Knowledge that Transforms

We investigate the impact of skewed demand on robust inventory management. Including skewness in calculations leads to cubic constraints that prevent a standard two-stage method from deriving an explicit and tractable objective function. To overcome this roadblock, we propose a joint optimization method to directly derive the robust solution in a closed form. The joint optimization method is widely applicable to various robust inventory models with different ambiguity sets. The notable advantage is that we can obtain the final solution without deriving the objective function, so many tedious intermediate steps are circumvented. We conduct numerical experiments on industry data to demonstrate that our moment-based policies deliver more consistent performance than divergence-based policies. After obtaining various closed-form solutions, we demonstrate that including skewness in the model improves the profit generated by the robust optimal order quantity if either demand is bounded or the cost-to-price ratio is low, even though the sample moments used may not equal the population moments. Furthermore, under these conditions in which skewness should be included in the calculations, the firm’s expected profit, under the most unfavorable distribution, increases with variance but decreases with skewness. This result stands in contrast to findings in the economics and finance literature, where distributional ambiguity is not considered.

Dynamic pricing is often complicated by strategic customer behavior. One tactic utilized by retailers to manage strategic customer behavior, known as Markovian pricing, is to offer price discounts at random intervals to prevent customers from predicting when the next discount will occur, thereby simplifying their strategic waiting behavior. In this article, we study Markovian pricing in competitive settings. We show that retailers can effectively adopt Markovian pricing in competitive environments, establish the optimality of flash discounts under competitive Markovian pricing, and find surprisingly that increased levels of competition may benefit both retailers. We confirm the robustness of these insights and also establish their limits of applicability in two model extensions. Our findings suggest that retailers engaging in competitive Markovian pricing should refrain from naïvely applying common wisdom toward third-party price-monitoring and comparison services and reconsider the efforts in growing their loyal customer base, and more broadly highlight the unique properties of competitive Markovian pricing.

We consider an assortment optimization problem for a class of online video games where the in-game virtual store has a unique structure with two sections: Featured and Just For You (JFY). All customers (players) are offered the same Featured section assortment, whereas the JFY section is used for personalized recommendations. We model customer choice under a constrained mixture-of-nested-logit model and propose different solution methods for the resulting assortment optimization problems. First, we introduce a novel mixed-integer nonlinear programming (MINLP) formulation. Numerical experiments show that the MINLP formulation generally obtains optimal solutions efficiently, using a variety of instances derived from conversations with our industry partner to mimic the environment found in their video game stores. In addition, we propose three approximate solution methods with theoretical performance guarantees: a fully polynomial time approximation scheme, a mixed-integer linear programming formulation, and a heuristic algorithm. To understand the impact of a shared Featured section, we analyze the distribution of display capacity between the Featured and JFY sections. Our numerical experiments highlight that the Featured section plays a critical role in balancing revenue and customer utility. To validate our use of a mixture-of-nested-logit model, we further conduct a simulation study based on ground-truth instances that are independent of the underlying structure of the consumer choice models we consider. The results indicate that our nested structure yields superior performance in terms of both capturing customer behavior and simulation revenue, compared with the mixture-of-multinomial logit model and the current practice of our industry partner. Overall, our paper is the first to study assortment optimization for the gaming industry under discrete choice models; it is also the first to devise both exact and approximate solution approaches for the constrained mixture-of-nested-logit model. Our results provide guidance for effective management of assortments in online video game stores and offer an “assortment” of solution approaches, allowing practitioners to choose one that best suits their environment.

We study a production planning problem with linear, nonlinear, deterministic, and/or stochastic production relationships between the production plans and actual production quantities. We start by introducing a stochastic dynamic programing formulation of the problem with a Markovian assumption on the production relationships. Under specific conditions, we establish the convexity of the optimal cost-to-go function and closed forms of optimal policies. To solve the original problem in the general case, we propose a solution framework based on sequential policy optimization and deep reinforcement learning. We discuss the theoretical properties of the framework and evaluate its numerical performance with linear and nonlinear production relationships. In the linear case, our framework performs in line with the state-of-the-art optimization-based methods with improved computational efficiency. In the nonlinear case, our framework achieves an optimality gap near 10%–20%. We also illustrate that the proposed methodology can also be extended to the problem of joint production planning and scheduling.

