Understanding Consumer Reactions to Assortment Unavailability
Understanding Consumer Reactions to Assortment Unavailability
Understanding Consumer Reactions to Assortment Unavailability
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
changes in the price level do not influence any relative comparison across brands. We recognize<br />
that estimates for the development of category sales in the control s<strong>to</strong>res will be affected by<br />
promotions, so <strong>to</strong> integrate for the presence of promotions, we construct a promotional indica<strong>to</strong>r.<br />
Because we know that promotions occur in all s<strong>to</strong>res at the same time, we base the promotional<br />
indica<strong>to</strong>r on the <strong>to</strong>tal sales across all s<strong>to</strong>res. To identify the weeks in which a promotion of some<br />
sort <strong>to</strong>ok place, we estimate a model with a cubic spline function for <strong>to</strong>tal sales across all s<strong>to</strong>res.<br />
We assume that a promotion occurred for each observation with a large positive error. We then<br />
reestimate the same model, which now includes the promotion indica<strong>to</strong>r, <strong>to</strong> identify those<br />
promotions that had a smaller impact.<br />
4.5 Empirical results<br />
4.5.1 Analysis 1: Total category sales<br />
We first focus on the weekly <strong>to</strong>tal category sales for each s<strong>to</strong>re, which can be directly obtained<br />
from the database by simple aggregation. In Figure 4.2, we show time series plots for the<br />
category sales in each s<strong>to</strong>re, which demonstrate a slight decrease in sales for all four s<strong>to</strong>res. This<br />
overall decrease in detergent sales cannot be attributed <strong>to</strong> the delisting because, in the control<br />
s<strong>to</strong>res, the number of available items remained constant. To assess the actual effect of the<br />
assortment reduction, we must compare the changes in the test s<strong>to</strong>res <strong>to</strong> changes in the control<br />
s<strong>to</strong>res.<br />
In Table 4.4, we provide the parameter estimates for Equations 2a and b, with which we<br />
model the <strong>to</strong>tal category sales per s<strong>to</strong>re. As regressors, we include the promotional indica<strong>to</strong>r <strong>to</strong><br />
control for promotional effects, which will lead <strong>to</strong> a better fit in the models and thus a smaller<br />
residual variation. We also include a dummy variable <strong>to</strong> correct for an influential outlier that<br />
corresponds <strong>to</strong> a week of extremely low reported sales in one of the s<strong>to</strong>res. The retailer informed<br />
us that this was due <strong>to</strong> an error in the data collection system and that the actual sales were higher<br />
but that the exact figures were unknown. Although the s<strong>to</strong>res were selected in advance for their<br />
similarities in detergent shelf metrics, the estimated s<strong>to</strong>re intercepts show some differences in<br />
baseline sales across the four s<strong>to</strong>res, which may be explained by the unique characteristics and<br />
environment of each s<strong>to</strong>re. The most interesting results appear in the final lines of Table 4.4,<br />
which display the estimated function value of f(t|γ) and g(t|θ) at the chosen knot points, as well<br />
as the associated standard errors. The results clearly show that the effect changes over time and<br />
105