732 CHAPTER 19. SETTING THE COST OF INTERNATIONAL CAPITAL
that the French, Japanese, Korean, and Mexican markets are at least partially
segmented from theuscapital market. Hietala (1989) studies the effects of the
Finnish law that prevented investors from investing in foreign securities and finds
that there is a significant difference between the returns on domestic assets required
by residents compared to foreigners. Gultekin, Gultekin, and Penati (1989) find
strong evidence that theusand Japanese markets were segmented prior to 1980.
However, while there were substantial controls on capital flows before the 1980s, this
is no longer true. Halliday (1989) already reports that even in those days there were
few constraints on investing in foreign stock markets. This was and is especially
true for investing in the markets of developed countries. For example, already in
the 1980s there were no controls on international investment into or from Austria,
Belgium, Denmark, Ireland, Italy, Japan, the Netherlands, theuk, theus, and
West Germany. The controls studied by Hietala (1989) and Gultekin, Gultekin,
and Penati (1989) were removed in 1986 and 1980, respectively. Also, looking at
restrictions that limit domestic residents from investing abroad, one sees that these
constraints are often not binding. For example, Fairlamb (1989) reports that in 1988
only 8 percent of Spanish funds were actually invested in foreign assets, while the
limit was 30 percent. Thus, while direct controls on foreign investment may have
been important in the past, they are probably no longer an important determinant
of portfolio choice and asset pricing in the mainOECDcountries.
Let us now discuss the more direct tests of international asset pricing models.
Solnik (1973), who did the first theoretical and empirical work in international asset
pricing, tests a special case of Equation [19.36], where the world market risk premium
and exposure risk premia could be merged into one single term. He concludes that
the data are consistent with his InternationalCAPM, although he does not test his
model against the single-country alternative.
An early test that does compare an international asset pricing model against
the single-country alternative was carried out by Stehle (1976). Specifically, Stehle
tries to find out empirically whetherusstocks are priced in a national market or
in a world market. He, too, uses a restricted version of Equation [19.36], assuming
thatλequals unity so that all currency risk premia disappear. The only remain-
ing difference between the international model (Equation [19.36]) and the national
model is the definition of the market portfolio. Specifically, in Equation [19.36], the
benchmark portfolio is the world market portfolio, while in Equation [19.24], it is
the national market portfolio. Stehle’s tests are not able to empirically reject one in
favor of the other, and Stehle concludes that asset pricing is done in a single-market
context. Dumas (1976), however, argues that when the data do not allow one to
distinguish between single-country asset pricing and international asset pricing, then
one ought to retain the simplest view—that is, one should conclude that there is
one international market instead of the many separate national markets.
There have been many additional empirical investigations, with a large portion of
them testing special restricted versions of Equation [19.36]. The conclusions tended
to be ambiguous. But more recent work has come up with more definite answers. As