Wiley-Online-Library: Naval Research Logistics (NRL): Table of Contents

Learning Optimal Reprocessing Policies for Degradable Inventory Systems: A Multi‐Armed Bandit Approach

Haiqi Shi — Fri, 12 Jun 2026 04:21:46 -0700

ABSTRACT

In this paper, we study the problem of online learning for a threshold-based inventory control policy in degradable inventory systems with reprocessing capabilities, where the demand is unknown but independent and identically distributed (i.i.d.) across periods. A key feature of our model is that degraded items can be held as inventory rather than being immediately salvaged or reprocessed, providing additional flexibility in operational decisions. Unlike traditional models that assume prior knowledge of the demand distribution, we develop an adaptive learning algorithm to determine the optimal reprocessing and production decisions without such information. Our analysis is motivated by medical supply chains where products like oxygen cylinders degrade over time but can be reprocessed through sterilization and refilling. The model incorporates key operational features including: (1) discrete quality degradation of inventory over periods, (2) the option to hold degraded items in inventory before reprocessing, and (3) the trade-off between production, reprocessing, and inventory holding costs. Through a novel notion of generalized multi-modularity tailored to our state-action structure, we establish the optimality of a state-dependent threshold policy with state-independent threshold parameters, governing both reprocessing and production decisions. When demand is unknown a priori, we propose an online learning algorithm and prove that the algorithm achieves a cumulative regret of O(logT·T)$$ O\left(\log T\cdotp \sqrt{T}\right) $$. This work contributes to both production-inventory coordination and online learning literature by providing: (1) structural analysis using a generalized multi-modularity framework to characterize the optimal policy in the setting with known demand, (2) the first learning-theoretic framework for reprocessable and degradable inventory systems where degraded items can be strategically held before reprocessing, and (3) theoretical performance guarantees through regret analysis. The methodology applies to various industrial settings where products degrade discretely over time and can be held in degraded states before being reprocessed or salvaged.

Financing the Supply Chain Sustainability: The Role of the Bank's Responsibility Level

Jintao Zeng, Huiling Zhong, Guitian Liang — Mon, 08 Jun 2026 03:30:45 -0700

ABSTRACT

Financial instruments have become an emerging solution to establish sustainable supply chains recently. This paper investigates the effectiveness of a new tool called the sustainability-linked loan (SLL), which contains a sustainability target, a regular interest rate as well as a discounted rate. If borrowers meet the target at the maturity of the loan, they can enjoy an interest discount, otherwise the regular rate is charged. We consider a retailer that procures a single product from a supplier, under which the supplier is capital-constrained and can exert costly effort to raise the product sustainability performance. We study two SLL models: the first one is called the supplier profit-responsibility (SPR) model, where a bank offers the SLL to the supplier, and the bank's objective is to maximize its own profit plus the supplier's effort. Compared with a general loan that contains only an interest rate, the SPR model always induces a higher effort level while weakening the supplier's profit. The second model is called supplier total responsibility (STR) model, under which the bank is totally responsible only to maximize the supplier's effort. We find that the bank's responsibility level (i.e., the weight of effort in the bank's objective function under the SPR model) significantly affects the comparison between the two SLL models. In particular, with a higher responsibility level, while the SPR model induces greater supplier effort, which benefits the retailer, the supplier and the bank obtain lower profits compared to those in the STR model. Furthermore, if the bank's responsibility level is intermediate, the SPR model produces a multi-lose situation: the sustainability level, the profit of the bank and the supply chain are all worse off compared to the STR model. Our results show that when providing the SLL, the bank's preference regarding supply chain sustainability plays a vital role.

Integrated Production and Delivery Scheduling With Multi‐Machine Open Shop Processing and Capacitated Vehicle Batch Deliveries

Jianming Dong, Weitian Tong, Yao Xu — Mon, 08 Jun 2026 01:51:18 -0700

ABSTRACT

We consider an integrated production and delivery scheduling (IPDS) problem where each job must be processed on multiple machines in an open shop environment and then delivered in batches to a single customer using a capacitated vehicle. The objective is to minimize the makespan, which captures both production and delivery stages. This model arises in several practical settings, including additive manufacturing and diagnostic laboratories, where tight coordination between production and outbound logistics is critical. Although IPDS has been extensively studied under single-machine, parallel-machine, and flow shop models, the open shop variant has received limited theoretical attention. This paper presents the first constant-factor approximation algorithms with provable worst-case guarantees for the multi-machine open shop IPDS problem. We first introduce a simple two-phase algorithm that achieves a 52+ϵ$$ \frac{5}{2}+\epsilon $$ approximation ratio for any fixed number of machines. We then develop a refined 32+ϵ$$ \frac{3}{2}+\epsilon $$ approximation algorithm based on a novel framework that combines instance scaling, schedule discretization, configuration enumeration, and LP-based assignment. In the special case of two open shop machines, our result significantly improves upon the best known 2-approximation by tightening the bound to 32+ϵ$$ \frac{3}{2}+\epsilon $$. Importantly, we observe that no algorithm can achieve an approximation ratio below 32$$ \frac{3}{2} $$ unless P=NP$$ \mathrm{P}=\mathrm{NP} $$, establishing that our algorithm is nearly optimal.

Dynamic Lot Sizing With Multiple Age‐Differentiated Demands of a Perishable Product

Jie Fan, Jinwen Ou, Vernon N. Hsu — Mon, 08 Jun 2026 01:47:23 -0700

ABSTRACT

In this paper, we study a dynamic lot sizing problem in which a manufacturer produces a perishable product and supplies it to several customers, each having a different requirement on the product freshness and only accepting inventory stock whose age is not older than a customer-dependent threshold. Both production and inventory holding cost functions are assumed to be concave. The manufacturer makes joint production and inventory allocation decisions to minimize the total cost. We explore structural properties and develop an optimal algorithm for solving the problem in polynomial time. More efficient optimal algorithms are also developed for two important special cases. We conduct computational experiments to demonstrate that the proposed algorithm significantly outperforms the commercial solver CPLEX. Through computational experiments, we also obtain managerial insights into the bottleneck customer, who has the highest freshness requirement.

Information Design for Early‐Stage Dose‐Finding Trials

Amin Khademi, Ningyuan Chen — Mon, 01 Jun 2026 22:18:35 -0700

ABSTRACT

To enhance enrollment rates in early-stage dose-finding clinical trials, we propose an information design approach, where the clinical investigator (CI) commits to an information releasing mechanism (IRM) based on the treatment's uncertain efficacy and toxicity to encourage patients to participate in the trial. The optimal IRM has a threshold structure, involving completely revealing/pooling the state (utility) when the realization is in the respective region. We analyze the comparative statics under the optimal IRM. Because general IRMs may be difficult to implement in practice, motivated by response-adaptive clinical trials where information from past patients will help drive decisions for future patients, we consider a practical IRM, where the CI decides on the number of patients to recruit, whose efficacy/toxicity outcomes will be used as information for future patients. For example, the proposed IRM may be the number of patients in the first batch in dose-escalation methods in Phase I, the expansion cohort at the end of Phase I, or the number of patients assigned to the maximum tolerated dose in Phase II. We show that this practical IRM can achieve at most 50% of the value of the optimal abstract IRM. Since patients' risk attitude toward toxicity may be private, we study the impact of belief about it on the optimal IRM and show that the structure of the optimal IRM can be in sharp contrast with that in the public information setting, depending on the distribution of risk attitude toward toxicity. To better understand the impact of the association between efficacy and toxicity on the optimal IRM, we study a bivariate Bernoulli model and show that the optimal IRM has a threshold structure, and the region in which it is a randomized recommendation, when the treatment is ineffective and toxic, shrinks when the association becomes larger.

A Heuristic for Bundle Pricing With Interrelated Products

Mon, 11 May 2026 23:41:57 -0700

ABSTRACT

Bundle pricing is a widely used strategy in marketing analytics and revenue management, yet most existing heuristics rely on restrictive assumptions about product valuations, such as independence or additivity, and about costs, such as being zero or additivite. We address the mixed bundling problem by proposing a general heuristic algorithm, the pricing-assignment heuristic (PAH), and its extension, E-PAH, which are designed to accommodate heterogeneous customer types and general forms of product interrelatedness, including complementarity and substitutability. The proposed approach alternates between pricing and assignment steps to iteratively improve the seller's profit, and can be initialized from any feasible solution or used as a post-processing tool to enhance existing heuristics. Through extensive numerical experiments, we show that our method consistently outperforms standard benchmarks such as component pricing, pure bundling, and bundle size pricing, achieving on average profit improvements of up to 43% for independent products and up to 53% when products exhibit interdependencies. The algorithm remains effective in the presence of general cost structures, highlighting its flexibility and practical relevance. Overall, the proposed heuristic provides a simple, scalable, and broadly applicable framework for mixed bundling in complex and realistic market environments.

Improving Affordability and Accessibility for Socially‐Beneficial Services via Government Incentives

Xiaoyan Zhao, Venus Lo, Stephen Shum — Sun, 10 May 2026 00:00:00 -0700

ABSTRACT

Many socially-beneficial services, like dental care, suffer from a dual challenge: Low affordability for citizens and long waiting times due to insufficient provider capacity. We model a government's problem of designing subsidy policies to address this issue. We analyze three financial interventions: Consumer vouchers, which are on the citizen side; and two provider-side subsidies aimed at increasing capacity: Fee-for-service subsidies and downtime rebates. Using a continuous-time Markov chain model, we capture the dynamic interactions between citizens and the service provider in response to these policies. Our findings yield several structural insights into subsidy design. We prove that subsidizing idle service capacity (a form of risk mitigation) always outperforms fee-for-service subsidies (a form of reward enhancement) in terms of cost-effectiveness. However, the choice between citizen-side and provider-side policies depends critically on the primary system bottleneck. To further improve the government's cost efficiency, we propose a mixed-subsidy policy. Although optimizing this policy is intractable, we develop an algorithm to find near-optimal solutions. Numerical experiments demonstrate that a mixed policy combining consumer vouchers with idle-time rebates can offer substantial cost savings compared to the best single-subsidy approach, highlighting the potential efficiency gains of a more integrated subsidy structure.

