In Eq. (5), n61
jt is the number of stocks in country j whose prices rise in period t,
n$08/
jt is the number of stocks whose prices fall, and T is the number of periods
used. We drop stocks whose prices do not move to avoid bias due to nontrading.
Thus, we de"ne fj as the average value of fjt, as defined in Eq. (1), across
periods. Values of fj must lie between 0.5 and 1.0.
Table 2 juxtaposes the ranking of countries by per capita GDP in Panel A
with their ranking by stock price synchronicity, as measured by fj , in Panel B.
Generally, high-income countries have asynchronous stock prices, and the U.S.
has the lowest fraction of stocks moving together, fj . In contrast, low-income
economies have the highest fj s. The "ve highest fjs are for Poland, China,
Taiwan, Malaysia, and Turkey. Our calculation of fj , the fraction of stocks in
each country that move together, are based on 1995 data. Our GDP per capita
variable is averaged over 1992 to 1994 to mitigate any transitory noise. Using
a three-year average of fjt gives similar results to those shown in Table 2.
An alternative way to distinguish firm-specific stock price movements from
market-wide price movements, following French and Roll (1986) and Roll
(1988), both of whom use U.S. data only, is to calculate the R2s of regressions of
the form