Food consumption regionality, seasonality, and sales promotion evaluation

Ronald Brian Larson, Purdue University

Abstract

Insufficient attention has been paid to regional and seasonal variations in food demand. Cluster analyses of consumption indices created market groups with regional patterns that were quite different from the geographies usually employed in demand studies. Regressions showed that the standard Census Bureau areas did not explain as much consumption variation as the clusters. This suggests that the Census boundaries should be revised or that new regions should be formed that represent the geographic patterns of food consumption. Although many studies allow demand intercepts to vary by geography or over time, there is evidence that slopes also vary. More interactions between regions or seasons and other variables should be added to demand models. A cluster analysis of seasonality indices produced regions with similar seasonal patterns, implying that region-season interactions may be important in some time-series analyses. Failure to follow recommended clustering procedures can reduce the likelihood that the true segment structure will be identified. Comparisons between the clusters constructed with standardized and unstandardized variables revealed major differences. Results using standardized variables were preferred. Because many demographic, psychographic, and geographic variables are related to food demand, marketers may boost profits by segmenting the market using these variables. Unfortunately, segmentation models do not suggest what tactics are best for each segment. Traditional techniques for evaluating price promotions compare their performance with unrealistic standards and may mislead marketers about the benefits from events. Without the promotion, prices and sales rates before the event are assumed to continue after it. Because firms are likely to change prices, traditional techniques will, in many cases, mis-estimate an event's profitability. A new model is proposed for evaluating and planning price promotions that allows price to vary and holds volume fixed with and without the event. This approach, based on price discrimination theory, uses demand relationships to predict the sales revenue without the promotion. Leakages, or limited arbitrage between segments, can significantly effect profits. When demand equations by segment are known, non-linear programming methods can solve the model for the profit-maximizing prices in each segment and promotional communication spending under different combinations of demand elasticities and leakage levels.

Degree

Ph.D.

Advisors

Akridge, Purdue University.

Subject Area

Agricultural economics|Marketing|Management

COinS