Simulation modeling in organizational and management research

Abstract

Simulation modeling provides a powerful methodology for advancing theory and research on complex behaviors and systems, yet it has been embraced more slowly in management than in some associated social science disciplines. We suspect that part of the reason is that simulation methods are not well understood. We therefore aim to promote understanding of simulation methodology and to develop an appreciation of its potential contributions to management theory by describing the nature of simulations, its attractions, and its special problems, as well as some uses of computational modeling in management research.Managerial behaviors and organizational outcomes increasingly are recognized to be the result of the interaction of multiple interdependent processes. Progress in understanding these phenomena depends, in part, on the ability to incorporate more complexity into management theory and to conduct research on the consequences of the resulting theory.Traditional approaches to theory development are limited in their ability to analyze multiple interdependent processes operating simultaneously. Even when the individual processes are well understood, analyzing their interdependent behavior poses difficulties, because the processes involved may interact in complicated and unforeseen ways. And because the interactions typically produce nonlinear system behavior with feedback, empirical analysis using the general linear model has limited value, especially when (as is typical) samples are sparse in the regions of greatest interest.In studying the complexities of managerial and organizational behavior, a more systematic methodology for theory development and analysis may prove useful. Specifically, we believe that simulation or computational modeling has unique advantages in this respect (Axelrod, 1997;Taber & Timpone, 1996). Well suited for the study of complex behavioral systems, simulations show greatest utility for gaining theoretical insight through developing theories and exploring their consequences (Cohen & Cyert, 1965).Yet researchers in the academic field of management have been slow to take advantage of simulation methods. Scholars in some related social science disciplines, most notably psychology, seem to be far ahead, and the application of simulations by management practitioners to set policy and study organizational problems is quite extensive (Carley, 2003). So management theorists have the opportunity to benefit by taking fuller advantage of simulation methods.Our aim in this article is to provide an explanation and overview of simulation methodology. By doing so, we seek to encourage management scholars to become users of simulation methods and to become better informed consumers of simulation-based research. An understanding of what simulations are and how they work is a prerequisite for appreciating the potential contributions of simulation analysis to management theory, as well as for identifying problems and shortcomings in simulation work.We thank Xia Zhao for research assistance and the anonymous ...

Agent-based modeling in managerial science: an illustrative survey and study

“…In any case, since simulation requires putting the system under investigation into a computational form, this also means that the models are precisely formalized (for a further debate on this point see Leombruni and Richiardi 2005;Epstein 2006b;Harrison et al 2007;Richiardi 2012). However, formalization and computational representation inevitably require finding abstract formulations of the system under investigation.…”

“…For this, we find it helpful to start with the particular strengths that simulation-based research is regarded to unfold in managerial science. As such, it has been argued that simulation in general, not only the agent-based approach, is particularly effective for developing theory ''when the research question involves a fundamental tension or trade-off'' (Davis et al 2007, p. 485) like, for example, short-versus long-term or chaos versus order, and when ''multiple interdependent processes operating simultaneously'' occur (Harrison et al 2007(Harrison et al , p. 1229; in a similar vein Davis et al 2007, p. 495).…”

“…For example, employing the idea of NK fitness landscapes requires that the underlying optimization problem can be appropriately represented in terms of an N-dimensional binary vector optimization problem, which is obviously a strong abstraction of many real-world problems. Hence, a major question of simulation is the extent to which the simplifications and abstraction which are prerequisites of a computational representation may induce a critical lack of external validity (Harrison et al 2007;Davis et al 2007).…”

“…Hence, ABM could be regarded in the light of simulationbased research in general as a method for theory development. According to Harrison et al (2007) simulation modeling allows researchers to capture ''complex multilevel, and mathematically intractable phenomena'' (p. 1240) by taking prior management theories and empirical data into account within the model and using computational technology; the computational experiments produce data which could bring about new empirical research (not least for validating and testing the simulation model) and new management theories-and these could be subject to further simulation research. Davis et al (2007) argue that simulation is especially useful ''in the 'sweet spot' between theory-creating research using such methods as inductive multiple case studies… and formal modeling…, and theory-testing research using multivariate, statistical analysis'' (p. 481), when the research question addresses some fundamental tensions as exemplified in Sect.…”

“…The latter means that simulation is used to develop structural insights and gain understanding of fundamental processes within a certain area of interest (e.g. Davis et al 2007;Dooley 2002;Harrison et al 2007;Gilbert and Troitzsch 2005). Axelrod regards simulation ''a third way of doing science '' (1997a, p. 3) or a ''third research methodology '' (1997b, p. 17) which has in common with deduction the explicit set-up of assumptions (though it does not prove theorems by the ''classical'' mathematical techniques) and which generates data from a set of rules (rather than from measurement of real world data as it is typical for induction).…”