How to Make the Most Productive Intervention in a Complex Economic System

Madsen, Björn (2015). How to Make the Most Productive Intervention in a Complex Economic System. PhD thesis The Open University.



Information about supply and demand propagates through supply chains in a queueing network with people and computers as batch information processors. As each batch processor delays propagation of information whilst pursuing optimal local decisions, the effect is delay and distortion of the information that is used to commit resources to actions in the supply chain. This thesis investigates the effect of delay and imperfect information as a source of error, to establish the case for change in research focus from optimal exploitation of physical constraints to optimal exploitation of information. In the context of real world supply chains, the thesis asks "How does one make the most productive intervention in a complex economic system?" and pursues a meta-intervention which perpetually minimises the discovered error-term. Evidence from literature indicates that agent-based modelling permits real-time peer-to-peer communication and distributed optimisation. Based on the literature the research project designs and develops an agent-based model which operates in real-time without batch-processes and can perform incremental multi-objective optimisation under realistic (chronologically progressive) conditions for decision making. The agent based model is then used to investigate two real-world supply chains, as case studies, which reveals a significant improvement of profitability and order-fulfilment. The thesis concludes that agent-based modelling is a very promising direction for "making the most productive intervention" as it reduces delay to a minimum. Finally it recommends that continuous improvement of decision making methods is a role better suited for humans, rather than operational decision making where computers cope much better with the high amount of detailed information.

Viewing alternatives

Download history


Public Attention

Altmetrics from Altmetric

Number of Citations

Citations from Dimensions

Item Actions