This is a replication file for Baillie and Bollerslev, "The Message in Daily Exchange Rates: A Conditional Variance Tale", JBES 1989, vol 7, pp 297-305. This does univariate GARCH models allowing for different means in the return based upon day of the week and different variances based upon day of the week and whether there was a skipped weekday. The GARCH model assumes that the unconditional variance (rather than the intercept in the GARCH process) changes based upon the dummies, so the sequential GARCH variance is more complicated than it would be if it was the variance intercept changing:
frml varf = alpha1*uu{1}+beta1*h{1}+$
w1*dd(1)+w2*dd(2)+w3*dd(3)+w4*dd(4)+w5*dd(5)+w6*skip+w7*skip{1}-$
(alpha1+beta1)*(w1*dd(1){1}+w2*dd(2){1}+w3*dd(3){1}+w4*dd(4){1}+w5*dd(5){1}+w6*skip{1}+w7*skip{2})
(For differing increments, eliminate the $ on the second plus all of the third line).
Baillie-Bollerslev JBES 1989 (GARCH with weekday effects)
Baillie-Bollerslev JBES 1989 (GARCH with weekday effects)
Last bumped by TomDoan on Mon Nov 20, 2023 8:30 am.