19.4. THECFO’S SUMMARYRECAPITAL BUDGETING 733
already mentioned, Gultekin, Gultekin, and Penati (1989) provide strong evidence
that theusand Japanese markets were segmented prior to 1980. However, they also
show that after the enactment of the Foreign Exchange and Foreign Trade Control
law in 1980, there is no longer any significant evidence against the hypothesis that
usand Japanese stocks are priced in an integrated market. A careful, and more
recent, test is by Dumas and Solnik (1991). They test the Solnik-Sercu International
CAPM, allowing for changes in risks and risk premia over time. Using data from major
OECDcountries, they reject Stehle’s hypothesis that the exposure risk premia,γi, are
zero, but they do not reject the full version (with non-zero risk premia for exchange
rate exposure). They also reject single-country asset pricing (with a purely local
benchmark). All of this lends support to the InternationalCAPM, at least for the
majorOECDcountries that do not impose explicit restrictions on capital movement.
There are also a few papers by De Santis and Gerard (1997, 1998) that allow for
autocorrelation in not just expected returns but also in variances and covariances,
modeling the fact that risk comes in waves. Their work confirms that exchange-rate
exposure is often non-zero and earns a statistically significant premium.
19.4 The CFO’s SummaryreCapital Budgeting
International asset pricing is potentially complicated by two extra issues: exchange
risk, and segmentation of capital markets. If the capital market of the home coun-
try and the host country are integrated, the cash flows of an investment project
can be valued in any currency, including the host currency. This simplifies capital
budgeting in the sense that no exchange rate forecasts seem to be necessary for
the translation. On the other hand, in integrated markets it becomes impossible
to observe the portfolio of risky assets held by the average investor in any of the
individual countries. Thus, the InternationalCAPMhas to be used, which means
that, in principle, exchange rate expectations and exposures still show up in the
cost of capital. In short, forecasts and exposures can only be eliminated by cutting
corners.
Thus, the first issue is whether or not there is integration. Having selected
either the single-countryCAPMor the InternationalCAPM, the next issue is to obtain
estimates of the model parameters. We need the stock market sensitivity or beta
and, in the InternationalCAPM, the exchange rate exposures. We also need the
expected return on the corresponding benchmark portfolios.
19.4.1 Determining the Relevant Model
If the capital market of the home country and the host country are segmented from
each other, the investing firm should set the cost of capital equal to the return that
is expected by its own shareholders. This means that a particular investment may
be profitable for a foreign firm but not profitable for a local firm.