A Joint Production and Delivery Problem in Multi‐Factory Multi‐Distribution‐Center Multiproduct Systems With Limited Capacities

Shenghui Zhai, Fang Liu, Zhuan Zuo — Thu, 07 May 2026 22:43:23 -0700

ABSTRACT

We consider a joint production and delivery problem in multi-factory multi-DC (distribution center) multiproduct systems with limited production and delivery capacities over a finite horizon. The objective is to minimize the system's expected total cost. Since the structure of the optimal policy is hard to find, we propose a Lagrangian relaxation heuristic to solve the problem. The proposed heuristic is based on solving a Lagrangian relaxation of the original problem. Although the Lagrangian relaxation problem remains challenging due to the joint production and delivery decisions, we identify a zero-inventory policy that enables further decomposition into independent single-product, single-DC subproblems, each of which can be solved independently. We evaluate the heuristic's performance by deriving a theoretical upper bound on its expected loss. In numerical experiments, we compare the Lagrangian relaxation heuristic with a benchmark myopic heuristic. The results consistently show that the Lagrangian relaxation heuristic achieves a significantly smaller expected relative loss and exhibits greater stability than the myopic heuristic.

Implementing Target‐Based Deferred Payments in Crowdfunding: A Signaling Perspective

Jiancheng Lyu, Yang Xu, Mabel C. Chou — Wed, 29 Apr 2026 23:59:13 -0700

ABSTRACT

In reward-based crowdfunding, asymmetric information about product quality can hinder both fundraising success and post-campaign sales. To mitigate this problem, Belavina (2020) introduces a target-based deferred payment mechanism, which withholds any funds raised above a prespecified funding target until the creator delivers the promised product to backers. This paper develops a game-theoretic signaling model to analyze how this mechanism influences strategic behavior in crowdfunding campaigns. The analysis reveals that target-based deferred payments not only lower the signaling costs for high-quality creators but also reshape their optimal signaling strategies. Specifically, when the fixed setup cost of producing a high-quality product is relatively low, high-quality creators optimally signal their type by offering low reward prices. Conversely, when the fixed setup cost is high, setting a high funding target becomes the dominant separating strategy. The paper further examines alternative formulations of the deferred payment mechanism and offers practical guidance for creators on designing campaigns that credibly signal product quality.

Online Monitoring of Irregularly Spaced Serially Correlated Univariate Processes

Xiulin Xie, Jiwoo Ha — Tue, 21 Apr 2026 22:06:13 -0700

ABSTRACT

Statistical process control (SPC) methods are commonly employed in various fields to detect distributional changes in sequential processes. Traditional SPC charts are typically developed under the assumption that in-control (IC) process observations are independent and normally distributed with identical parameters. However, when these assumptions are violated, recent research has shown that conventional control charts may become unreliable. To address these limitations, various alternative and flexible control charts have been developed to accommodate autocorrelated observations and nonparametric process distributions. Although existing methods can be reliable and effective when their assumptions hold, they still have some limitations. For instance, methods handling autocorrelated data often rely on parametric time series models or assume equally spaced observations, whereas nonparametric control charts that rely on data ranking or categorization typically suffer from information loss. Furthermore, the optimal performance of many control charts in detecting specific shifts often relies on the accurate specification of their parameters in advance. In this paper, we introduce a novel framework for Phase II online monitoring of univariate processes with irregularly spaced observation times and serial correlation, and the IC distribution cannot be adequately modeled by a parametric form. The method first estimates the IC covariance function for irregularly spaced time series using a local linear kernel smoothing procedure, then sequentially decorrelates the process observations. Next, the decorrelated observations are transformed based on their estimated IC distribution such that the transformed data are approximately standard normal. Finally, an adaptive CUSUM chart is employed to monitor the transformed data. Simulation results indicate that the proposed approach is effective across a variety of scenarios.

Partially‐Observable Sequential Change‐Point Detection for Autocorrelated Data via Adaptive Upper Confidence Region

Haijie Xu, Xiaochen Xian, Bo Zhang, Chen Zhang, Kaibo Liu — Tue, 14 Apr 2026 23:43:34 -0700

ABSTRACT

Sequential change-point detection for multivariate autocorrelated data is a widely encountered challenge in real-world applications. When sensing resources are limited, only a subset of variables from the multivariate system can be observed at each time point, giving rise to the problem of partially observable multi-sensor sequential change-point detection. To address this, we propose a novel detection framework called Adaptive Upper Confidence Region with State Space Model (AUCRSS). This approach models multivariate autocorrelated data using a state space model (SSM) and incorporates an adaptive sampling policy to enable efficient change-point detection and localization. A partially observable Kalman filter is developed for online inference of the system state, and based on this, a change-point detection procedure is constructed using a generalized likelihood ratio test. We analyze the relationship between detection power and the adaptive sampling strategy. Furthermore, by interpreting detection power as a reward signal, we establish a connection with the online combinatorial multi-armed bandit (CMAB) problem and introduce an adaptive upper confidence region algorithm to guide the sampling policy design. We provide a theoretical analysis of the asymptotic detection power, and we demonstrate that our proposed method significantly outperforms the baseline algorithms through extensive numerical experiments on both synthetic and real-world datasets.

Tactical and Strategic Risks From Supply Disruptions in Competing Supply Chains

Akhil Singla, Wallace J. Hopp, Seyed M. R. Iravani, Zigeng Liu — Sun, 12 Apr 2026 00:00:00 -0700

ABSTRACT

Supply chain disruptions can lead to both tactical (i.e., loss of short-term sales during a disruption) and strategic (i.e., loss of long-term market share) consequences. We model the impact of a supply disruption on competing supply chains in which two firms compete for a limited backup supply. We describe strategies for both firms in a two-stage game comprising (i) Preparation, which involves investment prior to the disruption to secure backup supply, and (ii) Response, which involves post-disruption purchasing from the secured backup supply for a component whose availability has been compromised. Firms maximize their long-run profit while simultaneously deciding their preparation and response strategies. We find the equilibrium strategy for firms in the two stages of the game. We describe the conditions under which a firm can use its preparation investment to not only minimize its disruption risks but also capture more market share. We also introduce a Leader-Follower-based game-theoretic model that helps measure each firm's risk exposure by estimating the benefit of preparation. We identify the primary factors that influence the firm's preparation investment and affect customer satisfaction, and show that these depend on the size of the firm and the length of the disruption. This enables us to characterize the appropriate balance between protecting market share and exploiting a disruption to gain market share.

Designing Optimal Incentives for Target‐Driven Projects

Xiaohan Zhu, Xu Sun — Tue, 07 Apr 2026 22:26:11 -0700

ABSTRACT

This paper develops an optimal incentive compensation scheme for a project with a predetermined target but no fixed deadline. A principal sponsors the project and hires an agent to execute it, offering a lump-sum payment that depends only on the project's completion time. The agent exerts a baseline effort level but may increase effort at a personal cost to accelerate progress, balancing the reward from completion against the cost of additional effort. The principal aims to maximize expected payoff, defined as the value of project completion minus the payment to the agent, while also internalizing the cost of delays. Project progress is modeled as a reflected Brownian motion with an agent-controlled drift rate. We solve the associated Bellman equation to characterize the agent's optimal effort and derive the principal's optimal incentive scheme. Extensions include settings in which the agent faces a delay penalty or the principal discounts future rewards. These create additional trade-offs between incentive provision and completion timing. Our numerical experiments further indicate that the principal's payoff is nonmonotonic in the payment level: very small payments produce slow completion, and excessively large payments reduce the net benefit. A finite-horizon extension incorporates project termination at a fixed deadline, which further highlights the role of timing incentives. Throughout, we provide numerical illustrations and managerial insights for designing incentive contracts in target-driven project environments.

Leveraging Public Transit for Robotic Deliveries: A Column Generation Approach

Yishay Shapira, Mor Kaspi — Thu, 02 Apr 2026 03:36:43 -0700

ABSTRACT

Autonomous mobile robots (AMRs) are small, electric, wheeled vehicles that operate at pedestrian speeds. In the last-mile delivery service considered in this study, a fleet of AMRs is deployed across multiple recharging depots within a service area, from which they depart to perform point-to-point deliveries. We consider an operational setting in which AMRs are allowed to travel onboard public transit vehicles, with the objective of extending the service range and reducing energy consumption. To model this problem, we propose two mixed-integer linear programming formulations: an arc-based formulation and a path-based formulation. For the latter, we develop a column generation approach coupled with a four-stage dynamic programming algorithm to efficiently solve the underlying pricing subproblem. This solution approach is further embedded within a rolling horizon framework to address dynamic and large-scale operational settings. A case study conducted in a subregion of Tel Aviv demonstrates the ability of the proposed methodology to handle large-scale instances based on real-world parameters. A sensitivity analysis highlights the effects of request time-window widths, public transit capacity, and AMR battery range on the number of requests that can be served. Finally, the results obtained under the rolling horizon framework confirm the feasibility and practical applicability of the proposed column generation approach.

Issue Information

Wed, 01 Apr 2026 23:36:08 -0700

Naval Research Logistics (NRL), Volume 73, Issue 4, Page 445-446, June 2026.

Parallel Machine Scheduling With Periodic Availability Constraints to Minimize Makespan

Lishi Yu, Zhiyi Tan — Wed, 01 Apr 2026 23:36:08 -0700

ABSTRACT

We investigate the scheduling problem on m$$ m $$ parallel and identical machines under periodic availability constraints. Availability periods and unavailability periods appear alternately on each machine. We propose an algorithm, PFFD, and demonstrate that its worst-case ratio is at most m3m+1(5+4β)$$ \frac{m}{3m+1}\left(5+4\beta \right) $$ for m≥3$$ m\ge 3 $$, where β$$ \beta $$ represents the ratio of the duration of an unavailability period to an availability period. Furthermore, we develop the PPTAS algorithm, which can achieve a worst-case ratio arbitrarily close to 1+β$$ 1+\beta $$ and runs in polynomial time when m$$ m $$ is a constant. When m$$ m $$ is part of the input, we show that there does not exist a polynomial time algorithm with worst-case ratio better than 5+4β4$$ \frac{5+4\beta }{4} $$ unless P=NP$$ P= NP $$.

