Knowledge that Transforms

As is the tradition at the journal, we are pleased to publish the extended abstracts from the winners of the 2005 MSOM Student Paper Competition. We do this to celebrate the achievements of these young scholars and to provide you with the opportunity to learn about their work in more detail. The 2005 prize committee was chaired by Serguei Netessine (University of Pennsylvania). The other committee members were Dan Adelman (University of Chicago), Philipp Afeche (University of Chicago), Yossi Aviv (Washington University), Fernando Bernstein (Duke University), Rene Caldentey (New York University), Jiri Chod (Boston College), Francis de Véricourt (Duke University), Laurens Debo (Carnegie Mellon University), Vinayak Deshpande (Purdue University), Wedad Elmaghraby (University of Maryland), Jeremie Gallien (Massachusetts Institute of Technology), Noah Gans (University of Pennsylvania), Vishal Gaur (New York University), Roman Kapuscinski (University of Michigan), Costis Maglaras (Columbia University), Ozalp Ozer (Stanford University), Rodney Parker (Yale University), Erica Plambeck (Stanford University), Ioana Popescu (INSEAD), Nils Rudi (INSEAD), Sergei Savin (Columbia University), Alan Scheller-Wolf (Carnegie Mellon University), Nicola Secomandi (Carnegie Mellon University), Xuanming Su (University of California, Berkeley), Julie Swann (Georgia Tech), Beril Toktay (Georgia Tech), Brian Tomlin (University of North Carolina), Tunay Tunca (Stanford University), Senthil Veeraraghavan (University of Pennsylvania), and Assaf Zeevi (Columbia University). The 2005 prize winners are: First Place—Owen Q. Wu (University of British Columbia); Second Place—Guillaume Roels (Massachusetts Institute of Technology) and Ping Josephine Xu (Massachusetts Institute of Technology); Honorable Mention—Richard Lai (Harvard University), Qian Liu (Columbia University), and Gigi Yuen (Northwestern University) Extended abstracts are presented for the following papers: (1) Optimal Control and Competitive Equilibrium of Production-Inventory Systems with Application to the Petroleum Refining Industry by Owen Q. Wu; (2) The Price of Information: Inventory Management with Limited Information About Demand by Guillaume Roels; (3) The Benefits of Re-Evaluating the Real-Time Fulfillment Decisions by Ping Josephine Xu; (4) Inventory and the Stock Market by Richard Lai; (5) Strategic Capacity Rationing to Induce Early Purchases by Qian Liu; and (6) Operations Systems with Discretionary Task Completion by Gigi Yuen.

Consider a firm that owns a fixed capacity of a resource that is consumed in the production or delivery of multiple products. The firm strives to maximize its total expected revenues over a finite horizon, either by choosing a dynamic pricing strategy for each product or, if prices are fixed, by selecting a dynamic rule that controls the allocation of capacity to requests for the different products. This paper shows how these well-studied revenue management problems can be reduced to a common formulation in which the firm controls the aggregate rate at which all products jointly consume resource capacity, highlighting their common structure, and in some cases leading to algorithmic simplifications through the reduction in the control dimension of the associated optimization problems. In the context of their associated deterministic (fluid) formulations, this reduction leads to a closed-form characterization of the optimal controls, and suggests several natural static and dynamic pricing heuristics. These are analyzed asymptotically and through an extensive numerical study. In the context of the former, we show that “resolving” the fluid heuristic achieves asymptotically optimal performance under fluid scaling.

We develop an approximation scheme for performance evaluation of serial supply systems when each stage operates like a single-server queue, and its planned inventories are managed according to a base-stock policy. We also present a near-exact matrix-geometric procedure for benchmarking our approximation relative to two other methods proposed in the literature. Through numerical tests, we demonstrate that our method is superior, both for performance estimation and for policy parameter optimization. Using this technique, we then perform experiments that address the following issues. What proportion of the optimal total inventory should managers allocate to upstream production stages to minimize the sum of inventory and backorder costs? If managerial action could lower holding cost rate or add capacity, which stages of the supply system should be targeted for maximum net benefit? Such concerns have been the subject of several recent studies relating to supply networks with constant and random independent lead times. We shine light on optimal actions for serial supply systems that experience congestion.