In this paper, we explore the question of whether an apparel manufacturer should incorporate a renting channel into its existing business model, which currently only includes a retailing channel. We also examine how such a change would affect product quality and sustainability. We develop several game-theoretical models where an apparel manufacturer, currently selling apparel products directly to consumers, may choose to rent them out either directly or indirectly through a third-party online platform. When both retailing and renting channels are available, the manufacturer will only sell the product if rental utility or brand quality is low. Otherwise, it will sell and rent out the product simultaneously. Our findings suggest that adding a renting option to a retailing-only business model could increase the manufacturer’s profit and product quality, despite potential demand cannibalization. However, the environmental impact of rental versus pure retail depends on the level of rental utility. The increase in profit and quality from incorporating renting is more significant when brand quality is higher. Additionally, this change could enhance consumer surplus and social welfare. We also analyze the effects of three common public policies on sustainability when a renting channel is added, finding that two of the policies can promote product quality. We extend our model by considering consumer heterogeneity in preferences for green consumption or discounted utility from renting, and find that our results remain robust. When examining demand cannibalization over time, we observe that product quality decisions may be influenced by the extent of cannibalization over time, while the environmental impact could worsen.

Real-time feedback applications are reshaping employee performance feedback in operations management. Their design and implementation significantly influence employee engagement, which is a key factor in the success of technology-driven business innovations. This study investigates an implementation strategy to optimize feedback app use by employing digital nudges to encourage regular engagement. Drawing on social norms theory and cognitive load theory, we examine how message framing and users’ cognitive states affect the quantity and content features of feedback, including ratings, review length, surprise, and recognition. We conducted two randomized experiments to evaluate the effectiveness of digital nudges. Experiment 1, a field study with over 250 users of the DevelapMe app in a financial firm, tested how message framing and timing influenced cognitive load. Experiment 2, an online randomized controlled trial, explored underlying mechanisms and validated cognitive load measures. The results show that while social norms-based nudges increase feedback volume, they are associated with shorter reviews and reduced recognition and surprise. Employees experiencing lower cognitive load provided more feedback but tended to give lower ratings. Importantly, the interaction between social norms and low cognitive load resulted in higher ratings and more detailed reviews. This suggests that reducing cognitive load can enhance the positive effects of social norms on feedback quality. This study underscores the moderating role of cognitive load in the effectiveness of digital nudges and offers insights into feedback design that promotes both participation and quality. Theoretically, it contributes to research on digital feedback systems by integrating social norms and cognitive load theories to explain employee feedback behavior. Practically, it provides guidance for managers in designing real-time feedback tools that strategically use digital nudges while minimizing cognitive load, fostering a culture of continuous feedback, strengthening engagement, and improving performance management systems.

The COVID-19 pandemic accelerated telemedicine adoption, offering a convenient alternative to in-person care. However, televisits may not fully address health concerns and sometimes require supplementary in-person visits, consuming resources that could have been saved if the initial visit had been in-person. As the pandemic subsides, in-person visits are regaining popularity, prompting providers to reorient resources toward in-person care. Transportation support (or subsidies) for patients, funded by providers or the government, plays a critical role in facilitating in-person visits. In this evolving landscape of telemedicine, we study how an outpatient care provider can optimally balance virtual and in-person services and whether, and how, to engage with transportation subsidies. We connect these two questions by examining how transportation subsidies reshape the provider’s optimal capacity allocation across service channels and, in turn, affect overall patient access. We develop a stylized queueing-game model to represent the operations of a revenue-maximizing provider serving patients who strategically choose between service channels. We find that provider size, measured by total capacity relative to demand, is key. Small and large providers perform best by focusing on one channel without offering subsidies, whereas medium-sized providers benefit from carefully balancing both channels alongside subsidies. Paradoxically, transportation subsidies, which make in-person care more accessible, may reduce overall patient access to care, even when fully funded by the government. This occurs because providers may shift capacity toward a higher-reimbursement channel, ultimately serving fewer patients. Differentiating payment rates between in-person and virtual visits can potentially prevent such reductions. Our study highlights the importance of capacity coordination between channels for providers and cautions policymakers that transportation support may unintentionally harm patient access. Properly designed financial incentives can help prevent such negative outcomes.

