Virtual markets : the application of agent-based modeling to marketing science
- Publication Type:
- Issue Date:
Advances in any science ultimately depend on the creation of instruments that can create observations from which theories can be hypothesized and tested. This dissertation proposes that significant advances in marketing science can be realized with the engagement of the advanced computational science technique of agent-based modeling. To support this proposition, the methodology is examined from first principles to concrete implementation. The ontological and epistemological bases for agent-based modeling are developed, and the evolutionary science paradigm as it applies to marketing (for which the method is most useful), is reinforced with extensive analysis of evolved universal human behaviors, especially behavior relevant to marketing. The concept of the narrative framework is then posited. The primary property of the framework is the central role of choice as an expression of value and resource allocation. This framework then explicates the notion of virtual market, and a appropriate definition of agent derived. The computing requirements and skills needed to actually building a virtual market are also proscribed. Then a detailed, operational example of a virtual market is laid out. Called AirVM, it portrays the dynamics of the market for passenger air travel by simulating the product definition and ticket purchasing process for every passenger travelling on every regularly scheduled commercial flight in the world over a week time period – over 40,000,000 passengers flying on thousands of flights, offered by hundreds of carriers. The synthetic populations of passengers (customers) and airlines (sellers) have empirically-derived distributions of salient properties, called incidence distributions, which are described in detail with empirical data to support their formulation and parameter estimation. The computing logic and samples of the interface are presented, and the system critiqued using appropriate agent-based modeling criteria. The major contributions of the research are the verification of the ontological suitability of agent-based modeling to marketing science, the empirical confirmation of the evolutionary basis for marketing behavior, the conceptual structure for construction of agent-based models in market research, and the proscription of how to construct a virtual market, illustrated with a detailed example. There are also several contributions to the airline passenger industry that emerge from the work. Finally, the dissertation contributes another example of the application of the technique to the burgeoning literature on agent-based modeling.
Please use this identifier to cite or link to this item: