An agent based model of the evolution of supplier networks
Authored by David C Earnest, Ian F Wilkinson
Date Published: 2018
DOI: 10.1007/s10588-017-9249-1
Sponsors:
No sponsors listed
Platforms:
No platforms listed
Model Documentation:
ODD
Pseudocode
Model Code URLs:
Model code not found
Abstract
We view supply chains as a type of complex adaptive system and develop
an agent based computer simulation model of the evolution and
performance of supply chains based on Stuart Kauffman's NK models of
fitness landscapes. Firms operate in networks in which they supply
products to some firms and source inputs from others. They seek to
maximize their own performance but they cooperate with other firms to
gain access to inputs. We model firm performance in terms of the fit of
its product with market demand and the contribution from first tier
suppliers. The model uses genetic algorithms to mimic the way firms
learn and adapt their products and supplier networks for more and less
complex products and different switching conditions. We find that (a) as
the complexity of the product increases, firms perform less well; and
(b) firms build supplier networks with higher average in-degree, greater
density, and significantly greater clustering to cope with product
complexity. Our findings suggest that firms using highly specific assets
or that face high switching costs are likely to pursue a supplier
strategy that relies more on multiple suppliers and more clustered
supply networks. Also, in industries characterized by highly specialized
training, plants and machinery dedicated to specific products and other
high product-specific transaction costs, we should observe more
specialization at low levels of product complexity but less at high
levels. The model contributes to our understanding of the evolution of
supply networks, which is an under-researched topic, provides the basis
for further extensions of the model and the development of more
realistic models of actual supply chains. The model also provides a
conceptual and methodological tool to assist firms and policymakers to
better understanding the nature of supply chains and to identify and
test strategies and policies.
Tags
Simulation
Agent based models
Evolution
Complex adaptive systems
Dynamics
Supply chains
perspective
context
Sociology
Business networks
Nk
fitness landscapes
Genetic
algorithms
Chain networks