Modelling of Supply Chain Risk Management for Sago Starch Agro-industry

Abstract

The supply chain system is increasingly exposed to various risks that may interrupt product, fund, and information flows among supply chain parties. Those supply chain risks (SCRs) may contribute to losses for the actors of the supply chain. In response, the supply chain risk management (SCRM) framework is employed to manage SCRs effectively. Sago is a strategic commodity in Indonesia as the largest sago producer in the world. However, the sago starch agro-industry encounter crucial issues, including supply reliability, quality consistency, and price. These issues were related to the existing risks across the sago starch supply chain. This research aims to develop an SCRM model for the sago starch agro-industry. Specifically, it has achieved three objectives; (1) identification of risks in the sago starch supply chain, (2) modelling of uncertain and interdependent SCRs in the sago starch agro-industry, and (3) modelling of coordination strategy with risk-sharing schemes for the sago starch supply chain. This research took a case study of the sago starch agro-industry supply chain in Kepulauan Meranti Regency, Riau Province, as the largest sago starch producer in Indonesia. It focused on investigating the supply chain of dried sago starch. Risk identification is the first stage of the SCRM framework as the basis for the risk assessment and mitigation stages. In this case, the system characteristics of the sago starch supply chain were identified and analyzed. The IPO (input-process-output) of the investigated system was described, including the actor's requirements. Furthermore, risk identification was started with literature reviews to identify the possible risks in the sago starch supply chain. Next, the risk candidates were confirmed to the related industry experts. Finally, the relevant SCRs were listed and classified into supply, operation, logistics, demand, price-financial, environmental, and external risks. The risk assessment stage focused on evaluating the SCR impacts on the performance of the sago starch agro-industry. The uncertainty and interdependency of SCRs were modelled into the Bayesian network (BN), a class of a probabilistic graph model. In its application, the risk and performance variables were selected to be modelled. The expert knowledge was utilized to generate the dataset through Monte Carlo simulations and rule bases. The directed acyclic graph (DAG) of the BN structure is constructed by applying hybrid methods. The Hill-climbing algorithm, a search and score-based method, was applied to learn datasets given to prior knowledge (rule of links among variables. In the next stage, Bayesian inference was applied to risk scenarios with sensitivity analysis to examine the impact strength of SCRs on the performance measures. The results of the simulation indicated that the natural and external-risk factors, supply and quality of logs and water, inbound and outbound-logistics factors might contribute to the industry performance. This research also recommended mitigation strategies to eliminate or at least minimize the adverse SCR impacts. Finally, risk mitigation stage, this research developed the coordination models with risk-sharing schemes to mitigate uncertain supply risks between sago collector (supplier) and sago mill (agro-industry). The system identification and literature review were carried as the basis for developing those coordination models. The developed models include non-risk sharing (NRS), undersupply risk-sharing (URS), oversupply risk-sharing (URS), and hybrid risk sharing (HRS) schemes. The Stackelberg game was applied, and their associated optimal payoff values are obtained for those models. The simulation results indicated that each model generated a bit different payoff value for both actors and supply chain profits. Overall, the HRS model generated a higher payoff value than the other models under three supply realization scenarios.