Radio-frequency identification (RFID) technology will reduce the costs of transactions to both customers and providers. This is a good thing, mostly. But it will be harder than people suppose, and it may have some adverse consequences. There are numerous technical and policy problems, including privacy and security issues. All these likely will be resolved, but only with time, experience, and money. Thus, it is unclear just how RFID will be used, just when various applications will become feasible. Also, as transaction costs decline, providers will be inclined to charge for things that are now free, such as public parks and congested roads. We may wait less at each transaction point, but such points may proliferate. Rather than obsess about this one technology, we should aim to think creatively about transactions and ways to improve them. Transactions affect our lives in many ways. A transaction is a process, not just an event. When people are involved, the process becomes intertwined with other intricate processes—product choice, social interaction, and the expression of power relations. RFID certainly offers the potential for improvement, but there are some lower-tech methods worth considering. This is not a systematic survey, but rather an essay, a sketch in broad strokes of an immense and varied landscape. Many topics are touched on lightly. The goal is to stimulate thought and discussion. More questions are raised than answered. In places the tone is personal, some might say eccentric.

This paper examines the interactions between routing and inventory-management decisions in a two-level supply chain consisting of a cross-docking warehouse and N retailers. Retailer demand is normally distributed and independent across retailers and over time. Travel times are fixed between pairs of system sites. Every m time periods, system inventory is replenished at the warehouse, whereupon an uncapacitated vehicle departs on a route that visits each retailer once and only once, allocating all of its inventory based on the status of inventory at the retailers who have not yet received allocations. The retailers experience newsvendor-type inventory-holding and backorder-penalty costs each period; the vehicle experiences in-transit inventory-holding costs each period. Our goal is to determine a combined system inventory-replenishment, routing, and inventory-allocation policy that minimizes the total expected cost/period of the system over an infinite time horizon. Our analysis begins by examining the determination of the optimal static route, i.e., the best route if the vehicle must travel the same route every replenishment-allocation cycle. Here we demonstrate that the optimal static route is not the shortest-total-distance (TSP) route, but depends on the variance of customer demands, and, if in-transit inventory-holding costs are charged, also on mean customer demands. We then examine dynamic-routing policies, i.e., policies that can change the route from one system-replenishment-allocation cycle to another, based on the status of the retailers’ inventories. Here we argue that in the absence of transportation-related cost, the optimal dynamic-routing policy should be viewed as balancing management’s ability to respond to system uncertainties (by changing routes) against system uncertainties that are induced by changing routes. We then examine the performance of a change-revert heuristic policy. Although its routing decisions are not fully dynamic, but determined and fixed for a given cycle at the time of each system replenishment, simulation tests with N = 2 and N = 6 retailers indicate that its use can substantially reduce system inventory-related costs even if most of the time the chosen route is the optimal static route.

We analyze a serial base-stock inventory model with Poisson demand and a fill-rate constraint. Our objective is to gain insights into the linkage between the stages to facilitate optimal system design and decentralized system control. To this end, we develop a closed-form approximation for the optimal base-stock levels. The development consists of two key steps: (1) convert the service-constrained model into a backorder cost model by imputing an appropriate backorder cost rate, and then adapt the single-stage approximation developed for the latter, and (2) use a logistic distribution to approximate the lead-time demand distribution in the single-stage approximation obtained in (1) to yield closed-form expressions. We then use the closed-form expressions to conduct sensitivity analyses and establish qualitative properties on system design issues, such as optimal total system stock, stock positioning, and internal fill rates. The closed-form approximation and most of the qualitative properties apply equally to the model with a backorder cost, although some differences do exist. Other results of this study include a bottom-up recursive procedure to evaluate any given echelon base-stock policy and lower bounds on the optimal echelon base-stock levels.

