Supply chain decisions are associated with design, planning and scheduling of various operations. Numerous analytical models and optimisation techniques have been developed and applied to tackle these issues. For example, analytical models have been used to assess and compare alternative solutions to production planning problems so that the preferred solutions can be selected and put into practice. Or sophisticated optimisation techniques have been used to address multi-objective and multi-criteria supplier selection problems.
The basic assumptions in all these tools and models are that the objectives and criteria are known in full, that they are quantifiable and prioritisable, and that a decision maker is perfectly rational and consistent. In practice, these are unrealistic assumptions. The use of tools and techniques with a strictly rational view may not be very effective when decisions are greatly affected by human behaviour. For example, studies show that decision makers are only “boundedly rational” and suffer from systematic biases. The term “bounded rationality” describes the phenomena where a person’s decision-making and logical reasoning is impaired by personal beliefs and cognitive limitations.
Of all the decision-making tools we have developed in the past decade, only about 20% have ever been fully implemented in the real world. The remaining 80% have been either impractical or used only to perform base analysis and forecasting the results of which are subsequently manually adjusted. This illustrates the shortcomings of the existing models which fail to take into account the behavioural influences. Designing practical decision making tools requires thinking outside the box and questioning the conventional assumptions.
For example, upstream supply chain, strategic sourcing decisions involve evaluation and selection of suppliers using a number of assessment criteria. The selection, measuring and weighting of these criteria typically involve cognitive biases due to the large number of influencing factors as well as the limited information processing capacities of individuals (i.e., when overloaded with information, individuals tend naturally to limit the amount of information taken into consideration, in this case the number of criteria or metrics). In addition, decision-making capabilities can be impaired when under stress. In particular, when things are going badly, people are more likely to make riskier decisions. A realistic supplier evaluation and selection model, therefore, needs to identify and eradicate such biases and cognitive limitations that can result in riskier decisions.
Another example is demand forecasting, downstream supply chain. Statistical system-generated forecasts often go through a number of manual adjustments made by forecasters and supply planners. Some of these adjustments can be essential and unavoidable to address exceptional circumstances and unexpected events (e.g., unforeseen sales promotions, climate/weather change, price change, product perishability, alterations in strategic plans, and new product developments). Yet, in most cases there are biases and inefficiencies in manual adjustments (such as over-optimism of the forecaster or over-reaction to a particular event) that may actually exacerbate the accuracy of the system-generated forecasts. There are approaches to identify and eradicate these biases, and improve the learning rate of forecasters for more effective adjustments. Some of these approaches may include the use of more advanced statistical forecasting systems that can standardise the magnitude and frequency of adjustments, more reliable market information, more coordinated S&OP and forecasting meetings for integrated forecasting decisions, and more effective training of forecasters.
Studies show that much of our decision-making is habitual and not a result of conscious deliberation. When learning to drive a car, a person focuses and pays conscious attention to all details. Once a seasoned driver though, he may find himself at the destination without any recollection of how he got there, nor the decisions he made during the journey. Similarly, supply chain decision-making can become routine, even though they may involve several biases and limitations. There are interventions to break such decision-making habits, the most effective of which is regulating behaviours (such as restricting a forecaster to no more than 10% data manipulation). Designing and implementing such regulations require a deep understanding of the contextual decision biases, the influencing factors, and the circumstances in which supply chain decision makers show more tendency to change their behaviours.
The Institute of Transport and Logistics Studies within the University of Sydney Business School has a research team specialised in investigating a range of topics in the area of behavioural supply chain decision-making. The team collaborates with companies in various industries to help supply chain practitioners make better behaviourally-informed decisions at the strategic, tactical and operational planning levels. Laboratory experiments, simulation studies, empirical field experiments, choice modelling experiments, and game theoretic approaches are used to identify the decision biases and inefficiencies, understand the underlying causes, and develop advanced behaviourally-informed decision support tools to effectively regulate human behaviour and change bad decision-making habits.