The obvious strategy for an investor deciding to diversify internationally but not
wishing to determine how to construct an international portfolio is to hold an inter-
national index fund. The parallel to holding a domestic index fund is to hold a value-
weighted portfolio of international securities. The Morgan Stanley Capital Interna-
tional index excluding the United States is a value-weighted index, and an investment
matching this index would be a value-weighted index fund.^11
If expected return is related to a market index and if securities are in equilibrium,
then bearing nonmarket or unique risk does not result in additional compensation.
The way to eliminate nonmarket risk is to hold an index fund. Even an investor who
believes that securities are out of equilibrium but does not profess to know which se-
curities give a positive or negative nonequilibrium return (has no forecasting ability)
should hold the index fund. In this case, bearing nonmarket risk on average does not
improve expected return because the investor on average selects securities with zero
nonmarket return. Thus the investor should eliminate nonmarket risk by holding an
index fund. If there was good evidence that individual securities’ expected returns
were determined by an international equilibrium model, and if a value-weighted
index was the factor affecting expected returns, a parallel argument could be pre-
sented for holding an international value-weighted index fund. However, the evi-
dence in favor of any international model determining expected return is still contro-
versial.
A disturbing aspect of an international index fund is the proportion that Japan rep-
resents of the world excluding the United States (about 25%). If one believes in an
international equilibrium asset pricing model and Japan represents about 25% of the
market portfolio, then this is appropriate. Otherwise it makes sense only if Japan is
expected to have an abnormally high return; for diversification or risk arguments it is
clearly inappropriate. The authors have heard a number of presentations suggesting
other weighting schemes, such as trade or gross national product (GNP) that lower
the percentage in Japan. The correct justification for any weighting should come from
equilibrium arguments; otherwise any weighting is as arbitrary as another.
If one is not willing to accept an international equilibrium model that partitions
risk into that part that results in higher expected return and that part that is unique, it
is appropriate for an investor without an ability to forecast expected returns to mini-
mize total risk. The risk structure is reasonably predictable through time. The low
correlation on average among country portfolios, and the pattern of relatively high
correlation among countries with close economic links (such as the United States and
Canada) is likely to continue in the future. Both Jorion (1985) and Eun and Resnick
(1989) have examined the stability of the correlation structure and have found pre-
dictability. Thus the past correlation matrices can be used to predict the future. Sim-
ilarly, Jorion (1985) has shown that standard deviations are predictable, and thus a
low-risk international portfolio can be developed.
If one wishes to develop an active international portfolio, then many of the same
considerations are involved as are present in developing an active domestic portfo-
lio. However, international investment adds two elements to the investment process
11 • 22 INTERNATIONAL DIVERSIFICATION(^11) Although the Morgan Stanley index is the most widely used index, differences by country in the
cross holdings of securities (one company owning shares in another) means that its weighting is very dif-
ferent than an index using the value of a country’s equity assets. Japan in particular is very much over-
weighted. In addition, the Morgan Stanley index is a sample of each country’s securities and the propor-
tion sampled varies from country to country. Thus it is not an appropriately weighted market index.