Consider two markets of different sizes but similar costs and fare structure. All other things being equal, is an airline’s expected revenue larger in the market with larger demand? If not, under what circumstances is it possible to compare expected revenues without carrying out a detailed analysis? In this article, we provide answers to these questions by studying the relationship between the optimal expected revenue and the demand distributions when the latter are comparable according to various stochastic orders. For the two-fare class problem with dependent demand we obtain three results. We show that airlines should prefer lesser positive dependence between fare classes when marginal demand distributions are the same. We also describe particular dependence structures under which stochastically larger marginal demand distributions improve optimal expected revenue. Finally, when the dependence between effective demands in the two fare classes arises due to “sell ups,” we show that stochastically larger marginal demand distributions should be preferred. (Sell ups occur when some lower-fare-class customers buy higher-fare tickets upon finding that the former tickets are sold out.) For a problem with an arbitrary number of fare classes and independent demands, we show that stochastically larger demand distributions should be preferred. Numerical examples demonstrating the effect of parameterized demand distributions (with appropriate stochastic ordering) and dependence structures are also presented.

As manufacturers in various industries evolve toward predominantly make-to-order production to better serve their customers’ needs, increasing product mix flexibility emerges as a necessary strategy to provide adequate market responsiveness. However, the implications of increased flexibility on overall system performance are widely unknown. We develop analytical models and an optimization-based simulation tool to study the impact of increasing flexibility on shortages, production variability, component inventories, and order variability induced at upstream suppliers in general multiplant multiproduct make-to-order manufacturing systems. Our results show that Partial flexibility leads to a considerable increase in production variability, and consequently in higher component inventory levels and upstream order variability. Although a modest increase in flexibility yields most of the sales benefits, production variability is reduced as more flexibility is added to the system. Consequently, investments in additional flexibility may be justified when component inventories are expensive, or simply by the benefits associated with the smoother production. The performance of flexible systems is highly dependent on the capacity allocation policies implemented. Policies that evenly distribute product demands to the available plants lead to consistently better performance because they avoid the misplacement of inventories by replicating the performance of a single-plant system. These insights and the simulation tool can be used by practitioners to guide the design of their flexible production systems, trading off the initial capital outlay versus the sales benefits and the expected operational costs.

We study determining prices and production jointly in a multiple period horizon under a general, nonstationary stochastic demand function with a discrete menu of prices. We assume that the available production capacity is limited and that unmet demand is lost. We incorporate discretionary sales, when inventory may be set aside to satisfy future demand even if some present demand is lost. We analyze and compare partial planning or delayed strategies. In delayed strategies, one decision may be planned in advance, whereas a second decision is delayed until the beginning of each time period, after observing the results of previous decisions. For example, in delayed production (delayed pricing), pricing (production) is determined at the beginning of the horizon, and the production (pricing) decision is made at the beginning of each period before new customer orders are received. A special case is where a single price is chosen over the horizon. We describe policies and heuristics for the strategies based on deterministic approximations and analyze their performances. Computational analysis yields additional insights about the strategies, such as that delayed production is usually better than delayed pricing except sometimes when capacity is tight. On average, the delayed production (pricing) heuristic achieved 99.3% (99.8%) of the corresponding optimal strategy.

Two critical decisions must be made daily when managing multiechelon repair and distribution systems for service parts: (1) allocating available repair capacity among different items and (2) allocating available inventories to field stocking locations to support service operations. In many such systems, procurement lead times for service parts are lengthy and variable, repair capacity is limited, and operational requirements change frequently—resulting in demand processes that are highly uncertain and nonstationary. As a consequence, it is common to have many items in short supply while others are abundant. In such environments, integrated real-time decision-support tools can provide significant value by reducing the impact of inventory imbalances and responding appropriately to the volatile nature of the demand processes. By “integrated” and “real-time,” we mean (respectively) tools that simultaneously consider key aspects of the current state of the operating environment in deciding what items to repair, where to ship available units, and by what mode to ship them. In this paper, we develop an integrated real-time model for making repair and inventory allocation decisions in a two-echelon reparable service parts system. We formulate the decision problem as a finite-horizon, periodic-review mathematical program, we show it can be formulated as a large-scale linear program, and we develop a practical heuristic method for solving the problem approximately. By simulating the operation of a service parts supply chain, we demonstrate the value of employing integrated decision models over using separate repair and inventory allocation rules for a range of environments where inventory imbalances exist. We also show that our heuristic approach is highly effective and that its inventory allocation subroutine, used as a stand-alone tool for making distribution decisions, outperforms a commonly used inventory allocation rule in most circumstances tested.