i s , f o r e a c h f i r m , w e t e s t t h e j o i n t h y p o t h e s i s , H 0 : 0 = = = = £ ¥ $ E E E E e u r o U S . 1 3 C o l u m n 5 o f

# Table 1 gives the fraction of firms in each country for which the hypothesis is rejected at

the five percent significance level. That fraction is well above five percent in most of the

countries. In those countries with many of their firms exposed to individual currencies –

# Indonesia, Korea, Malaysia, and the Philippines – the fraction is quite high, above 40

percent. Column 6 gives the median R ^{2 }of the regressions in each country; and column 7

gives the number of firms we observe in each country. Of course, as mentioned above,

these exposures are “residual” exposures. That is, because they are conditioned on the

local excess return, they do not reflect aggregate exposure experienced by the local market

as a whole. We estimate the exposure inclusive of the aggregate local effect in Section 4.

# Before proceeding to an analysis of exposure over time we address one potentially

important issue. Namely, we would like to know whether the results in Table 2 are

spurious. In particular, it is possible that we would find ‘significance’ in similar

regressions using four purely random variables in place of the returns on the dollar, the

euro, the yen, and the pound. That is, since we include all four exchange rates in the

regressions, we may be biasing upward our chances of statistical significance. To address

this concern we conduct a series of Monte Carlo simulations by substituting pseudo-

exchange rate changes in each regression. These simulations, reported in Appendix Table

3, confirm that the evidence in Table 2 is not spurious. Using simulated data we fail to

find abnormal levels of ‘exposure’. We now turn to an examination of changes over time

in measured exposure.

13 Normally an F-test, the test of exclusion restrictions is here a Ȥ^{2 }test since we use a covariance matrix that allows for heteroscedasticity and serial correlation.

8