The existence of a price promotion of a brand during a specific week is captured by
observing the price levels of the 10 brands selected and assuming that a significant
decrease in the price of a brand for a limited period (one or two weeks) must
correspond to a promotion. Using one variable for the regular price and another for
the price promotion is the best way of handling this promotional variable, as
Mulhern and Leone (1990) indicate. In this respect, to analyze brands’ relative
promotional discount size (in those weeks in which a price promotion was actually in
place), we calculated the difference between the price (in promotion) of each brand
that week and the price from the previous week when no price promotion was in
place, and divided this by the price from that previous week. When a brand is not
being promoted this variable equals zero, as both prices coincide.