Knowledge that Transforms

To mitigate supply and demand uncertainties, firms often design highly flexible production networks. This study investigates process flexibility in production systems subject to supply disruptions and stochastic demands with differentiated profit margins. In particular, we model supply disruptions as failures of arcs in the network, with node-based disruptions treated as a special case in which all arcs connected to a node simultaneously fail. To address the cost implications of such disruptions, we investigate the design of process flexibility under a robust optimization framework. We first develop a greedy algorithm under deterministic demand to efficiently evaluate the worst-case disruption scenario, and demonstrate the significant advantage of the alternate long-chain design under such disruptions. Subsequently, under stochastic demand, by introducing the marginal profit group index under disruption (MPGID), we characterize the worst-case total profit as a function of both the flexibility design and demand uncertainty, modeled through a partwise independently symmetric perturbation set. This representation enables direct performance comparisons across different flexibility configurations under disruption risk. For the case involving two products with distinct profit margins under supply disruption risk, we demonstrate that the alternate long-chain design outperforms all other long-chain configurations in terms of worst-case profitability. In addition, in certain cases, arc-based disruptions can be just as devastating as plant-node disruptions, particularly when they lead to the loss of high-margin demand. However, our fragility analysis reveals that this design becomes increasingly vulnerable as disruption risks intensify. To address this issue, we propose an MPGID-based heuristic that systematically generates flexible designs to mitigate both supply and demand uncertainties.

Considering human motivation, is it more effective to assign work tasks to shop floor workers or let them choose among a set of open tasks? This paper reports the results of a field experiment conducted with a manufacturer that uses wearable devices to distribute shop floor tasks. Drawing on theory on human motivation, we analyze 66,233 machine status reports and 31,429 work tasks completed in two manufacturing plants in Germany and Italy. We rely on a Difference-in-Differences approach in a 13-week-long field experiment. The results show that allowing workers to choose their next task shows no aggregate productivity difference but reveals heterogeneous behavioral effects. We find that workers respond more slowly to tasks (response time), but once accepted, tasks are completed faster (completion time). We discuss the potentially canceling effects and uncover evidence for behavioral mechanisms such as social loafing. This study has significant consequences for production managers in charge of implementing choice-based digital task assignment systems.

Empirical evidence indicates that consumers experience a relative loss of excitement over time and miss their habits when refraining from consumption under repeated purchases. Despite the practices of first-time trials and welcome-back offers, how to exploit this habituation behavior through customized dynamic pricing is overlooked in the existing literature. In this article, we analyze the long-run optimal customized dynamic pricing strategy in the presence of consumer habituation for time/frequency-based consumption. The consumer’s willingness to pay is determined by past consumption via the habitual level of consumption. We formulate the seller’s continuous-time profit maximization problem over an infinite horizon as an optimal control problem and derive the closed-form value function and optimal pricing strategy. We identify a key condition on the willingness-to-pay function that captures both the magnitude of the habituation effect and the saturated willingness to pay, which plays a central role in determining the seller’s optimal pricing policy. When consumers retain interest even at high levels of habitual consumption, always selling by pricing at their current willingness to pay is optimal. In contrast, if consumers lose interest when saturated, the optimal policy becomes a cutoff strategy, where the seller offers the product only when consumption remains below a critical threshold and prices out consumers beyond it. We further examine the resulting pricing and consumption dynamics and characterize an equilibrium habitual level of consumption at which randomized promotions can be optimal. Furthermore, our key findings qualitatively hold under random consumer valuation, but with richer patterns. Our work provides a tractable, closed-form characterization of the optimal dynamic pricing policy under consumer habituation, offering a more nuanced understanding of the strategic implications of an inverted U-shaped willingness-to-pay.

To increase their competitiveness, firms continue to upgrade their products and to buy technology upgrade components. While firms generally know the value of technology upgrade products (components), they do not know when the upgrade solution can be produced on a massive scale, who can produce it, and what the unit production cost is when they make their purchase decisions. To obtain the upgrade solution information and to reduce their purchase costs, some firms adopt the strategic supplier with bargaining mechanism, while other firms adopt the reverse auction with incentive mechanism. Recently, some firms (e.g., China National Petroleum Corporation) have adopted an innovative mechanism: they establish strategic partnerships with advanced suppliers and promise a proportion of the order to incentivize them to reveal the information on time. Meanwhile, they also adopt the reverse auction to determine the winning supplier and the purchase price for the remaining order to reduce purchase costs. Motivated by these firms’ practices, we propose a new procurement mechanism for technology upgrade products: the strategic auction mechanism . We analyze the optimal promised proportion of the order, the suppliers’ equilibrium policy, and the buyer’s equilibrium profit under this mechanism. In addition, we compare the performance of these three mechanisms and identify their application conditions.

