Selected Supply Chain Cluster Research Projects

This project closely adheres to the requirements of the “Guidelines” and emphasizes research on the following three main themes: mechanism design theory and empirical research in competitive scenarios, mechanism design theory and empirical research in cooperative scenarios, and mechanism design theory and empirical research under redistribution objectives. This project explores the following five topics within a unified framework: non-market competition design, non-market non-competition design, non-market design and public policy, information-related mechanism design, and mechanism design based on bounded rationality. We aim to address a series of progressively important scientific questions: firstly, to establish rigorous normative models to provide targeted and actionable recommendations for non-market design, supporting decision-makers in policy formulation. Secondly, through empirical research, to validate the effectiveness and significance of non-market design in policy applications. Thirdly, to advance the application and development of information-related mechanism design through economic experimental methods. Moreover, to introduce the concept of bounded rationality to extend, refine, and enhance traditional mechanism design theory, accompanied by corresponding experimental verification.

When the lead time is long, for products with a relatively short selling season (e.g., fashion items), firms may just stock their stores once before the season starts. In these circumstances, inventory transhipment between stores or distribution centers (DCs) often becomes the only way to reduce supply-demand mismatch. This project studies the initial stocking and in-season transhipment decisions of a firm selling its product through multiple stores/DCs.

Specifically, at the beginning of a selling season, the firm decides how much inventory to stock at each store. Over the season, demand to each store follows independent compound Poisson processes. If a store experiences stock-outs, it can tranship inventory from other stores at a cost. We will first apply stochastic dynamic programming to derive structural properties and characterize optimal policies for a three-location problem, then extend our approach to more general multi-location systems. Expecting challenges in high-dimensional state spaces, we will design simple algorithms and heuristics with provable near-optimal performance. Our revenue-loss analysis will show that the gap relative to the optimal policy is bounded by a constant. A numerical study using data from a firm storing and distributing high-value medical supplies will be conducted to test the performance. We will further examine the situation where exact demand distributions are unknown, developing data-driven methods for initial stocking and transhipment decisions.

This project fills an important gap in the field of stochastic service systems regarding human behaviors, and provides practical guidance for managing large-scale service systems. Through preliminary analysis, the project is expected to yield fruitful results in terms of paper publication, practical application, and training of students.

The uncertainties of returned product quantities and customer demand, and their time dependence, impose great operational challenges. We study infinite-horizon, periodic-review remanufacturing inventory systems with random returns and random demand, distinct from prior studies by considering demand-dependent returns. The firm decides on remanufacturing and manufacturing quantities to minimize the expected long-run average cost per period. Due to the “curse of dimensionality,” we focus on simple and implementable policies, such as base-stock-type policies (with or without disposal). We develop approximations for policy parameters that not only reveal useful structural insights but also enable easier implementation. The effectiveness of these approximations and simple policies is demonstrated numerically using both simulated and real data. We also study the impact of mis-specifying the return forecasting models and consider various extensions such as multiple types of cores, disposal options, and the joint management of new and remanufactured products.

We consider a firm offering two main trade-in schemes: trade-in-for-upgrade and trade-in-for-cash, targeting different customer segments. The firm refurbishes acquired products for resale while setting prices dynamically to maximize its total expected profit over a finite horizon. Customers with old devices choose which scheme to trade, and if they choose trade-in-for-upgrade, whether they prefer a new product. New customers without old devices decide to buy a refurbished product or a new product. We aim to develop effective algorithms for dynamically setting trade-in and resale prices. The problem is high dimensional, and so we employ approximate dynamic programming techniques to derive simple solutions and bounds. We collect data from representative companies to validate our model and demonstrate the effectiveness of our proposed solutions.

These problems are dynamic and include uncertain factors. We construct quantitative models to describe them and conduct thorough analysis to derive structural properties, from which we aim to characterize optimal control policies (e.g., warehouse space rental, inventory replenishment, product assortment). We will validate the models with real data and analyze practical implications. The project will advance warehousing and inventory management literature and provide solutions and guidelines on emerging problems in warehouse management.

These problems are dynamic and include uncertain factors. We construct quantitative models to describe them and conduct thorough analysis to derive structural properties, from which we aim to characterize optimal control policies (e.g., warehouse space rental, inventory replenishment, product assortment). We will validate the models with real data and analyze practical implications. The project will advance warehousing and inventory management literature and provide solutions and guidelines on emerging problems in warehouse management.

This project focuses on approximation algorithms for dynamic remanufacturing systems with stochastic product returns and demands that are correlated across periods — a realistic factor many prior studies omit. We construct and analyze novel policies and cost-accounting schemes to develop computationally efficient algorithms with constant performance guarantees. We examine simple policies that balance costs such as inventory holding, demand backlogging, remanufacturing, and manufacturing. The project also considers system extensions (e.g., disposal, capacity constraints, multiple product types), comparing the proposed approximation policies to naive or myopic approaches. The ultimate goal is to provide effective, implementable strategies for real-world remanufacturing systems.