Joint Scheduling of Production and Condition‐Based Maintenance Activities for Make‐To‐Order Deteriorating Manufacturing Systems Under Repairmen Constraint

Siqi Qiu, Danhong Tu, Mohamed Sallak — Wed, 01 Apr 2026 23:36:08 -0700

ABSTRACT

For a make-to-order manufacturing system, randomly fluctuating customer orders determine its production planning, and further influence the deterioration of machines in the manufacturing system. Maintenance activities are then conducted to recover machines to a better status by a limited number of repairmen with different skill levels. Production and maintenance are two conflicting activities for manufacturing systems, and they need to be jointly optimized to attain the global optimum. However, the joint optimization of production and maintenance for manufacturing systems considering randomly fluctuating customer orders with repairman constraints has not been studied yet. In this work, first, the reliability of production machines and a group maintenance model of the manufacturing system considering limited repairmen with different skill levels are constructed. Second, the production plan is first made according to the predicted customer orders and then updated by considering the influence of maintenance with limited repairmen. Third, the production plan and maintenance schedule are jointly optimized to minimize the entire operation and maintenance costs. Finally, the proposed method is applied to a real-world make-to-order manufacturing system of the shell of an air-conditioning compressor to prove its feasibility.

Gig or Employee? Platform's Joint Staffing and Pricing Decisions Under Hybrid‐Staffing Mode

Xuyan Xin, Jiayi Joey Yu, Tianjun Feng — Wed, 01 Apr 2026 23:36:08 -0700

ABSTRACT

In the rapidly evolving landscape of on-demand services, platforms are increasingly adopting a hybrid-staffing mode that combines full-time employees with part-time gig workers. This approach leverages the stability of full-time employees and the flexibility of gig workers, enabling platforms to efficiently meet fluctuating consumer demands. To identify optimal staffing and pricing strategies, we develop a model analyzing these joint decisions. Our findings highlight several key insights. First, the platform's staffing choice is influenced by the salary cost of employees and the labor pool size of gig workers. When both factors are moderate, a hybrid-staffing mode is optimal. Additionally, as the frequency and level of demand surge increase, the platform is more inclined to adopt a hybrid-staffing approach. Second, within the hybrid-staffing framework, gig workers can fulfill multiple roles. They not only act as supplementary resources during peak hours but also maintain operational flexibility in both high- and low-demand states in some cases. This dual role necessitates adaptive staffing and pricing strategies, which may exhibit non-monotonic patterns in response to market changes. Third, the platform consistently benefits from a hybrid-staffing mode, achieving higher profit compared to a single-staffing mode. Moreover, this mode enhances labor and consumer welfare, consistently outperforming the gig-only approach. Our study offers valuable insights for platforms to optimize hybrid staffing and pricing strategies, emphasizing the impact on stakeholder welfare within the ecosystem.

Who Should Donate? A Socially Responsible Supply Chain With Prosocial Customers

Yongbo Xiao, Xiuyi Zhang, Xinyue Cai, Fei Gao — Wed, 01 Apr 2026 23:36:08 -0700

ABSTRACT

Charitable donations have been widely practiced by upstream manufacturers and downstream sellers involved in the supply chain to exercise corporate social responsibility. In addition to generating social benefits (i.e., donation value), donations can also contribute to building a brand's reputation and expanding the customer base, as today's consumers are increasingly socially responsible. This study addresses the question of whether a firm should engage in direct donations or opt for a “free ride” by leveraging supply chain partner's donation initiatives to improve operational performance. We consider a socially responsible supply chain consisting of a manufacturer and a product seller, who seek to maximize his/her mixed objective, that is, a weighted sum of their respective profit and donation value. Both firms consider donation as a means of managing demand, which depends on the total amount donated by the supply chain members. Following a Stackelberg game setting, we study the interactive donation and inventory decisions involved in the supply chain. To encourage both firms to participate in donation and enhance their overall performance, we then investigate several collaborative donation strategies. We show that under equilibrium, only one firm donates and the other simply takes a free ride without donating anything. Furthermore, a firm could offer monetary incentives to its supply chain partner in the form of price discounts or refunds, contingent upon the partner making charitable donations. Such a collaborative donation approach can always benefit the firm offering the monetary incentives, while paradoxically it may harm its supply chain partner (who receives the discounts or refunds).

Coordinating International Humanitarian Inventory by Stochastic Dual Dynamic Programming

Penghui Guo, Jianjun Zhu — Wed, 01 Apr 2026 23:36:08 -0700

ABSTRACT

International humanitarian organizations face challenges in inventory management due to unpredictable disasters, evolving emergencies, participant coordination, long planning horizons, and broad geographical coverage. This paper develops an international humanitarian inventory coordination framework, using multi-stage stochastic programming to make long-term (monthly or quarterly) procurement, inventory, and transportation decisions. As a counterpart of monetary expense in the objective, a target-based disutility, that is monotonically non-increasing and convex, is proposed to measure the suffering caused by insufficient consumption. The model is solved by the generalized Stochastic Dual Dynamic Programming (SDDP), which allows convex recourse functions. The SDDP takes historical demands as input and generates an optimal policy for making future decisions without knowing exact demand information. Unlike deterministic equivalent formulations based on scenario trees, this policy is implementable for out-of-sample data. Extensive numerical experiments are conducted with publicly available data from the United Nations Humanitarian Response Depot, the United Nations Office for the Coordination of Humanitarian Affairs (OCHA), and the EM-DAT international disaster database. The method can generate a policy in under two hours using 216 months (18 years) of data from the 34 most disaster-vulnerable countries or territories where the OCHA works. The SDDP policy offers up to 21% cost savings over myopic or deterministic policies. Results demonstrate that good out-of-sample coordination results can be achieved with a moderate sample size, a reasonable number of iterations, and within the current OCHA organization structure.

Personalized Differential Privacy for Ridge Regression Under Output Perturbation

Krishna Acharya, Franziska Boenisch, Rakshit Naidu, Juba Ziani — Wed, 01 Apr 2026 23:36:08 -0700

ABSTRACT

The increased application of machine learning (ML) in sensitive domains requires protecting the training data through privacy frameworks, such as differential privacy (DP). Traditional DP enforces a uniform privacy level ε$$ \varepsilon $$, which bounds the maximum privacy loss that each data point in the dataset is allowed to incur. However, this one-size-fits-all approach fails to reflect the reality that individuals often have different privacy expectations-depending on factors such as the sensitivity of their data, legal requirements, or personal risk tolerance. As a result, uniform DP can either overprotect some data (hurting utility) or underprotect others (violating privacy needs). In this work, we introduce our Personalized-DP Output Perturbation method (PDP-OP) that enables us to train Ridge regression models with individual per data point privacy levels, in the central privacy model. Our method relies on a technique known as output perturbation, that was introduced by Chaudhuri and Monteleoni, augmented with re-weighting data points according to their privacy levels. Additionally, we provide rigorous privacy proofs and accuracy guarantees for PDP-OP. Thereby, our work distinguishes itself by providing theoretical accuracy guarantees in personalized DP settings in ML, whereas similar previous work only provided empirical evaluations. To demonstrate how our theoretical bounds hold in practice, we evaluate PDP-OP on synthetic and real datasets and with diverse privacy distributions. We show that by enabling each data point to specify their own privacy requirement, we can significantly improve the privacy-accuracy trade-offs compared to non-personalized DP. Finally, we also show that PDP-OP outperforms the personalized privacy techniques introduced by Jorgensen et al. that rely on subsampling as opposed to reweighting.

Competing With Copycats in Reward‐Based Crowdfunding

Feiyang Shen, Weili Xue, Xiaolin Xu, Xiaoqiang Cai — Wed, 01 Apr 2026 23:36:08 -0700

ABSTRACT

In reward-based crowdfunding, creators possess private knowledge regarding the true quality of their innovative products. This presents creators with a trade-off between signaling their true quality level to inform backers (crowdfunding investors) and concealing such information to counteract copycats. In this paper, we establish a stylized model to study the effect of copycats on creators' signaling strategy and their corresponding profits, as well as evaluate countermeasures for combating copycats. Our main findings are as follows. First, the presence of copycats facilitates quality signaling by lowering the cost of separation. Moreover, in contrast to conventional wisdom, both high- and low-type creators may engage in signaling: High types tend to distort reward prices, while low types strategically adjust funding goals. Second, product imitability and potential market size play key roles in the creators' signaling strategy. Creators can only effectively signal their quality when the product imitability is moderate, while the potential market size is huge. However, although the presence of copycats in general is conducive for creators to signal their quality, it diminishes creators' profits due to price competition in the potential market. Third, we evaluate three commonly used countermeasures for combating copycats: A lenient information policy by the platform, as well as a high product inimitability or a quick market entry by creators. We show that a lenient information policy may hurt creators, while improving product inimitability or expediting market entry only works under certain conditions.

