Table
3.4
SmithKline Beecham Price Differentials and Market Movements
This table reports regressions estimates of the equation:where
rSKA
−SKB,
is the difference between the log returns of SmithKline Beecham A shares (London) and E shares (New York); S&P andt
FTSE, are returns on the S&P and
Financial Times
Allshare index, respectively, expressed in their native currencies; and $/£ represents
log changes in the dollar-to-pound exchange rate. Specification 1 includes leads and lags (shown above) to allow for nonsynchron
ous
trading. Specification 2 employs a more restricted set of leads and lags (based on actual time differentials). Specifications 3 a
nd 4 are the
same as Specifications 1 and 2, but include a lagged dependent variable on the right-hand side. Durbin’s Alternate H (DAH) is re
ported
in place of the Durbin–Watson (DW) statistic for Specifications 3 and 4. Specifications 5–8 employ 2-, 5-, 15-, and 50-day return
s. For
these specifications, leads and lags of independent variables are dropped. All regressions are OLS, with standard errors that al
low for se-
rial correlation and heteroskedasticity. Where there is only a single coefficient, standard errors are in parentheses.
Return DW or
Lagged
Specification
Horizon
R
2
DAH
DOF
Dep. Var.
S&P
FTSE
$/£
1, 7/89–12/95
1 day
0.221
2.70
1665
−
0.270
c
0.291
c
0.119
c
2, 7/89–12/95
1 day
0.216
2.69
1668
−
0.390
c
0.390
c
0.215
c
3, 7/89–12/95
1 day
0.311
−0.54
c
1665
−
0.335
c
−
0.508
c
0.458
c
0.212
c
4, 7/89–12/95
1 day
0.307
−0.43
c
1667
−
0.318
c
−
0.541
c
0.365
c
0.214
c
5, 7/89–12/95
2 days
0.118
2.70
834
−
0.466
c
0.409
c
0.184
c
rS
P
FTSE
titi
i
jtj
j
ltl t
l
SKA SKB,
$/£
−+
=−
+
=−
+
=−
=+
∧ +
+
+
∑∑ ∑
αβ
δ
γ
ε
1 1
1 1
1 1