Modeling complex entrepreneurial processes: A bibliometric method for designing agent-based simulation models

摘要

Purpose: The purpose of this paper is to suggest a bibliometric method for designing agent-based models (ABMs) in entrepreneurship research. The application of this method is illustrated with an exemplary agent-based modeling and simulation (ABMS) regarding the early venture growth process. This bibliometric approach invigorates the utilization of ABMS as a viable research methodology in process-oriented entrepreneurship research. Design/methodology/approach: In the bibliometric method, a domain corpus composed of scholarly articles is established and systematically analyzed through co-word analysis to discern essential concepts (i.e. agents, objects, and contexts) and their interrelations. The usefulness of the bibliometric method is elucidated by constructing an illustrative ABMS. Findings: The bibliometric method for designing ABMs identifies essential concepts in the entrepreneurship literature and provides contexts in which the concepts are interrelated. The illustrative ABMS based on these concepts and interrelations accurately and consistently reproduces the emergence of power-law distributions in venture outcomes consistent with empirical evidence, implying further merit to bibliometric procedures. Practical implications: The proposed method can be used not only to build simple models with essential concepts, but also to build more complex models that take a large number of concepts and their interrelations into consideration. Originality/value: This study proposes a bibliometric method for designing ABMs. The proposed method extends similar procedures that are limited to thematic or cluster analysis by examining the semantic contexts in which the concepts co-occur. This research suggests that ABMS from bibliographic sources can be built and validated with empirical evidence. Several considerations are provided for the combined utilization of the bibliometric method and ABMS in entrepreneurship.