Socially Responsible Newsvendor

Chen Hu, Ming Hu, Yongbo Xiao — Wed, 01 Apr 2026 23:36:08 -0700

ABSTRACT

With the advocacy on corporate social responsibility (CSR), it is common for firms to integrate profit objectives with social responsibilities, such as with an aim to boost consumer welfare. We focus on a socially responsible firm that is concerned with its profit as well as consumer surplus and examine four different types of pro-social behavior by the firm: optimizing a weighted average of the expected profit and consumer surplus (referred to as the mixed-objective model), negotiating with pro-social executives (referred to as the Nash bargaining), charitable donations after profit maximization (referred to as the donation), and ensuring the portion of consumer surplus to be a given fraction of the social welfare (referred to as the fairness model). Our results show that under all behaviors, there is a more substantial boost to consumer surplus at the expense of a slight decrease in profit when consumer surplus consideration (referred to as the CSC level) is lower. Among those four behaviors, while maintaining the same profit level, a donation is not the most consumer-surplus-enhancing pro-social behavior among those four behaviors, when the overhead cost is sufficiently high or when a high enough profit level needs to be maintained. This finding challenges Milton Friedman's advocacy that socially responsible businesses should indirectly fulfill their societal duties by first focusing on profit maximization and then redistributing the generated profit for social causes. Our results imply and quantify the managerial insight that in balancing consumer surplus against profit loss, a little commitment to consumers can go a long way. We also shed light on when the firm should choose a decentralized pro-social behavior, such as donations, and when it should incorporate consumer surplus consideration into operational decisions for consumer surplus enhancement.

Offering Electric Vehicle Battery Swapping as a Service: Electric Vehicle Manufacturers or Battery Producers?

Zhong‐Zhong Jiang, Kunyang Li, Christopher S. Tang, S. Alex Yang — Sat, 28 Mar 2026 01:16:04 -0700

ABSTRACT

To accelerate electric vehicle (EV) adoption and overcome the limitations of traditional EV charging—such as long charging time and lack of access to home chargers for urban residents—Battery as a Service (BaaS) has emerged as a promising alternative. Implementing the BaaS model requires investment and operation of battery swapping stations. As both EV manufacturers and battery producers venture into this domain, it raises the question: which party is better positioned to build and operate such infrastructure? We develop a game-theoretical model with one battery supplier and one vehicle manufacturer to compare two BaaS operating models: “manufacturer-operated” model (Model-M) and “supplier-operated” model (Model-S), which differ fundamentally in supply chain structure. In the base model, Model-S induces a larger number of battery swapping stations built. However, Model-M entices more customers to adopt EVs and generates higher profits for the manufacturer. Interestingly, Model-M may also be preferred by the supplier, despite requiring the supplier to cede some decision-making authority. Extending the analysis to a setting with two competing EV manufacturers, we show that the relative efficiency between Model-M and Model-S depends on the degree of downstream competition. Specifically, Model-M tends to be socially optimal in low-competition environments, whereas Model-S gains ground as competition intensifies. Finally, we show that as battery-related costs account for a larger share of total network costs for each additional station built, Model-S becomes the preferred structure for both the manufacturer and supplier over a larger range of parameters. Our paper offers insights for industry leaders and policymakers in determining which party should lead the effort of investing and operating the BaaS model.

On‐Time Meal Delivery Assisted by Drone Resupply

Wenqian Liu, Lindong Liu, Xiangtong Qi — Fri, 20 Mar 2026 03:46:24 -0700

ABSTRACT

Ensuring timely delivery is crucial with the increasing competition in online meal delivery services. This requires the industry to adopt new technologies and the corresponding operational models, including the use of drones. Concerning the desired features of meal delivery, such as safety and reliability, we propose an operational model that incorporates the usage of drones into the current rider-based delivery model. In our approach, known as drone resupply, drones transport meals from restaurants to riders, and riders then deliver them to customers. We aim to address two key issues when implementing this approach. First, at the operational level, models and algorithms are developed to effectively coordinate rider routing and drone scheduling. These algorithms are tailor-made by leveraging the short routes in the meal delivery industry. Second, at the tactical planning level, we reveal managerial insights to aid meal delivery platforms in making informed decisions regarding the implementation of drone resupply solutions. Particularly, drone resupply proves to be more efficient than rider-only mode across diverse order volumes and service ranges, and remains competitive when the promised delivery time is extended. The effectiveness of drone resupply is closely tied to the fleet configuration of riders and drones, as they have different yet complementary roles in achieving on-time delivery. Additionally, restricting one single order per drone trip does not compromise the effectiveness of drone resupply delivery, but necessitates more demanding drone schedules.

Generalized Survival Signature for Repairable Systems: Optimal Allocation of Minimal Repairs

Weiyong Ding, Gaofeng Da, Peng Zhao — Thu, 19 Mar 2026 00:57:13 -0700

ABSTRACT

The minimal repair model is a fundamental framework for analyzing repairable systems. Optimizing the allocation of limited repair resources under this model is both important and challenging. This study introduces an efficient approach based on system signatures to address this problem. We first propose the generalized survival signature for repairable systems (rGSS) and derive a key mixture representation of system reliability. This extends the traditional signature concept beyond repairable series systems to general coherent systems. Leveraging the rGSS, we develop optimal repair allocation strategies between two specific components with comparative criticality via minimal cut (path) sets in arbitrary coherent systems, as well as among all components in four common system structures: Series, parallel, parallel-series, and series-parallel. Numerical examples are provided to illustrate the optimality of the proposed strategies and explore their potential for generalization.

Optimal Maintenance Planning for Mission‐Oriented Systems Considering Dynamic Mission Duration

Kai Li, Yi Luo, Xiujie Zhao, Xun Xiao — Wed, 11 Mar 2026 06:51:21 -0700

ABSTRACT

In practical applications, systems often face variable mission durations influenced by dynamic factors such as maintenance personnel availability, environmental conditions, and real-time operational demands. This paper proposes a condition-based maintenance (CBM) strategy for mission-oriented systems (MOS), addressing the complexities of stochastic mission durations and system degradation. We design a multistate system (MSS) reliability framework that explicitly models dynamic transitions between discrete performance states defined by mission profiles via a discrete-time Markov chain (DTMC) and degradation levels via a Wiener process. Unlike traditional binary-state models, our approach captures degradation state shifts influenced by mission duration, enabling adaptive maintenance policies for systems operating in multistate conditions. The maintenance optimization problem is formulated as a Markov decision process (MDP) via backward dynamic programming to minimize expected maintenance costs. Numerical simulations and sensitivity analyses validate the model's efficacy and adaptability in optimizing maintenance for unmanned aerial vehicles (UAVs). The findings underscore the importance of minimizing maintenance and inspection time, and tailoring strategies to mission characteristics and system costs.

Implications of Strategically Communicating Social Responsibility to Consumers in Online Retailing

Bin Dai, Yurong Liang — Mon, 02 Mar 2026 21:23:06 -0800

ABSTRACT

Recent research on social responsibility (SR) communication has focused on upstream supply chains; however, in this study, we analyze how to strategically communicate SR information to consumers in online retailing. We present the optimal SR communication strategies in agency selling and reselling modes and further explore how strategic SR communication strategies affect optimal prices, supply chain members' profits, the platform's SR standard of reselling mode, and the supplier's selling mode selection. The results indicate that both the platform and supplier choose strategically overstated SR communication when the supplier's SR level is moderate, and overstated SR communication is more likely to occur in agency selling with a low marginal selling cost. Interestingly, we find that the overstated SR communication strategy may create a win–win–win situation for the supplier, platform and consumers. Furthermore, SR communication motivates the supplier to choose agency selling. This phenomenon is driven by the profitability of SR overstating, which stems from consumers' additional willingness to pay for SR. The platform can strategically set a proper SR standard for the reselling mode to regulate overstating and encourage the supplier to choose the reselling mode. Counter-intuitively, the optimal SR standard of the reselling mode is the minimum acceptable SR level of socially conscious consumers for products with a high commission rate and a relatively high SR level.

Selling Formats Choice in the Presence of Third‐Party Resellers: The Roles of Platform Empowerment

You Zhao, Yi Tao, Rui Hou, Dongyuan Zhan — Mon, 02 Mar 2026 21:14:48 -0800

ABSTRACT

Recent years have seen online retail platforms engage in empowerment to boost demand for both their own channels and those of third-party sellers. Under such a context, this paper investigates the selling formats choice problem within a distribution channel where a manufacturer distributes products through both a platform and a third-party reseller, who subsequently sells to consumers via the platform. Our study examines two hybrid selling modes that differ in the selling relationship between the manufacturer and the platform: a manufacturer–platform reselling format, in which the platform purchases products from the manufacturer and resells them to consumers, and a manufacturer–platform agency selling format, in which the manufacturer sells directly to consumers via the platform under a commission contract. In both modes, a third-party reseller sells to consumers through the platform and pays a commission. This phenomenon cannot be fully explained by existing theory on traditional one-to-one e-commerce supply chains. To address this gap, we develop a game-theoretic model to analyze equilibrium selling formats choices, wholesale prices, empowerment levels, and sales quantity decisions. Furthermore, we derive several noteworthy findings. For instance, three firms prefer a manufacturer–platform agency selling format under a high commission rate and a moderate cost of empowerment. Conversely, three firms prefer a manufacturer–platform reselling format under a low commission rate and cost of empowerment. These findings stand in contrast to the cases where the absence of platforms empowerment or platforms only empower themselves. Moreover, under a manufacturer–platform reselling format, our results demonstrate that the platform would empower the third-party reseller if and only if cost of empowerment is high. For a particular product category, our study offers valuable insights for managers grappling with selling formats choices in the presence of third-party resellers when the platform exerts empowerment.

Multi‐Agent Reinforcement Learning for Joint Police Patrol and Dispatch

Matthew Repasky, He Wang, Yao Xie — Fri, 27 Feb 2026 21:27:08 -0800

ABSTRACT

Police patrol units need to split their time between performing preventive patrol and being dispatched to serve emergency incidents. In the existing literature, patrol and dispatch decisions are often studied separately. We consider joint optimization of these two decisions to improve police operations efficiency and reduce response time to emergency calls. We propose a novel method for jointly optimizing multi-agent patrol and dispatch to learn policies yielding rapid response times. Our method treats each patroller as an independent Q$$ Q $$-learner (agent) with a shared deep Q$$ Q $$-network that represents the state-action values. The dispatching decisions are chosen using mixed-integer programming and value function approximation from combinatorial action spaces. We demonstrate that this heterogeneous multi-agent reinforcement learning approach is capable of learning joint policies that outperform those optimized for patrol or dispatch alone. Policies jointly optimized for patrol and dispatch can lead to more effective service while targeting demonstrably flexible objectives, such as those encouraging efficiency and equity in response.

