where n61jt is the number of stocks in country j whose prices rise in week t and
n$08/ jt is the number of stocks whose prices fall. For each country j we calculate
R. Morck et al. / Journal of Financial Economics 58 (2000) 215}260 219
f
US!f
j
. The variance of the estimate is approximately
f
US(1!f
US)
n
US
#f
j (1!f
j
)
n
j
, (2)
assuming that f
US and f
j are uncorrelated. By the Central Limit Theorem, the
statistic
( f
US!f
j
)
Jf
US (1!f
US)/n
US#f
j
(1!f
j
)/n
j
(3)
is approximately normal for su$ciently large sample sizes such as n
US and n
j
.
The null hypothesis that the fraction of stocks moving together in the U.S. is the
same as that in the emerging markets can be rejected in 43 out of 52 weeks for
China, 37 weeks for Poland, and 45 weeks for Malaysia. In contrast, the null
hypothesis can be rejected in only 18 weeks for Denmark and New Zealand, and
in only 2 weeks for Ireland.
The di!erences are economically as well as statistically signi"cant. Using the
weekly data for the whole of 1995, Eq. (1) shows that 79% of the stocks in China
move together in an average week. The same calculation indicates that 77% of
the stocks in Malaysia move together in an average week of 1995, as do 81% of
the stocks in Poland. In contrast, in the United States, Denmark, and Ireland,
the fraction of stocks gaining value in a given week typically barely exceeds the
fraction of stocks losing value.