A model of demand in a repeated purchase situation

Abstract

Abstract Purpose – The purpose of this paper is to propose a model for consumer demand for vertically differentiated products. The approach is resolutely dynamic. Consumers are especially placed in situations of repeated purchases. They may reflect their past purchases in their decision. The suppliers adapt to the demand by amending their offer price. A simulation of the model of consumer behavior is made on the Champagne wine market. It results in a stable market that validates the theoretical choices. Design/methodology/approach – The methodology used is multi‐agent simulation. It is little used compared to cross‐cutting approaches such as multiple regression, joint analysis and constrained optimization. The multi‐agent simulation is a metaphor of the real world that makes virtual agents (consumers and suppliers) compete; these agents are provided with features that can vary over time and with predetermined behavior. This longitudinal approach allows in particular the capture of the effects of time on the choices observed in the habit phenomena, and also allows a description of nonlinear relationships. Findings – The use of a variable personal capital leads to the creation of a simple dynamic model of consumer behavior and fulfils the simulation of the demand in a market of vertically differentiated products. Originality/value – The originality of this work is based both on the formalization of the dynamic decision process and the methodology used, based on multi‐agent simulation. It helps to explore the evolution of the behavior of agents in the long term by taking into account past experiences. The simulation allows us to show that, in situations of repeated purchase, habits and involvement put into perspective the impact of salient cues of choice (reputation, price).

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 7 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.