Dynamic Rental Allocation with Condition‐Based Usage Loss

Zimeng Li, Yixuan Xiao, Quan Yuan — Tue, 24 Feb 2026 21:27:33 -0800

ABSTRACT

We study a rental firm's optimal inventory and rental allocation problem considering random inventory loss due to usage. Each product has two conditions: good and bad, both can satisfy demand. After each rental, good products have a depreciation rate to become bad, and bad products have a depreciation rate to become useless. The firm chooses its inventory before the rental season starts and decides how to allocate its good and bad products to satisfy demand in each period during the rental season. If the total inventory of the good and bad product is no greater than the demand, the firm rents out all inventory. Otherwise, the optimal rental quantities are governed by two thresholds that depend on the weighted sum of inventory of the good and bad products adjusted by the demand. Based on the two thresholds, the firm's optimal rental decision can be classified into three cases: rent the bad product first, good product first, or a mix of both products. We also analyze two priority allocation policies: Good-First (GF) and Bad-First (BF), and the cost difference between them and the optimal policy. When the total inventory is moderate, we propose a modified Bad-First (MBF) policy that only optimizes the rental allocation in the last two periods and uses the BF policy for the remaining periods. Such policy performs well and significantly reduces computation complexity. Our numerical study shows that the usage-based loss rates can have a non-monotone impact on the initial inventories and significantly increase the firm's cost.

Customer Reviews Subject to Reporting Bias: Its Influence on Customers, Firms, and Platform

Fengfeng Huang, Pengfei Guo, Yulan Wang — Tue, 10 Feb 2026 22:25:17 -0800

ABSTRACT

Customers tend to share extreme experiences more than moderate ones, a phenomenon known as reporting bias. Reporting bias diminishes the visibility of moderate experiences and polarizes customer opinions. It raises the following questions: How does reporting bias affect customers' evaluations of product quality? How should firms adjust their pricing strategies to address this reporting bias? What can online platforms do to mitigate its impact? We consider a firm selling a product of uncertain quality through an independent platform. Product quality can be high or low, and the probability of high quality is the firm's private information. Customers with heterogeneous preferences arrive sequentially and infer quality based on observed reviews. After consumption, customers decide whether to leave a review, with their review decisions subject to reporting bias. We assess the effectiveness of two common practices: review-solicitation programs and platform interventions that automatically assign positive reviews to unreviewed transactions. We show that customers cannot learn the high-quality probability from reviews subject to reporting bias: they make downward-biased estimations if the high-quality probability exceeds a threshold and upward-biased estimations otherwise. The firm's optimal pricing ultimately converges to a static price that maximizes the expected current profit. Reporting bias hurts a high-quality firm (i.e., a firm whose high-quality probability is above the threshold) but benefits a low-quality firm. While review-solicitation programs can alleviate reporting bias, only a high-quality firm is interested in participating. Platform intervention does not necessarily alleviate reporting bias and, worse yet, may harm high-quality firms. Our findings suggest that online platforms should implement review-solicitation programs to mitigate reporting bias. These programs enhance the quality of information for customers, facilitating more informed purchasing decisions and allowing high-quality sellers to signal their product quality through participation.

Surrogate‐Free Annealing Random Search for Continuous Stochastic Optimization

Feng Xu, Jianqiang Hu, Xiangyu Yang — Fri, 06 Feb 2026 23:06:49 -0800

ABSTRACT

Optimizing blackbox stochastic systems, where only outputs are observable, is challenging due to difficulties in estimating objective function values. Surrogate-based methods, such as interpolation, are widely used but struggle with stochastic noise and high computational costs. To overcome these limitations, we propose surrogate-free annealing random search (SFARS), a novel algorithm that eliminates explicit surrogate models. SFARS employs a value aggregation mechanism based on a predefined discrete point set, enabling efficient Monte Carlo estimators. Theoretical analysis establishes a finite-time probability error bound and guarantees almost sure global convergence with a sub-exponential rate. Numerical experiments demonstrate superior efficiency and robustness, particularly in high-noise environments.

Financing Overconfident Retailers: Bank Loan or Trade Credit?

Jinhong Feng, Ting Zhang, Tsan‐Ming Choi — Wed, 04 Feb 2026 05:17:49 -0800

ABSTRACT

Overconfidence is a cognitive bias of believing an uncertain event to be more certain than it is. This bias is prevalent in small retailers whose operations (e.g., product ordering) rely on financing such as bank financing (where banks finance retailers' product ordering) or trade-credit financing (where retailers order products from their upstream manufacturers without immediate payments). We study which one of the two financing modes should be used to finance an overconfident retailer. Our findings reveal that a higher level of overconfidence harms the retailer under bank financing but does not harm the retailer under trade-credit financing. Under either financing mode, a higher overconfidence level can benefit the supply chain. We further uncover that the manufacturer should offer trade-credit financing when the retailer exhibits low levels of overconfidence and should not offer it when the retailer's overconfidence level is high. When the retailer's overconfidence level is high, trade-credit financing will yield a higher expected profit for the retailer and the supply chain than bank financing. We show that our main findings are robust in a variety of model extensions. Our paper explores the complex interplay between a prevalent psychological factor (overconfidence) in retailers' financing strategies, enriching the literature on supply chain finance and providing insights into the operations of overconfident retailer supply chains.

The Role of Heterogeneous Robots: Operating Policies of Warehousing Systems With the Lift Robot and Ground Robot Collaboration

Zhengmin Zhang, Yeming Gong, Wanying Chen — Sat, 31 Jan 2026 01:40:54 -0800

ABSTRACT

Recently, heterogeneous robots have been adopted in the same warehouse to enhance flexibility and improve system efficiency. Our paper is inspired by a novel heterogeneous robotic warehouse system, namely the lift robot and the ground robot collaborative (LRGR) warehousing system. In the LRGR system, lift robots store and retrieve totes on single deep storage racks. Meanwhile, ground robots transport totes between lift robots and workstations, navigating both the aisles and the space beneath the racks. The performance of an LRGR system is predominantly determined by its operational policies, especially the dwell point and junction point policies that regulate the interactions between lift and ground robots. We propose a fork-join queueing network to assess the performance of LRGR systems under various collaboration policies. An improved matrix-based approximation method is proposed to solve the model. The accuracy of the analytical models is verified by simulation. Our numerical experiments show that implementing the service completion junction point policy in combination with the service completion dwell point policy significantly boosts system efficiency and reduces energy consumption. Our model can provide new perspectives on effective collaboration policies for heterogeneous robotic systems.

Algorithmic Decision‐Making Safeguarded by Human Knowledge

Ningyuan Chen, Ming Hu, Wenhao Li — Mon, 26 Jan 2026 03:44:54 -0800

ABSTRACT

Commercial AI solutions provide analysts and managers with data-driven business intelligence for a wide range of decisions, such as demand forecasting and pricing. However, human analysts may have their own insights and experiences about the decision-making that are at odds with the algorithmic recommendation. In light of such a conflict, we study problems in which humans and AI interact in the decision-making process and characterize the conditions under which human knowledge adds value to AI decision-making. In this paper, we provide a general analytical framework for studying the augmentation of algorithmic decisions with human knowledge: the analyst uses the knowledge to set a guardrail by which the algorithmic decision is clipped if the algorithmic output is out of bounds and appears unreasonable. We show that when the algorithmic decision is asymptotically optimal with large data, the non-data-driven human guardrail usually provides no benefit. However, we point out two common pitfalls of the algorithmic decision: (1) lack of domain knowledge, such as market competition, and (2) model misspecification. In these cases, even with sufficient data, the augmentation from human knowledge can still improve the performance of the algorithmic decision. We propose a model to capture a practical and pervasive type of human–AI interaction in the decision-making process. We derive insights into when the human analyst should follow the algorithmic recommendation. We conclude that even in the era of big data, human knowledge can still play an essential role in decision-making.

Prioritization Competition in a Three‐Sided Shipping Market: A Tri‐Level Game‐Theoretic Approach

Xinyue Pu, Xi Lin, Xiwen Bai — Thu, 15 Jan 2026 14:19:53 -0800

ABSTRACT

With the rapid growth of trade and port services, competition among ports has intensified, leading to strategic measures such as berthing priority. This paper examines the impact of prioritization on port congestion, stakeholder competition, and social welfare in duopoly and fragmented carrier markets. Using a tri-level game-theoretical model, we analyze the equilibrium in a port-carrier-cargo-owner system, focusing on the interactions between decisions of different stakeholders. Key findings include: (1) In a duopoly market, equilibria either show no priority or both ports prioritizing the smaller carrier, while in a fragmented market, the Nash equilibrium converges to not providing priority; (2) Priority solidifies carriers' competitive landscape, prompting fleet adjustments that improve efficiency and thus reducing congestion; (3) While priority can reduce congestion, it may also disadvantage the more competitive port and raise inter-modal transportation costs for some cargo owners; (4) Priority brings benefits for ports in markets with stronger carrier market power but offers little advantage in perfectly competitive markets, aligning with the fact that prioritization is mainly observed in container markets.

Financing Format in an Agricultural Supply Chain With Default Risk

Meiling Zhou, Xianpei Hong, Ying‐Ju Chen — Thu, 15 Jan 2026 00:00:00 -0800

ABSTRACT

This study addresses the gap in existing research by examining how default risk and bankruptcy risk influence financing format choices in agricultural supply chains characterized by yield uncertainty and diseconomies of scale. Using a two-period model, we analyze the equilibrium outcomes of direct financing (DF) and guarantee financing (GF) for a representative family farm facing production-period default risk and sales-period bankruptcy risk. Key findings reveal that lower default risk incentivizes the family farm to increase investment, benefiting both parties, while bankruptcy risk emerges conditionally—depending on output distribution—prompting strategic input reduction to avoid insolvency. The family farm prefers GF when the agricultural firm offers a high risk-free rate, and both the bank rate and default risk are low; otherwise, DF dominates. For the agricultural firm, GF is optimal under high diseconomies of scale or moderate diseconomies with elevated default risk. Notably, DF and GF can achieve win-win outcomes under specific conditions, even with discrete output distributions. The introduction of zero-interest early payment financing (EP) reshapes preferences: EP becomes a viable alternative that aligns incentives, often outperforming DF and GF. This research contributes by systematically integrating agricultural-specific risks into financing decisions and demonstrating how strategic financing design reconciles conflicting objectives in agricultural supply chains.

