indexes are only slightly larger than the
correlation between two securities in the United States and less than the correlation
between two securities in most other markets. The correlations shown in Exhibit 11.3
are very similar to those found in other studies. Thus Exhibit 11.3 is representative
of typical correlation coefficients.^3 The numbers in Exhibit 11.3 are somewhat
higher than those found five years earlier, 0.48 rather than 0.40. This is primarily due
to the increased correlation among countries within the European Monetary Union
because of the elimination of exchange rates charges and greater integration of the
economies.
Exhibit 11.4 shows the correlation between the Salomon Brothers long-term bond
indexes of eight countries for the years 1990–2000. These indexes are value-
weighted indexes of the major issues in each country. Once again the correlations are
very low relative to the correlations of two intracountry indexes or bond portfolios.
The average correlation between countries shown in Exhibit 11.4 is 0.54. In contrast,
Kaplanis and Schaefer show an average correlation between countries of 0.43 for
long-term bond indexes in their sample period, and Chollerton, Pieraerts, and Solnik
(1986) find 0.43. This can be contrasted with the correlation between two typical
American bond mutual funds of 0.94 and the correlation between the U.S. govern-
ment and corporate bond index of 0.98.
Finally, Exhibit 11.5 shows correlation coefficients for short-term bonds, in par-
ticular, monthly returns of three-month debt. The average correlation for the same
eight countries shown in Exhibit 11.4 is 0.34. The low correlation across markets for
stocks, bonds, and Treasury bills (T-bills) is the strongest evidence in favor of inter-
11 • 6 INTERNATIONAL DIVERSIFICATION
(^3) Similar results have been found by other researchers. For example, Solnik (1974a) studied the 15-
year period 1971–1986 and found an average correlation of 0.35 between countries. Similarly, Kaplanis
and Schaefer (1996), studying the period February 1978–June 1987, found an average correlation of 0.32.
Furthermore, Eun and Resnick (1988), studying the period 1973–1982, found an average correlation of
0.41.