Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

SA17

2 - Inducing Fresh Food Spending Through Optimal Incentives Elisabeth Paulson, MIT, Retsef Levi, Georgia Perakis We consider a two-stage scenario where a planner first allocates a fixed budget across possible interventions, and the consumer then decides how to allocate his food budget between fresh food and other options. The goal of the planner is to maximize the consumer’s spending on fresh food. We formulate this game as a two-stage optimization problem with equilibrium constraints, and consider both single-period and stationary multi-stage versions. We present structural results on the planner’s optimal budget allocation, validate our results on a nationally representative data set, and compare these results to the current status quo of federal spending. 3 - Pricing For Heterogeneous Products: Analytics for Tickets Reselling Rim Hariss, Massachusetts Institute of Technology, Operations Research Center, 1 Amherst Street, Cambridge, MA, 02142, United States, Max R. Biggs, Charles Herrmann, Michael Li, Georgia Perakis In collaboration with a major secondary ticket exchange, we study a trading strategy for buying and reselling tickets for events. We look at classifying whether each individual ticket will sell at a given price, in the presence of confounding factors. We construct an optimization based ticket trading strategy that is the reseller is piloting as well as a novel method for heterogeneous treatment effect estimation for classification. On synthetic datasets this approach beats state-of- the-art machine learning approaches for estimating treatment effects. We show using our approach to trade on NBA ticket listings yields a seller’s revenue up to $9.6 million (27% return) for a single season. 4 - Performance Guarantees for Revenue Maximization in Online Type Matching Elaheh Fata, Massachusetts Institute of Technology, 125 Massachusetts Avenue, Aeronautics and Astronautics, Cambridge, MA, 02139, United States, Will Ma, David Simchi-Levi In today’s online marketplaces, the problem of dynamically matching agents in real-time is becoming increasingly prevalent. It captures important tradeoffs faced by the emerging sharing economy, as well as classical inventory and production decisions faced by the transportation industries. The dynamic matching of agents is an inherently challenging task due to the uncertainty in the future agents’ arrival. We consider this problem from the perspective of a central platform who, through historical data, has established the types of agents that could potentially arrive and the relationships between them. The platform’s goal is to maximize the total reward earned by matching agents over time. Platform Ecosystem Management Sponsored: Revenue Management & Pricing Sponsored Session Chair: Hemant K. Bhargava, University of California, Davis, CA, 95616, United States Co-Chair: Ramnik Arora, Facebook, Menlo Park, CA, 9, United States 1 - Salesforce Incentives for Selling Platforms Hemant K. Bhargava, University of California, Gh-3108, Graduate School of Management, Davis, CA, 95616, United States, Olivier J. Rubel We present a model of saleforce incentives for selling platforms, considering both direct and cross-market network effects, which distort the traditional tensions inherent in designing salesforce compensation plans. 2 - Impact of Platform Venture Capital Investments on the Entry and Exit of Complementary Applications in Ecosystems Arvind Karunakaran, McGill University, Montreal, QC, H3H 2P2, Canada, Joey Van Angeren We examine how platform venture capital (PVC) investments affect the entry and exit of complementary applications. We hypothesize that complementors are more likely to introduce and less likely to retract apps in their market niche following a PVC investment. We test this using data collected from a B2B platform. 3 - Economic Analysis of Peer-to-peer Insurance Hong Guo, University of Notre Dame, 356 Mendoza College Of Business, University of Notre Dame, Notre Dame, IN, 46556-5646, United States, Yu-Chen Yang, Yong Jin Peer-to-peer (P2P) insurance is an emerging form of insurance contract enabled by InsurTech advancements. Building upon classic principal-agent model with moral hazard, this paper studies the key characteristics of P2P insurance as a form of insurance contract and compares it to traditional insurance contracts. We identify conditions under which P2P insurance alleviates moral hazard and evaluate its efficiency. n SA19 North Bldg 128B

n SA17 North Bldg 127C Additive Manufacturing and Innovation in a Supply Chain Sponsored: Manufacturing & Service Oper Mgmt/Supply Chain Sponsored Session Chair: Jayashankar M. Swaminathan, University of North Carolina- Chapel Hill, Chapel Hill, NC, 27599-3490, United States Co-Chair: Ali Kemal Parlakturk, University of North Carolina Kenan-Flagler, Chapel Hill, NC, 27599-3490, United States 1 - Retailing with 3D Printing Yao Cui, Cornell University, 401N Sage Hall, Ithaca, NY, 14853, United States, Li Chen, Hau Leung Lee In this paper, we study the impact of adopting 3D printing in a dual-channel (i.e., online and in-store) retail setting, on a firm’s product offering and pricing decisions for the two channels, as well as supply chain operations. 2 - Decentralized Customization with 3D Printing: Drivers of Retail Level 3D Printing Nagarajan Sethuraman, UNC Chapel Hill, Chapel Hill, NC, 27514, United States, Ali Kemal Parlakturk, Jayashankar M. Swaminathan In this paper, we study the trade-offs involved in decentralized customization enabled by 3D printing at retail stores. We develop an analytical model that considers in-store 3D printing as a component of the firm’s broader product line strategy. Managerial insights from our novel model can guide the adoption of 3D printing at retail stores. 3 - Product Management and Innovation in a Supply Chain Hyoduk Shin, Junghee Lee, Vish Krishnan In a technology-intensive supply chains, intellectual property invented by an upstream firm must be embedded in a manufactured subsystem which is then integrated into a full system sold to end consumers. Upstream technology providers face key business model decisions about how to monetize their innovation, which is the focus on this study. They typically employ a royalty- driven business model, but the royalty-based approach has gotten complicated in industrial multi-lateral supply chains. We consider these business model decisions in the context of the industry structure of the subsystem and full system markets. 4 - Printing Spare Parts at Remote Locations. Fulfilling the Promise of Additive Manufacturing Rob Basten, Associate Professor, Eindhoven University of Technology, Eindhoven, Netherlands, Bram Westerweel, Geert-Jan Van Houtum We investigate the benefits of on-site printing of spare parts at remote geographical locations. We consider a periodic-review inventory control system with two supply sources. Regular replenishments occur according to a fixed interval. In between replenishments, spare parts can be printed that possess a lower reliability than regular parts. We characterize the optimal inventory control policy and conduct experiments to investigate the benefits of on-site printing. Using case studies with the Royal Netherlands Army we show that simple AM technology can create significant benefits. Our results show that remote locations are a prime candidate for implementing AM technology to print spare parts. n SA18 North Bldg 128A New Problems in Supply Chain Sponsored: Manufacturing & Service Oper Mgmt/Supply Chain Sponsored Session Chair: Georgia Perakis, Massachusetts Institute of Technology, Cambridge, MA 1 - Value of Failure Prediction Models for Efficient Network Inspection Operations Mathieu Dahan, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Room 1-253c, Cambridge, MA, 02139, United States, Steven B. Link, Saurabh Amin, Georgia Perakis Can infrastructure utilities significantly improve their network inspection operations using predictive failure models? To address this question, we formulate a multi-stage stochastic problem for scheduling regions to visit and inspecting each region. The objective is to maximize failure localization performance under timing, sensing, and routing constraints. We apply techniques for solving submodular function optimization under matroid constraints, and evaluate the solution quality under offline prediction, dynamic customer calls, and strategic probing. We estimate the value of diagnostic information in these cases using data from SF Bay area’s gas pipeline inspection operations.

8