Wildfires—also known as wildland fires, bushfires, and forest fires—are large-scale, uncontrolled fires that occur in vegetated areas, triggered by natural or human causes. They have significant economic, environmental, and social impacts. As with most disasters, the location, timing, and magnitude of wildfires are unpredictable. Wildfire management (WFM) has a long history in forestry research; however, it remains relatively underexplored by Operations Management (OM) researchers, partly due to the specialized knowledge required. We propose a macro-level framework to review the existing WFM literature and identify four key areas for future research: Risk Assessment, Fuel Treatment, Surveillance and Detection, and Initial Attack and Suppression. We also examine techniques for managing these functions and present cross-tabulations to map research coverage. This work enables OM researchers to identify research opportunities that align with their interests in WFM functions and their expertise in specific methodologies. We highlight research gaps and outline opportunities for empirical and analytical contributions, including predictive and prescriptive analytics. Practitioners will also benefit from discussing WFM functions to address real-world challenges.

Despite significant advancements in predictive artificial intelligence (AI), organizations continue to grapple with determining the most effective integration of AI into their operations, particularly in combining AI with human capabilities in task execution. This paper introduces a diagnostic system aimed at minimizing costs while maintaining patient safety in hospitals by efficiently allocating mammography interpretation tasks between AI algorithms and radiologists. The optimal diagnostic system employs algorithm-generated risk scores for mammograms to determine if additional assessment by a radiologist is necessary. It evaluates the costs of two approaches—automation, where AI completely replaces human interpretation, and delegation, where AI and radiologists share interpretation tasks—compared to the current expert-alone strategy. When AI performance does not exceed that of radiologists, the optimal design is a simple two-threshold policy: AI recommends no follow-up for low-risk cases, recommends follow-up for high-risk cases, and delegates ambiguous cases in between to radiologists; the thresholds are analytically derived and state-independent. This two-threshold policy is state-independent and that optimal thresholds are not contingent on the diagnostic system’s prior history. We demonstrate our system’s performance by back-testing against real-life scenarios, utilizing data from a mammography AI contest and real-world cost and performance metrics. Backtesting demonstrates potential cost savings of up to 20.9% compared to the expert-alone approach. Beyond radiology imaging, our work holds significant implications for the design of workflows in the AI era and human–machine collaboration contexts.

In service systems, customers tend to overutilize resources, resulting in high congestion. We study a non-coercive way to alleviate congestion based on customers’ awareness of the negative waiting externalities they impose. To this end, we revisit Naor’s rational-queue model by adding a term that captures the cost a customer associates with the waiting time generated for other customers encountered during their sojourn. First, we show that this modification preserves the threshold structure of customers’ joining strategy. Using lattice-path counting techniques, we derive a closed-form expression for the expected generated waiting time, conditional on the number of customers observed upon arrival. The generated wait is increasing and concave in the queue length at arrival. Viewing expected waiting time as allocated across arrivals, the queue length associated with an arbitrary unit of expected generated waiting time is first-order stochastically lower than the queue length associated with a unit of expected personal waiting time. We prove that the maximum generated wait is increasing and convex in the joining threshold and provide tight bounds with exact light/heavy traffic asymptotics. Internalizing social awareness (weight on generated wait) lowers the joining threshold and reduces waiting; in high-demand regimes, it can also increase social welfare. When the operator can also influence the weight on personal wait, that lever reduces congestion more strongly than waiting externality awareness. Finally, we quantify the awareness level needed to reach Naor’s social threshold and show that it is lowest at a non-extreme demand level.