Robust Assortment Planning With Operational Considerations for E‐commerce Retailers: A Three‐Phase Solution Framework

Song Jiu, Dan Wang, Zujun Ma — Fri, 09 Jan 2026 05:35:12 -0800

ABSTRACT

With the rapid expansion of e-commerce marketplace, major e-tailers are implementing a two-layer distribution network with regional and forward distribution centers (FDCs) for timely delivery. We study an e-commerce assortment planning problem that incorporates operational considerations over a multi-period horizon. Before the horizon starts, an e-tailer determines the assortment presented online and a partial assortment stored at each FDC. In each period, she first replenishes products from the supplier, then allocates inventory to FDCs, finally fulfills realized product demands adaptively, with an objective of maximizing the expected total profit. Given the distributional ambiguity in practice, we formulate a distributionally robust optimization model based on customers' multinomial logit choices, which presents the challenges of fractional nonlinearity and computational complexity. To address them, we propose a tractable three-phase solution framework that integrates the strengths of conic programming, linear decision rule, and the exact branch-and-Benders-cut algorithm. Numerical experiments suggest its good computational performance from various aspects, and explore the values of incorporating operational considerations and assortment selection. A case study using real data from JD.com demonstrates the practical applicability of our framework, which increases profit by 102.6% over JD.com's status quo policy. The profit increment consistently increases over time, suggesting the significant potential long-term benefits from implementing our framework.

Which Model to Choose for E‐Commerce Platforms in Emerging Markets: Centralized or Decentralized?

Jingyi Peng, Feng Yang, Zihao Zhang — Thu, 08 Jan 2026 09:52:43 -0800

ABSTRACT

E-commerce is booming and has gradually become the mainstream shopping method worldwide. In this paper, we focus on two distinct yet thriving models. One is the centralized model (e.g., Amazon and Taobao), the leading player in E-commerce. The other is the decentralized model (e.g., Shopify and WooCommerce), which has become a rising star in recent years. We explore why they succeed and highlight their key characteristics, offering insights for emerging markets on which platform type better fosters growth and social welfare. We develop a stylized model to capture the essence of the centralized model, where transactions occur only through the platform, which offers advertising services to sellers. The decentralized model establishes independent websites for sellers and provides value-added services to improve website display quality. We first consider the equilibrium when the two models operate independently without interference. Then, we analyze a competitive scenario. Here are some key findings. The centralized platform's optimal equilibrium always leads to advertised sellers achieving full or even overexposure. In the decentralized platform, when display quality is lower, it charges lower value-added fees to attract more sellers; when quality is higher, it charges higher fees to target high-value sellers. An interesting finding is that sellers tend to benefit more on the centralized platform when baseline traffic is high, but this advantage diminishes under competition. Under competitive conditions, sellers' and buyers' surplus peaks in centralized platforms when traffic is moderate. Comparing social welfare and platform profit, we find that most value in e-commerce is captured by platforms.

Estimating Confidence Intervals and Regions for Quantiles in Steady‐State Simulations

Lei Lei, Christos Alexopoulos, Yijie Peng, James R. Wilson — Tue, 06 Jan 2026 12:03:26 -0800

ABSTRACT

We propose methods based on batching, sectioning, and generalized likelihood ratios (GLRs) for computing confidence intervals (CIs) and confidence regions (CRs) for quantiles in steady-state simulations. Based on central limit theorems for quantile estimators, CIs and CRs can be computed from a single batch of simulated output using a GLR method to consistently estimate the unknown density function. This paper makes the following contributions: (1) We derive a GLR estimator for distribution sensitivities in the steady-state setting and, under the geometric-moment contraction (GMC) conditions, we establish the uniform consistency of the GLR estimators and the asymptotic validity of the respective CIs and CRs for quantiles. (2) We also establish the asymptotic validity of CIs and CRs for quantiles by batching and sectioning methods for steady-state simulations. Numerical experiments demonstrate the validity of the aforementioned methods as the coverage rates of the CIs and CRs approach the target levels for appropriately large sample sizes. In the steady-state setting, the sectioning and GLR methods demonstrate their respective advantages in different examples.

Manufacturer Incentive Provision for Consumer Deliberation in a Supply Chain With a Fair‐Minded Retailer

Song Huang, Li Bie — Sun, 04 Jan 2026 23:36:22 -0800

ABSTRACT

This paper studies a manufacturer's incentive provision strategy for consumer deliberation in a decentralized supply chain with a fair-minded retailer who is concerned with how the supply chain profit is allocated. Consumers who are initially uncertain about the quality of the product may incur fixed deliberation costs to ascertain their actual valuations of the product to make the proper purchase decisions. We develop a game-theoretic model that accounts for the filtering effect exerted by a fair-minded retailer on the upstream manufacturer's strategic pricing manipulation that intends to influence consumer deliberation, as well as the effect of their interactions on the dynamics of channel relationships. Some interesting results are obtained. First, our findings reveal that the manufacturer has a stronger incentive to lower the wholesale price driven by the retailer's fairness concerns when the deliberation cost is low or high. This incentive becomes especially salient if the retailer is highly averse to inequality such that, under specific conditions, channel coordination can always be achieved through a wholesale price policy in the presence of consumer deliberation. Second, the positive effect of the retailer's inequality aversion on the wholesale price can align her interest with the manufacturer's, such that her profit may be maximized even at a high deliberation cost. Third, when the retailer requires a low profit allocation ratio, the manufacturer may prefer a low wholesale price to motivate the retailer to inhibit consumer deliberation instead of a high wholesale price to provide an incentive for inducing it in the absence of retailer fairness concerns. Consequently, both channel members can benefit from retailer fairness concerns. Finally, the retailer becomes less inclined to fully empower the consumers if she is sufficiently inequity averse, whereas the manufacturer's empowerment decision remains unaffected.

The Weak Core, Partition‐Based Universal Stability, and Their Risk Associations Through A Partial Order

Jian Yang — Wed, 31 Dec 2025 03:35:07 -0800

ABSTRACT

We are concerned with the stability of a transferable-utility cooperative (TU) game. First, the concept of core can be weakened so that the blocking of changes is limited to only those with multilateral backings. This principle of consensual blocking, as well as the traditional core-defining one of unilateral blocking and one straddling in between, can then be applied to partition-allocation pairs. Each such pair is made up of a partition of the grand coalition and a corresponding allocation vector whose components are efficient and individually rational for the various constituent coalitions of the given partition. Among the resulting stability concepts, two are universal in that any game has stable solutions. For a game with strictly positive values, furthermore, its imputations possess fractional interpretations. These would allow a certain ranking between games, in what we call “centripetality”, to imply a clearly describable shift in the games' stable solutions. When coalitions' values are built on both random outcomes and a common positively-homogeneous law-invariant reward function characterizing players' enjoyments from their shares, such comparative statics can help explain the phenomenon that aversion to risk promotes cooperation. Our results can have direct impacts on operations management settings like inventory centralization.

Two‐Stage Adaptive Robust Hub Location Problem Under Demand Uncertainty

Haifeng Zhang, Jian‐Cai Wang, Yuan Gao — Wed, 24 Dec 2025 04:01:08 -0800

ABSTRACT

This paper studies a two-stage adaptive robust hub location problem with multiple assignments under demand uncertainty. In our setting, the capacitated hubs are strategically located in the first stage to minimize the worst-case scenario cost over a budgeted uncertainty set, and the routing decisions, which are adaptive to uncertainty realizations, are made in the second stage to transport all commodities. For the large-scale instances of the problem, we develop a novel column-and-constraint generation approach that integrates Benders decomposition. In this approach, we design a customized Benders decomposition to efficiently solve the master problem involving a subset of uncertain scenarios, in which a tailored cutting plane algorithm is developed to solve Benders dual subproblems and a cut refinement strategy is proposed to generate strong Benders cuts. Besides, to quickly identify possible uncertain scenarios, we reformulate the second-stage problem into a more tractable form, which is further simplified by significantly reducing the number of redundant variables and constraints. Extensive computational experiments on the well-known instances with up to 200 nodes are conducted to evaluate the effectiveness of proposed model and the performance of the solution method. The computational results demonstrate that our developed solution method outperforms the conventional column-and-constraint generation or Benders decomposition. Compared with the two-stage stochastic programming model, our proposed model can provide more reliable and robust solutions with superior out-of-sample performance. The results also illustrate the effect of uncertainty budgets and highlight the advantages of incorporating features such as hub capacity and multiple assignments into our model. Moreover, we make extensions by applying our framework to handle more general polyhedral uncertainty sets.

The Role of Consumer Preferences Information Sharing in Agricultural Products Quality Improvement

Fei Ye, Wenzhuo Li, Wenhui Fu, Yina Li, Qiang Lin — Wed, 24 Dec 2025 00:00:00 -0800

ABSTRACT

In recent years, agribusinesses have shifted their focus towards enhancing the quality of agricultural products. This shift is driven by consumers' growing interest in high-quality agricultural products. However, many agribusinesses face challenges in making optimal quality improvement and production decisions due to limited access to accurate consumer information. To address this, intermediary platforms, equipped with consumer information, not only provide an opportunity for agribusinesses to directly sell their products to consumers, but also consider whether and how to share consumer information with agribusinesses. Motivated by these circumstances, this study considers the supply chain consisting of an agribusiness and an intermediary platform. We investigate the agribusiness's quality improvement strategies and the platform's information-sharing strategy. We find that the quality improvement preferences of the agribusiness are significantly influenced by the platform's information-sharing strategy. Without information sharing, the agribusiness tends to improve quality. However, with information sharing, the agribusiness's willingness to improve quality depends on initial product quality, acceptance level, and quality investment cost. Furthermore, improving quality boosts platform profits when the initial quality is relatively low, the quality investment cost is low, or the product is already positioned at the high end. Information sharing can further enable the platform to gain profits from the agribusiness's quality improvement. Interestingly, the platform's information-sharing strategy also shapes the agribusiness's choices of quality improvement, facilitating a “win-win-win” outcome for the platform, the agribusiness, and consumers. These findings emphasize the importance of information sharing in promoting effective quality improvement, while also highlighting the critical impact of quality improvement and information sharing on the profits of platforms and the development of agribusinesses.