Capacity configuration and allocation problems in production networks under demand uncertainty have been extensively studied in the literature. In addition to demand uncertainty, capacity risk is also prevalent in practice. This article considers a bipartite production system consisting of multiple plants and a set of customers located in different areas. In particular, these plants will use their configured capacities, facing capacity risk, to serve those customers who have differentiated fill-rate requirements with uncertain demands. Our objective is to optimize capacity configuration and allocation decisions by formulating the problem as a two-stage stochastic program that explicitly incorporates capacity risks and budget constraints. First, we develop a novel “randomized edge-weighted max-flow” priority rule that integrates capacity risk, penalty costs of unmet demands, and allocation costs to allocate the realized capacities to satisfy fill-rate targets, with a tunable parameter introduced to balance service levels and costs. In particular, we derive necessary and sufficient conditions for identifying plants with zero capacities and obtain some practical insights, which can help reduce the potential redundancy of the network system. When partial information and correlation of distributions are known, we implement robust optimization techniques to obtain the ranges of robust capacities under various scenarios. Some practical managerial implications are derived by theoretically discussing the impacts of capacity risk and demand correlations on these ranges. From numerical experiments, we found that the long-chain configuration no longer has the “almost-as-good-as-the-fully-flexible-network” advantage when the capacities are random. In addition, we also investigate the impacts of correlations among realized capacities or demands on the associated flexibility. Finally, some interesting and useful insights for practitioners can be obtained by analyzing the impacts of costs, fill rates, and uncertainty on the total optimal capacity and total configuration cost.

The rise of generative AI (GenAI) technology is revolutionizing firm operations and supply chain management, profoundly influencing firms within the GenAI supply chain network. While a few emerging studies have explored the impact of GenAI initiatives on firms’ stock performance, it remains unclear how such disruptive technology initiatives affect the stock price movements of their supply chain partners. This paper empirically investigates the impact of firms’ GenAI initiatives on the abnormal stock returns of their suppliers by analyzing data on news, supply chain relationships, and daily stock prices. The analysis indicates that, on average, a firm's GenAI announcement leads to a positive abnormal return of 0.27% for its suppliers on the announcement day, with additional evidence of sustained long-term value creation. Heterogeneity analyses reveal that this spillover effect is more pronounced among suppliers with higher R&D intensity, stronger sales growth, closer geographic proximity to the focal firm, and operations in less competitive industries. Furthermore, the positive abnormal returns are significantly greater when the GenAI initiative pertains to product innovation rather than process improvement. Our findings have important implications for supply chain stakeholders navigating the adoption of GenAI technologies.

Online asset-selling businesses, such as used cars and real estate platforms, have experienced remarkable growth in recent years. Unlike general retail operations, which make decisions at the stock-keeping unit level, asset selling operates at the individual unit asset level. Practical operational constraints (e.g., infrequent price adjustments within a limited timeframe) set asset-selling platforms apart from general retail. Further complicating decision-making are real-world uncertainties, such as volatile demand and unknown latent value of the asset. We present a dynamic pricing framework that captures the salient characteristics of the asset-selling business while leveraging consumer online behavioral data to maximize the payoff of individual assets. We develop practical algorithms for solving the dynamic pricing problem, including a mean approximation (MA) algorithm that uses forecast mean values as proxies for future customer arrival rates and online learning algorithms that integrate learning of the latent value of an asset with dynamic pricing decisions. To evaluate these algorithms, we propose an asymptotic regime suitable for the online asset-selling business context, one that scales up customer demand arrival rate within a finite time horizon. The key findings are that, under mild conditions, the expected value of selling an asset is concave and increasing at a logarithmic rate with respect to demand rates and increasing no faster than a linear speed in the asset’s latent value. These properties allow us to derive the performance bounds of our policies. An extensive numerical study and a real-data calibrated case study demonstrate the practical value of our proposed algorithms, suggesting that those simple heuristics can achieve strong performance in the asset-selling environment. Moreover, our integration of the learning of an asset’s latent value with dynamic pricing decisions, alongside asymptotic analysis, provides a robust framework for data-driven decision-making and demonstrates the potential of consumer behavior data as a strategic asset for online asset-selling platforms.

