Agentenbasierte Simulation zur Analyse und Lösung betriebswirtschaftlicher Entscheidungsprobleme

Abstract

ZusammenfassungIn komplexen ökonomischen Systemen ergeben sich oft Fragestellungen bzw. Entscheidungsprobleme, die mit klassischen Analyse-und Optimierungstechniken nur schwer adressierbar sind. Märkte, auf denen Akteure mit heterogenen Zielsetzungen dynamisch interagieren, oder Unternehmen, in denen vielschichtige Prozesse ineinandergreifen, sind typische Beispiele für solche Systeme in unserer global vernetzten Wirtschaft. In den letzten Jahren hat mit der agentenbasierten Simulation (ABS) eine Technik zunehmende Verbreitung gefunden, die hilft, solche Systeme zu simulieren und inhärente Wirkungszusammenhänge besser zu erfassen. Im vorliegenden Beitrag stellen wir die Möglichkeiten der ABS insbesondere zur Analyse und Lösung betriebswirtschaftlicher Problemstellungen näher vor und geben eine systematische Übersicht von Anwendungsfällen aus der Forschung und der unternehmerischen Praxis. Abschließend analysieren wir den Anwendungsprozess zur Nutzung einer ABS, von der Konzeption und Modellierung bis hin zur Implementierung und dem Ablauf der Simulation.Abstract Complex systems frequently reveal decision problems which cannot be addressed by classical analytical or optimization methods. Markets consisting of players that interact with heterogeneous goals or enterprises with multi-faceted intertwined processes are typical examples of such systems within the global economy. During recent years, the agent-based simulation (ABS) has spread as a technology that helps to simulate such systems and better understand their inner causes and effects. In this work, we present the potential of ABS with respect to business applications and review a broad set of case studies from academia as well as from corporate practice.A. Deckert ( ) · R. Klein To conclude, we analyze the application process for the utilization of an ABS, spanning from the conception and modeling to implementation and the processing of the simulation.Schlüsselwörter Simulation · Agenten · Betriebswirtschaftliche Anwendungen

Simulation-based research in management accounting and control: an illustrative overview

“…Classical deduc-4 Please notice that we focus on the application of simulation techniques in research on management accounting and management control. Computerized simulations, however, are not only utilized in the context of research, but also for education purposes (Keys and Wolfe 1990) as well as in corporate practice (e.g., Deckert and Klein 2010;Hilgers 2008). A prominent example is presented by Wynder (2004), who used a business simulation in a second year management accounting course.…”

“…They claim that business simulations provide opportunities to develop higher-order thinking skills, which appears to be a critical factor of success in business. With respect to the application of simulation methodologies in corporate practice, Deckert and Klein (2010) review the literature and provide a set of case examples for both academic exercises closely related to corporate practice and studies that directly stem from practice. Exemplary fields of application are energy markets, production and logistics, retail markets, financial business, and telecommunication markets.…”

“…In their seminal paper, Davis et al (2007) provide five clusters of simulation approaches; namely (i) system dynamics, (ii) NK fitness landscapes, (iii) genetic algorithms, (iv) cellular automata, and (v) stochastic processes. 5 Deckert and Klein (2010) provide a rather general classification and distinguish between continuous, dynamic, and Monte-Carlo simulation methodologies (cf. also Domschke and Drexel 2005).…”

“…Typically, with system dynamics the system is modeled by a set of simple processes with circular causalities (causal loops) among them and the behavior of the system under investigation results from the interrelated processes. Basically, system dynamics-based simulation experiments are very close to the continuous simulation methodology (Deckert and Klein 2010). Typical research questions investigated with this class of models pertain to the effect of causality and system instability on a specific notion of performance.…”

“…In the context of management research, this class of models is frequently used for research on diffusion processes (see, e.g., Goldenberg and Efroni 2001;Kiesling et al 2012). Classes (ii)-(iv) usually consider the adaption of a system or individuals over time, they can be classified as dynamic simulation methodologies (Deckert and Klein 2010).…”