Informs Annual Meeting 2017

TA23

INFORMS Houston – 2017

3 - A Variational Equilibrium Network Framework for Humanitarian Organizations in Disaster Relief: Effective Product Delivery under Competition for Financial Funds

3 - Earning and Learning with Varying Cost Ying Zhong, City University of Hong Kong, Kowloon Tong, Hong Kong, yzhong4-c@my.cityu.edu.hk, Jeff Hong, Guangwu Liu The earning-and-learning problem refers to the profit maximization problem where one optimizes the profit by choosing a price and learns the demand at the same time. Traditionally, the cost is fixed and the problem may be formulated as a multi armed bandit problem, which is known to have an O(logT) lower bound on the expected regret, where T is the number of periods. In this paper, we consider the case where the cost changes over periods. Then, the optimal pricing decision becomes a function of the cost. We propose an upper-confidence-bound type of algorithm to solve the problem. We find that, surprisingly, the expected regret may be bounded by a constant under certain conditions. 4 - Forecasting and Price Optimization for a Beverage Company: Theoretical Questions Arising from Practice Will Ma, Massachusetts Institute of Technology, 71 School Street, Cambridge, MA, 02139, United States, willma353@gmail.com, Hanwei Li, David Simchi-Levi, Jinglong Zhao We discuss our experience conducting demand forecasting and price optimization at a global beverage company, which controls its pricing through multiple retailers. We discuss many different substitution patterns observed in the data: internally within the company’s products, externally with competitors’ products, over time, and between different retailers. The business contstraints on pricing also give rise to various unsolved theoretical questions. 342E Revenue Management for Marketing and Service Systems Sponsored: Revenue Management & Pricing Sponsored Session Chair: John G Turner, University of California-Irvine, Irvine, CA, 92697-3125, United States, john.turner@uci.edu 1 - Optimal Signaling Mechanisms in Unobservable Queues with Strategic Customers David Lingenbrink, Cornell University, Ithaca, NY, United States, dal299@cornell.edu, Krishnamurthy Iyer We consider an unobservable M/M/1 queue offering a service at a fixed price to a Poisson arrival of delay-sensitive customers. The service provider can observe the queue, and may share information about the queue with arriving customers. The customers incorporate this information into their prior beliefs before making the decision to join or balk. We pose the following question: which signaling mechanism and what price should the service provider select to maximize her revenue? We formulate this problem as an instance of dynamic Bayesian persuasion, characterize the structure of the optimal signaling mechanism, and compare its revenue with that of the optimal state-dependent pricing mechanism. 2 - Price Discrimination over Quality via Aggregate Rebate Programs Amir Ajorlou, Massachusetts Institute of Technology, 32 Vassar Firms selling network goods can enjoy a marginal gain in profit due to the network effect. Early adopters improve the social value of the product and firm can raise the price when the product becomes more popular. Motivated by this observation, we pose the question as to whether it is beneficial to a firm selling a product with no network effect to artificially inject externality? To this end, we propose an aggregate reward program where the reward paid to the consumers is a function of the sales volume. By analyzing the equilibria of the induced global game among consumers, we show that the optimal reward program pays back the buyers a fixed rebate, if the product fails to reach certain level of popularity. 3 - Waterfall Yield Optimization for Online Advertising Dmitri Arkhipov, University of California, 23 Bethany Drive, Irvine, CA, 92603, United States, darkhipo@ics.uci.edu, John G. Turner, Michael Dillencourt, Ameila C. Regan In programmatic advertising, advertisements are matched to users. Quality matches may take time to compute, time being a scarce resource particularly for publishers who would like their webpages to load quickly. Policies for ad placement must execute very quickly; this necessitates policies are either simple or are pre-computed offline. The order in which advertisers are offered ad spaces as these become available is referred to as a waterfall and the process of finding a good or optimal such ordering is referred to as yield optimization. We formulate a stochastic optimization problem to solve a publisher’s waterfall yield optimization problem. TA23 Street, 32-D569, Cambridge, MA, 02139, United States, ajorlou@mit.edu, Ali Jadbabaie, David Simchi-Levi

Anna B. Nagurney, University of Massachusetts Amherst, Isenberg School of Management, Dept of Operations & Information Mgmt, Amherst, MA, 01003, United States, nagurney@isenberg.umass.edu, Patrizia Daniele, Emilio Alvarez Flores, Valeria Caruso

In this paper, we present a new Generalized Nash Equilibrium (GNE) model for post-disaster humanitarian relief that captures competition and with novel financial funding and altruism functions. The common shared constraints associated with the relief item deliveries at points of need are imposed by an upper level humanitarian organization or regulatory body and consist of lower and upper bounds to ensure the effective delivery of the estimated volumes of supplies to the victims of the disaster. We formulate the variational equilibrium and illustrate the game theory model through a case study focused on tornadoes hitting western Massachusetts, a highly unusual event that occurred in 2011. 4 - Practice-based Heuristics for Humanitarian Transportation Planning Erica L. Gralla, George Washington University, 800 22nd st NW, Rm 2680, Washington, DC, 20052, United States, egralla@gwu.edu Transportation is a critical challenge in humanitarian response, but adoption of optimization-based decision support has been limited due to the difficulty of implementation. We evaluate the performance of heuristics based on current planning practices, identify key weaknesses, and suggest ways to improve planning while maintaining implementability. 5 - Supply Chain Modeling of Large Scale Humanitarian Food Aid Operations with Different Transfer Modalities Keziban Rukiye Tasci, Northeastern University, 77 Westland Avenue #5, Boston, MA, 02115, United States, tasci.k@husky.neu.edu, Ozlem Ergun Large humanitarian organizations strive to serve millions of beneficiaries spread around the globe by delivering many different commodities and services on time while abiding by organizational and governmental regulations. This study describes a time expanded network model that determines the cost-effective ways to supply pre-decided basket of commodities via using traditional food delivery and cash & voucher modalities for one of the largest humanitarian aid agencies in the world. Through these models and analysis of historic procurement data we identify significant improvement opportunities for the organization’s supply chain operations. 342D Applications of Revenue Management Sponsored: Revenue Management & Pricing Sponsored Session Chair: Maxime Cohen, maxccohen@gmail.com 1 - Price and Availability Discrimination, Exclusive Goods and Social Networks Ruslan Momot, INSEAD, Boulevard de Constance, Fontainebleau, 77305, France, ruslan.momot@insead.edu, Elena Belavina, Karan Girotra We study the value of price and availability discrimination when selling goods whose value derives from exclusive ownership in a social network. We build a stylized model of socially interacting customers who differ in their network position (degree) and their proclivity for social comparisons (conspicuity). We find that firm’s optimal strategy is to set price proportional to product of customer’s degree and conspicuity and exclusively offer the product to only high conspicuity customers with sufficiently low degree. We find that price and availability discrimination are substitutes. While availability is useful instrument on its own, it has limited values when price discrimination is possible 2 - Pricing and Assortment Strategies with Product Exchanges Laura Wagner, IESE Business School, Calabria, Barcelona, Spain, LWagner@iese.edu, Victor Martinez de Albeniz Return policies enable consumers to return or exchange products they are unsatisfied with. We develop a search framework where consumers sequentially learn about products’ true value and decide whether to keep, exchange or return them. We show that when pricing is not a decision, the assortment problem is NP-complete. When pricing is endogenous, the optimization becomes tractable and it is not necessarily optimal to set all products at equal margins. Finally, we find that retailers prefer to pass all return costs on to the consumers, and that this decision can also be beneficial to consumers. TA22

277