Investigating emergent nested geographic structure in consumer purchases: a Bayesian dynamic multi-scale...

Area of Science:

• Marketing Science
• Spatial Analysis
• Econometrics

Background:

• Spatial modeling of consumer response data is increasingly important in marketing.
• Existing models often lack integrated spatial and temporal dimensions.
• Understanding consumer behavior across geographic and time scales is crucial for businesses.

Purpose of the Study:

• To extend the spatial multi-scale model by incorporating both spatial and temporal dimensions.
• To develop a dynamic multi-scale spatiotemporal modeling approach.
• To analyze consumer purchase data and identify emergent geographic structures.

Main Methods:

• Empirical application using US company catalog purchase data (1997-2001).
• Development and application of a dynamic multi-scale spatiotemporal model.
• Utilized a scalable and computationally efficient Markov chain Monte Carlo (MCMC) method.

Main Results:

• Identified a nested geographic market structure transcending state borders.
• Revealed spatial clusters of consumers with similar spatiotemporal purchasing behavior.
• Demonstrated superior estimation and prediction performance compared to other spatial and spatiotemporal models.

Conclusions:

• The dynamic multi-scale spatiotemporal model effectively captures emergent geographic and nested structures.
• The model highlights the importance of dynamic patterns in multi-resolution analysis.
• The MCMC method enables the analysis of large spatiotemporal consumer datasets.