Application of Agent Based Simulation to analyze the impact of tax policy on soybean supply chain

Abstract

This work seeks to explore and demonstrate the use of Agent Based Simulations (ABS) in modelling and simulating supply chains. Such methodology was applied to develop a model to evaluate the impact of current tax policies in soy supply chain in Brazil. The model brought interesting insights on how the country’s current tax structure induces logistics and tributary trade-offs, therefore generating a suboptimal grain distribution. This is accomplished by going through the conception and implementation of an Agent Based Model. First there is the definition and delimitation of the main agents acting upon soy’s supply chain, such as producers, trader and consumers. Those agents then have their behaviors studied and translated into programable patterns. Finally, the model considers the environmental interactions with the mentioned players, including the effects of infrastructure capacities, transportation costs, storage costs and tax legislation. After quantitative and behavioral validation, the simulation is then able to mimic the actual allocation of corn, soy and soymeal productions in their respective supply chains. This would allow inferring how the system could work in different tax conditions, thus quantifying the tributary impact in terms of congestions, idle infrastructure and delays. The analysis of such results points out that a path dependant tax system may induce agents to opt for inefficient logistic solutions, if such alternatives are cheaper when taking taxes into account. From those simulations it is possible to conclude that there are opportunities for supply chain efficiency gains in the design of a new tax policy.