Conditional Generative Modeling for Enhanced Credit Risk Management in Supply Chain Finance

Qingkai Zhang, L. Jeff Hong, Houmin Yan — Sat, 20 Dec 2025 08:36:51 -0800

ABSTRACT

The rapid expansion of cross-border e-commerce (CBEC) has created significant opportunities for small- and medium-sized sellers, yet financing remains a critical challenge due to their limited credit histories. Third-party logistics (3PL)-led supply chain finance (SCF) has emerged as a promising solution, leveraging in-transit inventory as collateral. We propose an advanced credit risk management framework tailored for 3PL-led SCF, addressing the dual challenges of credit risk assessment and loan size determination. Specifically, we leverage conditional generative modeling of sales distributions through Quantile-Regression-based Generative Metamodeling (QRGMM) as the foundation for risk measures estimation. We propose a unified framework that enables flexible estimation of multiple risk measures while introducing a functional risk measure formulation that systematically captures the relationship between these risk measures and varying loan levels, supported by theoretical guarantees. To capture complex covariate interactions in e-commerce sales data, we integrate QRGMM with Deep Factorization Machines (DeepFM). Extensive experiments on synthetic and real-world data validate the efficacy of our model for credit risk assessment and loan size determination. This study explores the use of generative models in CBEC SCF risk management, illustrating their potential to strengthen credit assessment and support financing for small- and medium-sized sellers.

Reconsidering Volume‐Based Drug Procurement Policy: The Consequences of Manufacturers' Optimal Production Planning and Breach Strategies

Nani Zhou, Tong Wang, Guohua Wan — Sat, 20 Dec 2025 08:26:09 -0800

ABSTRACT

This paper examines the impacts of Volume-Based Procurement (VBP) policies on pharmaceutical supply chains, with a focus on the strategic behaviors of manufacturers. VBP aims to reduce costs and stabilize supplies by centralizing procurement through competition based on volumes and prices. However, the VBP policy could also inadvertently incentivize manufacturers to engage in strategic behaviors that may disrupt supply continuity. We develop a comprehensive model to analyze manufacturers' production planning policies under strategic, patient, and reliable modes and the strategy selection decision. For the manufacturer adopting the strategic mode, we establish optimal thresholds of the remaining commitments, above which it is optimal to breach the contract, and further develop a simple yet effective heuristic policy. We also characterize the optimal base-stock ordering policies for the manufacturer under patient and reliable modes and derive upper bounds for optimal order quantities when commitment is extremely large. By numerical studies, we show the criticality of contract parameters in mitigating drug shortages, highlighting that high commitment levels encourage reliability due to steep default penalties, while low commitment levels lead manufacturers to fully fulfill the commitment before supplying higher-priced alternatives. The mid-range commitment, however, tends to foster strategic behaviors, increasing shortage rates. We apply Pareto analysis to demonstrate that, despite criticism for leading to shortages, VBP policies yield net positive outcomes within a wide range of terms in the Oseltamivir case, as the advantages of lower prices and the reinvestment of penalties into public welfare outweigh the drawbacks of manufacturers' strategic behaviors. Furthermore, increasing the commitment quantity and allowing a broader range of higher bidding prices could enable Pareto improvements over current practices. However, simply raising the commitment quantity alone, despite leading to higher breach penalties, is unlikely to yield these improvements without adjustments to other contract terms.

Energy‐Efficient Scheduling of Multi‐Shuttle Automated Storage and Retrieval Systems Considering Heterogeneous Unit Loads

Peiran Tao, Rong Wang, Peng Yang, Yeming Gong — Wed, 17 Dec 2025 18:56:14 -0800

ABSTRACT

We study the mixed storage and retrieval scheduling problem of multi-shuttle automated storage and retrieval systems (AS/RSs), considering heterogeneous unit loads to minimize energy consumption. Different from travel time, the energy consumption incurred by crane movement between any two locations is not fixed and can not be easily calculated beforehand. As energy consumption is related to the weight of all unit loads on shuttles, the energy consumed between any two locations is not constant but a variable affected by the scheduling of both storage and retrieval operations at other locations. Considering this aspect of the objective, we formulate the energy-efficient mixed storage and retrieval scheduling problem as a mixed integer nonlinear programming model and propose a branch-and-price algorithm framework to solve the problem. In the pricing subproblem of the framework, we tailor a labeling algorithm based on a proposed dominance judgment policy and embed the Kuhn-Munkres algorithm in the labeling algorithm. The results of computational experiments show that the proposed algorithm outperforms the Gurobi solver in terms of both solution quality and computational efficiency. Based on the results, we give suggestions on the design and operation of warehouses. Our work will be insightful for warehouse managers aiming to operate multi-shuttle AS/RSs with reduced energy consumption and carbon emission and contribute to sustainable warehousing practices.

A Novel Two‐Stage Flexible Flow Shop Batch Scheduling Model for Grinding Workshops

Jun Xu, Xuanru Pan, Chao‐Bo Yan, Guo‐Qiang Fan, Zhixin Liu — Wed, 17 Dec 2025 00:00:00 -0800

ABSTRACT

The growing need for data storage in data centers has increased the demand for mechanical hard disks due to their low cost and high reliability. Aluminum substrates are the most popular base plates for mechanical hard disks because of their high hardness and low cost. The subtractive grinding process used to produce aluminum substrates consists of two stages: rough grinding and fine grinding. Each stage uses the same number of grinding batch machines, characterized by identical machine capacity and stage-dependent fixed grinding speeds. The grinding speed during rough grinding is higher than that during fine grinding. The removal thickness for each aluminum substrate falls within a certain interval, and these intervals must overlap when processed in the same batch. The allocation of removal thickness between the two stages determines the batch processing time for each stage, making it a decision variable within a specified interval. Inspired by this process, this paper develops a novel two-stage flexible flow shop batch scheduling model to improve grinding efficiency and proves it is NP-hard. Subsequently, we provide two lower bounds. Based on different removal thickness allocation strategies and batch assignment rules, we design five approximation algorithms and analyze their worst-case performance ratios. Computational experiments demonstrate the varying performance of the proposed algorithms and provide managerial insights for real workshop production.

Integrated Production and Transportation Problem With Order Waiting

Yuejuan Zhu, Zhixue Liu, Feng Li, Julong Wang — Tue, 16 Dec 2025 05:45:51 -0800

ABSTRACT

We study production and transportation integration in a make-to-order environment with time-dependent waiting and inventory holding costs. In this problem, manufacturers receive orders from customers, produce and then transport the products to customers, resulting in associated production and transportation costs. Orders, upon receipt, are not immediately produced, and likewise, once produced, they are not instantly transported. This delay in processes incurs additional costs, specifically order waiting costs resulting from the gap between receipt and transportation, and inventory holding costs due to the storage of products preceding their transportation. The objective is to determine an integrated plan of production and transportation that minimizes the total cost of production, waiting, inventory holding, and transportation. We first show that the problem is strongly NP-hard, and then develop a primal-dual heuristic algorithm with a worst-case bound of two. The computational results demonstrate that our algorithms perform well based on randomly generated instances. Finally, we incorporate both limited production capacity and unit production cost into the problem, and extend the proposed algorithm to solve the problem. Additionally, we conduct computational experiments to demonstrate the efficiency and efficacy of the algorithm.

Strategic Joining and Optimal Pricing in a Single‐Server Batch Arrival Queue With Different Information of Batch Size

Kaili Li, Jinting Wang, Zhe George Zhang — Sun, 14 Dec 2025 05:19:37 -0800

ABSTRACT

This study explores the strategic behavior of customers in a single-server batch arrival queue, where the batch size is a random variable. Each customer makes a decision to either join or balk under a linear reward-cost structure. The utility of each customer is contingent on her individual decision and the decisions of her companions. The research considers two scenarios distinguished by the level of information: the batch size observable case, where customers are aware of their batch size information, and the batch size unobservable case, where customers lack information upon arrival. In both cases, a unique Nash equilibrium joining strategy and a socially optimal joining strategy are derived. To bridge the gap between the individual equilibrium joining strategy and the socially optimal strategy, a proposed fee imposed on customers is introduced. By comparing outcomes across both information scenarios, we derive managerial insights regarding optimal information-disclosure policies. The key insight from our numerical experiments is that regulating the mean and variance of the batch size distribution is key to maximizing overall system throughput and social welfare.

On Order Restricted Inference in Multi‐Step Stage Life Testing for a General Family of Distributions

Erhard Cramer, Ayan Pal, Debashis Samanta — Fri, 12 Dec 2025 16:59:26 -0800

ABSTRACT

Recently, k$$ k $$-step stage life testing (SLT) has been proposed by Laumen and Cramer (2021) as a natural extension of progressive censoring with fixed censoring times (PC-FCT) as well as of simple step-stress accelerated life testing (SSALT). In this context, a failure-rate based multi-step SLT model is explored where the stress-specific time-to-failure- distributions of the experimental units conform to the flexible proportional hazard (PH) family of distributions. In a multiple SSALT experiment, with an elevation in the stress level, the load on the experimental units increases and hence the tendency of the units to fail increases. Naturally, it gives rise to an ordering among the mean times-to-failure of the stress-specific failure time random variables. We propose to use the method of generalized isotonic regression in order to carry out the associated order restricted inference. The order restricted maximum likelihood estimates (MLEs) of the unknown model parameters are obtained employing the principle of order restricted maximization of the likelihood function. In addition, the interval estimates of the model parameters are computed based on the observed Fisher information matrix. An extensive simulation study is performed to see the effectiveness of the proposed method. Finally, two datasets are analyzed in detail to illustrate the proposed estimation methodology.

