2015 Informs Annual Meeting

MB53

INFORMS Philadelphia – 2015

MB55 55-Room 108B, CC Applications of DEA II Cluster: Data Envelopment Analysis Invited Session

2 - An Analytical Framework for Value Co-Production in Services Guillaume Roels, Associate Professor, UCLA, 110 Westwood Plaza, Los Angeles, CA, 90095, United States of America, guillaume.roels@anderson.ucla.edu, Uday Karmarkar Although services are often defined as co-productive of value, the concept of value is often difficult to measure. Yet, measuring value is not necessarily a prerequisite for service process improvement. In this paper, we propose a general framework for the modeling and analysis of services with co-production. The framework identifies three major process stages: (i) the production stage, which involves co-production, (ii) the output sharing stage, and (iii) the consumption stage.

Chair: Alan Pritchard, University of Maryland, Robert H. Smith Scholl of Business, Van Munching Hall, College Park, MD, 20742, United States of America, apritchard@rhsmith.umd.edu 1 - Nurse Staffing Performance Evaluation: Data Envelopment Analysis vs. Expert Assessment Fan Tseng, University of Alabama in Huntsville, Dept of Mgt, Mkt, & IS, Huntsville, AL, 35899, United States of America, tsengf@uah.edu, Karen Frith, Faye Anderson, Patricia Patrician When using Data Envelopment Analysis (DEA) to evaluate the efficiency of nurse staffing, the results are greatly influenced by the selection of input and output metrics. To evaluate different DEA models for their usefulness, we enlisted experts in nurse administration to evaluate the performance of nursing units using data from multiple hospitals. We compare the results between experts and the models, and discuss the issues in DEA modeling for evaluating nurse staffing performance. 2 - It Productivity Paradox: A New Frameworks Integrating Configuration Theory and Dynamic DEA While some research argues that information technology (IT) can improve organizational productivity, others maintains that the impact of IT may be negative. This paper advances a new perspective based on data envelopment analysis (DEA) to investigate the IT productivity paradox. We propose a new theoretical framework based on dynamic two—stage network DEA models, considering multiple periods, multiple inputs and multiple outputs, to study and understand IT productivity paradox. 3 - An Oligopolistic Emissions Trading System with Uncertain Demand Alireza Tajbakhsh, PhD Candidate, DeGroote School of Business, McMaster University, 1280 Main St. W, Hamilton, ON, L8S 4L8, Canada, alirezt@mcmaster.ca, Elkafi Hassini We propose a static Cournot oligopoly game to investigate a perfectly competitive market in which supply chains compete in a non-cooperative manner in their product markets. Partners of each supply chain engage in a cooperative triopoly game where initial permit allocations of the pollutants are given on the basis of their sustainability performance that is derived from a data envelopment analysis model. 4 - Product Variety and Productivity: Evidence from the North American Beverage Industry Alan Pritchard, University of Maryland, Robert H. Smith School of Business, Van Munching Hall, College Park, MD, 20742, United States of America, apritchard@rhsmith.umd.edu, Martin Dresner, Xiang Wan Using data taken directly from a major North American soft drink beverage bottler and distributor, we examine distribution center (DC) productivity. First, we employ data envelopment analysis (DEA) and a double bootstrapping procedure to estimate the relative efficiency of 108 DCs over a four year period (2008-2011). Then, we use a Tobit regression model to investigate the factors that influence DC productivity – that is, unexplained variation in efficiency, over time. Liu Jiawen, PhD, Huazhong University of Science and Technology, 1037 Luoyu Road, Wuhan, 430074, China, jiawen_liu@hust.edu.cn, Yeming Gong

MB53 53-Room 107B, CC Social Media, Sales and Pricing Sponsor: Behavioral Operations Management Sponsored Session

Chair: Wedad Elmaghraby, Associate Professor, University of Maryland, 4311 Van Munching Hall, College Park, MD, 20742, United States of America, welmaghr@rhsmith.umd.edu 1 - Scarcity Strategies under Quasi-Bayesian Social Learning Nitin Bakshi, London Business School, Regent’s Park, London, United Kingdom, nbakshi@london.edu, Yiangos Papanastasiou, Nicos Savva The introduction of popular experiential products is often accompanied by temporary stock outs. This paper proposes a mechanism based on an empirically- motivated behavioural model of social learning. We show that such strategies may be beneficial for the firm and may also increase consumer surplus. 2 - Integrating Social Media Metrics Wendy Moe, University of Maryland, 3469 Van Munching Hall, College Park, MD, United States of America, wendy_moe@rhsmith.umd.edu, David Schweidel The primary goal of this paper is to offer a modeling approach that integrates multiple social media metrics. We do this by jointly modeling the number of mentions, the number of co-mentions and expressed sentiment across brands in a given market as a function of a latent map that represents the underlying competitive landscape in the industry. We demonstrate how a brand can use this model to establish benchmark metrics and calculate a measure of differentiation. 3 - Optimizing Donation Campaigns with Social Media Shawn Mankad, Assistant Prof Of Business Analytics, University of Maryland, 4316 Van Munching Hall, College Park, MD, 21201, United States of America, smankad@cornell.edu, William Rand, Chen Wang The rising popularity of social media has resulted in organizations of all types attempting to use the social streams to inform managerial decisions. However, using social media data can be challenging due to its varied and dynamic nature. In this work, we discuss show donations to a major nonprofit organization can be substantially increased by integrating Twitter usage around crisis events to determine the timing and targeting of marketing communications.

MB54 54-Room 108A, CC Markov Decision Processes

Cluster: Tutorials Invited Session Chair: Andrew J. Schaefer, University of Pittsburgh, 3700 O’Hara Street, Benedum Hall 1048, Pittsburgh, PA, 15261-3048, United States of America, schaefer@ie.pitt.edu 1 - Tutorial: Markov Decision Processes in Healthcare Andrew J. Schaefer, University of Pittsburgh, 3700 O’Hara Street,

MB56 56-Room 109A, CC Recent Advances in Location Analysis Sponsor: Location Analysis Sponsored Session

Chair: Sibel Alumur, University of Waterloo, 200 University Avenue West, Waterloo, ON, N2L 3G1, Canada, sibel.alumur@uwaterloo.ca 1 - Green Charging Station Location Problem Okan Arslan, Bilkent University, Department of Industrial Engineering, Ankara, Turkey, okan.arslan@bilkent.edu.tr, Oya E. Karasan We deal with ‘charging station location problem’ as a variant of ‘flow refueling location problem’ (FRLM) by additionally considering the hybrid vehicles such as PHEVs. The objective is to maximize the environmental benefits through maximizing electricity usage in transportation. To solve the problem, we propose an arc-cover model, and apply Benders decomposition. The structure of this formulation allows us to construct Pareto-optimal cuts without having to solve any linear programming problems.

Benedum Hall 1048, Pittsburgh, PA, 15261-3048, United States of America, schaefer@ie.pitt.edu

The last decade has seen a large number of Markov decision processes (MDPs) applied to various healthcare settings. In this tutorial we review some of the healthcare decisions for which MDPs may be appropriate. We discuss some of the unique challenges that arise in healthcare modeling. Finally, we discuss future directions for MDPs in healthcare.

190