Experimental Pricing As an Approach to Demand Analysis 31
of concentrate in each store may have exerted some influence
upon sales differences, intuitively, this effect would seem negli-
gible. Presumably, differences in store traffic did not contribute
to the store effect, because-it will be recalled-the analysis was
performed using per customer unit data.
In reference to the general demand function, the store effects
(6,,), which are expressed in terms of logarithms, may be easily
translated into percentage shifts in demand.33 The positive and
negative parameter estimates, of course, are to be interpreted
as upward and downward shifts in demand, respectively.
Week Effect.-The week effects (kI,.,s) designate proportional
shifts in demand attributable to differing demand conditions
among weeks within a particular price "age." As implied by
the signs of the week parameters, for a given price "age," condi-
tions prevailing in certain weeks served to shift demand upward
in some instances and downward in others. Naturally, both
positive and negative shifts within each age were to be expected,
because of the restriction imposed upon the analytical model
that the ^k-j-' (as well as the other categorical variables) sum
to zero.
Effect of Price "Age".-It will be remembered from the sta-
tistical analysis that the age of price had no significant effect
upon the slope of the demand function. However, price "age"
presumably did effect shifts in demand as reflected by the .
Since there were perceptible, though not significant, differences
in the slopes of demand functions fitted to the separate price
"ages" [see form (D)], perhaps a more sensitive model would
have yielded a demand curve for each price "age"-a result
apparently more in keeping with economic intuition.
Because slope differences in the demand function were not
manifested by the analysis, the .,s constituted the sole measure
of the "carry-over effect" of a change in price.34 Measures of
Percentage shifts in the general demand function may be determined
readily by use of the following simple formulae:
For positive 6j's: antilog(6 + 2) 100 = percentage upward shift in de-
mand,
For negative si's: 100 antilog ( 6 + 2) = percentage downward shift in
demand.
3 Technically, there would seem to be no difference between the two
concepts, age of a price change and age of a price, since one implies the
other.