00Thaler_FM i-xxvi.qxd

3 .Empirical Hypotheses and Tests

Our null hypothesis is that relative twin prices should be uncorrelated with
everything. Our alternative hypothesis is that markets are segmented, so
that relative market shocks explain movements in the price differential.
Specifically, we hypothesize that stocks that are most intensively traded on
a given market will comove excessively with that market’s return and cur-
rency.
To measure the relative comovement of twin prices, we regress the twins’
log return differential on U.S., U.K., and Dutch market index log returns
plus the relevant log currency changes:

(1)

where A and B represent the twin pair. Because of the cross-border aspects
of these markets, we include currency changes as well as local-currency
stock returns as market factors in Eq. (1). The null hypothesis is that all of

rSPFTSE DI

gl gl

titi

i

jtj

j

ktk

k

ltl

l

mtmt

m

AB, &

/$ /£ ,

−+

=−

+

=−

+

=−

+

=−

+

=−

=+ + +

++ +

∑∑ ∑

∑∑

α βδ λ

γυε

1

1

1

1

1

1

1

1

1

1

108 FROOT AND DABORA



  

   





  







  







  







  







  











Figure 3.3. Log deviations from SmithKline Beecham parity. Note: This figure
shows on a percentage basis the deviations from theoretical parity of SmithKline
Beecham H and E shares are traded on the NYSE. Data are from the Center for Re-
search in Security Pricing (CRSP).