Blessing or Curse: Third‐Party Product Variety for First‐Party Product Development in Online Platforms

Jinzhi Li, Lin Tian, L. Jeff Hong — Fri, 05 Dec 2025 06:26:51 -0800

ABSTRACT

In recent years, to leverage their information advantage and expand revenue, many online platforms have built their own brands and developed products themselves. There is a widespread concern that these first-party products, utilizing third-party information, may gain popularity and encroach upon the market share of third-party manufacturers. Consequently, some manufacturers have started taking countermeasures to protect their information by reducing the variety of their products. This paper presents an analytical model to examine the strategic interactions of first-party product development in an online marketplace involving a platform and a manufacturer. The focus is on the role of the manufacturer's product variety to gain a better understanding of how its trade-offs impact market outcomes. Our analysis reveals three key insights. First, in terms of the platform's entry into first-party product development, the manufacturer's product variety has non-monotonic effects on the platform's and the manufacturer's profits. Second, in highly competitive markets, when the manufacturer's product variety is high, the platform will launch first-party products; whereas when the manufacturer's product variety is low, it is not optimal for the platform to do so. Third, the manufacturer can benefit from strategically reducing the product variety to discourage the platform's first-party product development. Furthermore, in anticipation of the manufacturer's strategic decision to reduce product variety, the platform ought to refrain from initiating first-party product development in highly competitive markets. Otherwise, the platform's entry intention may provoke the manufacturer's strategic countermeasures and cause tangible loss to all parties, including the manufacturer, consumers, and the platform itself. These results provide useful managerial and regulatory implications.

Integrated Routing and Scheduling Problem for the Airborne Deployment of Heterogeneous Multistatic Sonar Networks

Fri, 21 Nov 2025 03:36:36 -0800

ABSTRACT

In this paper, we study the routing problem of an airborne carrier deploying a heterogeneous Multistatic Sonar Network (hMSN) for Anti-Submarine Warfare (ASW). An hMSN combines monostatic (co-located source and receiver) and bistatic (separate source and receiver) sonar systems from different buoy pairings. In fine, the goal is to determine the optimum deployment sequence for these limited-life buoys, air-dropped by the carrier. That is, we seek the deployment that achieves the optimum average insonification rate over a pre-determined mission duration, which could be several consecutive hours of on-site flying. This average coverage is calculated via hMSN snapshots taken at regular time steps, hence the scheduling aspect corresponding to the buoys actually active at time t$$ t $$, that is, previously deployed and not yet terminated. This new problem in the literature is addressed through a two-phase approach with (i) the spatial (static) optimization of the hMSN and (ii) the temporal (dynamic) optimization of the hMSN deployment sequence. In this way, the solution obtained as output for the coverage problem (i) is used as input data for the routing-scheduling problem (ii). To solve it, we propose an exact method by means of a Mixed-Integer Linear Program (MILP) and an approximate method by means of an ad-hoc constructive heuristic. The results show that it is possible to use these methods to efficiently find a deployment sequence of these networks of operational dimensions with a discretization as fine as a minute, and for missions lasting up to 8 h.

A Robust Bayesian Framework for Degradation State Identification in the Presence of Outliers

Ancha Xu, Juan Wang, Di Zhu, Zhen Chen, Yijun Wang — Wed, 19 Nov 2025 05:37:51 -0800

ABSTRACT

Accurate degradation state estimation is critical for predictive maintenance, yet it is often compromised by measurement outliers and parameter uncertainty. Existing methods either assume Gaussian measurement errors, which are sensitive to outliers, or overlook parameter uncertainty, leading to overconfident predictions. To address these challenges, we propose a Bayesian online degradation state estimation framework that integrates robust error modeling with parameter uncertainty quantification. Specifically, we model measurement errors using a Student's-t$$ t $$ distribution to handle outliers and employ variational Bayes with Laplace and Gamma approximations to estimate posterior distributions of degradation states and parameters efficiently. This framework enables real-time updates, ensuring adaptability to dynamic operating conditions. Furthermore, based on the estimated degradation states, we derive real-time remaining useful life predictions and dynamic maintenance strategies under a cost function model. Numerical experiments and case studies demonstrate the framework's robustness, computational efficiency, and practical applicability.

Strategic Information Disclosure and Encroachment in a Co‐Opetitive Supply Chain: Bayesian Persuasion Perspective

Erbao Cao, Yaodan Zhang, Ying‐Ju Chen — Wed, 19 Nov 2025 00:00:00 -0800

ABSTRACT

We investigate the optimal information disclosure policy in an environment where an online platform, possessing private market demand information, interacts with an original brand manufacturer (OBM) and a contract manufacturer (CM) with asymmetric brand power. The OBM enjoys a brand advantage, while the CM may encroach on the market if the profitability justifies the encroachment cost. The platform, leveraging data advantages, influences the manufacturers' decisions through strategically designed disclosure policies. The optimal disclosure policy depends on the CM's encroachment cost and the OBM's brand advantage. First, full disclosure is optimal when the encroachment cost is sufficiently high or low. Second, partial disclosure is optimal for distorting the posterior expectation about market demand downward when encroachment costs are moderately high, or upward when they are moderately low. Third, partial disclosure can induce encroachment if the brand advantage is small or deter it if the brand advantage is large. The platform profits from the optimal policy, which may come at the expense of the OBM and CM. We demonstrate that the optimal policy can enhance profits for the entire supply chain under intense competition and a large brand advantage. This study provides a critical framework and practical insights for understanding strategic decisions and channel management.

Information Sharing, Selling Modes, and Logistics Service Strategies in an E‐Commerce Supply Chain

Xiaogang Lin, Zichao Liao, Shuai Yan, Yiwen Bian — Wed, 12 Nov 2025 01:21:50 -0800

ABSTRACT

E-commerce sales have become increasingly common and important. Manufacturers generally wholesale products to platforms that subsequently resell to customers (reselling), or directly sell to customers through platforms by paying a commission rate (agency selling). In both selling formats, manufacturers can choose between the platforms' logistics and their own logistics for product delivery, generating four typical modes: (1) reselling+platform logistics (RP); (2) reselling+manufacturer logistics (RM); (3) agency selling+platform logistics (AP); and (4) agency selling+manufacturer logistics (AM). Motivated by the fact that platforms possess superior demand information and apply it to improve operational decisions, this paper investigates how a platform strategically shares information to induce a manufacturer to adopt a more efficient mode. We find that the manufacturer's mode preference is significantly influenced by information asymmetry. If the platform possesses highly accurate information (i.e., the information accuracy level is high), the manufacturer prefers the AP mode, and this preference strengthens as the platform's logistics investment efficiency improves. Otherwise, the manufacturer prefers the AM (RP) mode if the logistics improvement efficiency is low (high). However, when the platform's information accuracy level is high but investment efficiency is low, the platform will commit to sharing information to induce the manufacturer to adopt the AM (RP) mode rather than the AP mode if logistics improvement efficiency is low (high). Notably, as the platform's logistics investment efficiency gradually improves, these induced changes in mode selection can be reversed under certain conditions, while the negative impact of information sharing on consumer surplus and social welfare becomes increasingly significant.

Platforms' Search‐Pattern Preference: The Role of Assortment

Qingwei Jin, Lin Liu, Yi Yang — Sun, 09 Nov 2025 00:00:00 -0800

ABSTRACT

Many digital platforms provide a search environment for consumers to evaluate sellers' products. We investigate a strategic platform's preference over two classical search patterns—parallel versus sequential—keeping in check consumers' search behavior (how many products and attributes to evaluate) and sellers' strategies (price and assortment decisions). In the benchmark model with exogenous assortment level, our results show that the platform prefers a parallel (sequential) pattern when the search cost is small (large) or when the assortment level is high (low). However, when the assortment level is a decision by sellers, the platform's preference will be altered qualitatively: The platform prefers a parallel (sequential) pattern when the search cost is large (small), and the analytical predictions are generally consistent with observations in practice. We have identified several novel effects that are built on the fundamental difference between parallel and sequential patterns and use them to explain the platform's search-pattern preference. Interestingly, our paper shows that the platform can strategically use operational means (assortment prevention effect) and marketing means (pricing prevention effect) to manipulate consumers' search to maximize its profit.

Optimal condition‐based parameter learning and mission abort decisions

Li Yang, Yuhan Ma, Fanping Wei, Qingan Qiu — Tue, 20 Aug 2024 00:00:00 -0700

Abstract

Unexpected failures of safety-critical systems during mission execution are not desirable in that they often result in severe safety hazards and significant financial losses. Prompt mission abort based on real-time degradation data is an effective means to prevent such failures and enhance system safety. In this study, we focus on safety-critical systems that experience cumulative shock degradation and fails when the degradation exceeds a failure threshold. Real-time degradation measurements are obtained via sensor monitoring, which are stochastically related to the hidden degradation parameters that vary across components. We formulate the optimal mission risk control problem as a sequential abort decision-making problem that integrates adaptive parameter learning, following which a dynamic Bayesian learning approach is exploited to sequentially infer the uncertain degradation parameters by utilizing real-time sensor data. The problem is constituted as a finite horizon Markov decision process to minimize the expected costs associated with inspections, mission failures and system failures. We derive a series of structural properties of the value function and demonstrate the existence of optimal abort thresholds. In particular, we establish that the optimal policy follows a state-dependent control limit policy. Additionally, we study the existence and monotonicity of control limits associated with both the number of inspections and degradation severities. We demonstrate the performance of the proposed risk management policy through comparative experiments that show substantial superiorities over risk-induced loss control.