With the rapid growth of omnichannel retailing and the takeaway delivery economy, the classic point-to-point mode for on-demand delivery is deficient in delivery capacity, coverage area, dispatching efficiency, and courier safety assurance. Inspired by the success of Dabbawala, a historical Indian company for lunch delivery, we propose a novel public on-demand delivery service system that uses the public transit network to satisfy stochastic delivery demands. In particular, the proposed system includes a radial public transit network for intermediate transshipment, as well as couriers with e-bikes for terminal pick-up and drop-off. Our research aims to generate system design that minimizes the sum of penalty costs from lost sales and operational costs associated with courier terminal delivery distance. Solving the integrated system optimization problem relies on incorporating operational details, especially the allocation strategies of the lines’ capacity and the couriers’ terminal traveling modes. For the former, we propose a novel flexible design, called dual long-chain design, to improve flexibility. For the latter, we propose an elegant approximation of optimal service region partitioning that minimizes the expected terminal delivery distance and the resulting costs, without compromising delivery timeliness. Leveraging the theoretical results of the operational strategies, we simplify the integrated optimization problem and propose an efficient approximation algorithm. Finally, we validate the advantage of the proposed system over classic point-to-point delivery in satisfying demands and reducing costs through extensive numerical experiments, providing managerial insights in handling massive on-demand delivery demands and utilizing the idle capacity of the public transit system.

We study a distributionally robust pricing model aimed at maximizing the worst-case profit under limited knowledge of consumers’ valuation distribution. In addition to mean and variance, we incorporate asymmetric distributional information through semivariance, and then explore the robust optimal posted and randomized pricing strategy. We obtain the robust posted price and show that several series of asymptotic three-point distributions can approach the worst case gradually. The inclusion of asymmetry enables sellers to design three distinct pricing strategies, each corresponding to a different level of asymmetry with explicitly defined thresholds. For the randomized pricing problem, we derive a near-optimal strategy by identifying three key pricing candidates. To avoid the complexity of implementing the optimal randomized price, we also explore a compromise strategy that employs a finite set of pricing candidates, which we formulate as a second-order cone program. Furthermore, we provide numerical evidence highlighting the advantages of incorporating asymmetric distributional information over the mean–variance approach, from both the worst-case profit and actual profit perspectives. Our numerical experiments also reveal that posted pricing is more effective for low-cost products, while randomized pricing is preferable for higher-cost items.

We consider a dynamic resource provisioning problem for a supplier in which the availability of the provisioned resource is subject to random disruptions whose distribution is only indirectly observable through samples. To signal its commitment to service quality, the supplier adopts a service level agreement contract that specifies both the target service level and the associated penalty for violation over a finite contract period. The supplier needs to dynamically determine resource provisioning decisions with the objective of minimizing operational costs and penalties incurred due to service-level agreement violations. We construct a Wasserstein-based distributionally robust dynamic programming framework to model and solve the dynamic resource provisioning problem under a service-level agreement. In particular, we provide a convexification algorithm that enables us to solve the nonconvex robust dynamic programming problem in a backward manner. We further examine a special case where service shortages depend linearly on the provisioned resources, enabling the problem to be reformulated into a sequence of linear programs. This linear shortage model naturally connects to residual-based robust formulations, which facilitate us to accommodate nonlinear relationships between resource provisioning and service shortages. We propose several approximation algorithms to improve computational efficiency. To mitigate the possibly over-conservativeness, we explore radius adjustment strategies based on sample size, state, stage, and cumulative cost information, which yield consistent out-of-sample performance. We perform a case study of a cloud computing example to demonstrate the effectiveness of the proposed solution approach and elicit managerial insights. The results suggest that suppliers should provide fewer backup servers when cumulative downtime is low or when approaching the end of the planning horizon. The dynamic resource provisioning policy significantly reduces the total cost compared to the best static policy. Furthermore, applying appropriate radius adjustments can further enhance the out-of-sample performance.