Chicken or the egg? Forecast and order adjustments in supply-chain planning (joint work with Enno Siemsen) click here
Many operational decisions increasingly rely on the guidance of (AI-based) algorithms. However, human judgment remains an essential element. In supply chain planning forecasters create demand predictions, which become inputs into inventory plans. They also adjust these demand predictions based on their judgment. Inventory planners also adjust the resulting orders based on their judgment. Should firms allow forecasts or inventory orders to be adjusted? In a series of behavioral experiments, we compare forecast to order adjustments. Surprisingly, order adjustments are a better avenue to incorporate human judgment. We examine the causes of this difference and discuss its implications.
Behavioral considerations of automation (joint work Iman Moosavi and Jan Fransoo) click here
This work explores human-algorithm interaction in a planning context, focusing on interactions resulting from task automation. Unlike typical augmentation interactions —where algorithms provide recommendations to human decision-makers for a single task— we investigate scenarios in which a human and an algorithm collaborate across sequential decision-making tasks. To examine this interaction, we designed an experiment involving a demand-forecasting task followed by a newsvendor inventory planning task. Our findings show that the behavior of both human planners and human forecasters change when they interact with an algorithm versus a human taking the other role. Furthermore, we show that automating the inventory planning task does not lead to an overall increase in profit due to the changes in the behavior of human demand forecasters who provide demand information for the planning task. Our findings highlight that when deploying algorithms in decision-making tasks, managers should consider not only the improvements within the specific task for which the algorithm is implemented, but also the behavior of human decision-makers in other parts of the organization who are responsible for related tasks.
Lead Time Optimism: The Impact of Supply Timing Uncertainty on Ordering Behavior (joint work with Simone Balvers and Jan Fransoo) click here
Decision-makers face different forms of supply uncertainty when making ordering decisions. This paper focuses on two forms of supply timing uncertainty: lead time uncertainty and review period uncertainty (i.e., uncertainty in the time between two order moments). We motivate the importance of these two forms of uncertainty from the nanostore retail channel. Using an experimental design, we find that subjects order more under lead time uncertainty than under review period uncertainty, for both high and low margin products. We identify as an underlying mechanism the tendency of decision-makers to underestimate the expected lead time, while this is not necessarily the case for the expected review period. Interestingly, the underestimation of the expected lead time is not mitigated by reducing the lead time uncertainty.
Volume guarantees in global health and economies of scale: the effect on current and future prices (joint work with Alexander Rothkopf) click here
Recently global health procurement organizations grant volume guarantees to pharmaceutical companies for the procurement of donor funded medicines and medical devices (e.g. pentavalent vaccine, contraceptive implants) and have been successful in achieving, among other things, lower prices. One main driver for substantial price reductions associated with volume guarantees are economies of scale. However, awarding one or more manufacturers a longer-term contract may inhibit recurring competition and also disincentivize new manufacturers to enter the market. This paper focuses on how the design of a volume guarantee (its size and split among incumbent suppliers) may affect competition and prices, currently and in the future, when technology is mature and fixed production costs are high. We find that due to economies of scale a) second period bids, a proxy for future prices, do not monotonically increase in the guarantee, b) first round bids, a proxy for current prices, do not necessarily decrease in the guarantee, as one would intuitively expect, and c) there is benefit from splitting the current volume among suppliers. Total procurement cost over both periods, and hence the success of the guarantee, depends heavily on whether or not suppliers behave strategically.