2016 INFORMS Annual Meeting Program

TC77

INFORMS Nashville – 2016

TC75 Legends C- Omni Behavioral Operations III Contributed Session Chair: Sam Kirshner, UNSW Business School, Level 2, West Wing, Quadrangle Building, UNSW, Sydney, 2052, Australia, s.kirshner@unsw.edu.au 1 - The Price Of Privacy: Experimental Evidence For The Value Of Privacy With Respect To Social Norms Rachel Cummings, California Institute of Technology, 1200 E California Boulevard, MC 305-16, Pasadena, CA, 91125, United States, rachelc@caltech.edu In a series of multi-player experiments, we measure people’s willingness to pay for privacy, and how this value depends on behavioral social norms. Each player is given the option to ``steal’’ monetary payments from the other players; the information shared about these decisions varies across treatments, which includes revealing partial or noisy information information about the player’s actions. By varying the information sharing policy, we can measure how people trade off money for privacy. We also measure how people’s willingness to steal changes as stealing behavior becomes more prevalent. 2 - Behavioral Analysis Of Consumers’ Purchase Timing Decision Ilhan Emre Ertan, PhD Candidate, University of Texas at Dallas, 430 Southwest Pkwy., Apt 2102, College Station, TX, 77840, United States, emre.ertan@gmail.com From a long history of markdown sales, consumers have expectations that retailers will provide markdown discounts during a selling horizon. The consumer purchase timing decision is analyzed by using discounted expected utility theory. The consumer’s sequential decision-making process is formalized under uncertain product availability and several behavioral biases. An optimal purchase timing policy is identified in a market environment, in which a strategic customerknows the markdown pricing scheme, available inventory level, and remaining time to the end of the selling horizon. 3 - A Behavioral Experiment On Sharing Advance Warning Of A Supply Chain Disruption Sourish Sarkar, Assistant Professor, Pennsylvania State University - Erie, 5500 Copper Dr, Apt 104, Erie, PA, 16509, United States, sourishs@gmail.com, Sanjay Kumar Using the beer distribution game in a controlled laboratory setting, we explore the effect of advance warning of a supply chain disruption. Effect of sharing that information with supply chain partners is also investigated. Considering both upstream and downstream disruptions, we summarize the results from our experiment with several scenarios: disruption with no advance warning but information sharing; disruption with no advance warning and no information sharing; disruption with advance warning but no information sharing; disruption with advance warning and sharing of this warning information. 4 - The Behavioral Traps Of Making Multiple, Simultaneous, Newsvendor Decisions Shan Li, Assistant Professor, Baruch College, City University of New York, Baruch College, 55 Lexington Avenue, New York, NY, 10010, United States, shan.li@baruch.cuny.edu, Kay-Yut Chen We conducted an experimental study to explore behaviors of newsvendors who make order decisions for two stores simultaneously. While the two stores are independent, we find that order decisions are impacted not only by the history from the same store, but also by the past information from the other store. 5 - The Impact Of Reference Points On Supply Chain Coordination

TC76 Legends D- Omni Decision Analysis I Contributed Session Chair: Chih-Yang Tsai, Professor, State University of New York at New Paltz, 1 Hawk Drive, New Paltz, NY, 12561-2443, United States, tsaic@newpaltz.edu 1 - General Model For Dynamic Learning Ambiguity vs Bayesianism Mohammad Rasouli, University of Michigan, 430 South Fourth Ave, Ann Arbor, MI, 48104, United States, rasouli@umich.edu We propose a general framework for adaptive learning that can model both Bayesian and non-Bayesian learning. We show how different objectives including expected outcome, minmax, and min regret can be modeled in this framework. This framework gives a unified view to the existing results in learning. We complete the existing results by proposing new sufficient statistics and dynamic programming techniques. The connection with zero-sum games is discussed. We will discuss conditions under which pure strategies can achieve optimal performance. 2 - Remove A Paradox In Data Envelopment Analysis Dariush Khezrimotlagh, Dr., Miami University, Oxford, OH, 45056, United States, khezrid@miamioh.edu Data Envelopment Analysis (DEA) is a non-parametric linear programming tool to assess the relative efficiency of a set of homogenous firms with multiple input factors and multiple output factors. DEA has been used in thousands of published papers and books in well-known qualified journals and by reputable publishers since 1978. DEA assumes that the relationships between the factors can be varied from one firm to another. This article proves that this assumption has a contradiction with the homogeneity of firms and concludes that the provided scores by DEA models are not relatively meaningful and should not be used to rank or benchmark firms. The instructions to remove the paradox in DEA are illustrated. 3 - Reproducing Kernel Hilbert Space Approach To Stochastic Frontier Semiparametric Estimation Carlos Felipe Valencia Arboleda, University of los Andes, Cra 1 Este No 19A - 40, Edificio Mario Laserna, Bogota, 11001000, Colombia, cf.valencia@uniandes.edu.co We develop a nonparametric estimator for the Stochastic Frontier Analysis problem based on the Reproducing Kernel Hilbert Space approach. We prove minimax optimality of convergence for the frontier estimator, and semiparametric efficiency for the finite dimensional parameters. Using Sobolev Hilbert Spaces, we implement the method under monotonicity and concavity constrains. We perform a simulation study to show the benefits of our estimators.

TC77 Legends E- Omni Opt, Integer Programing III Contributed Session

Chair: John Shane Lyons, PhD Student, Western University-Ivey Business School, 23 Pine Ridge Drive, London, ON, N5X 3G7, Canada, jlyons.phd@ivey.ca 1 - Computer-assisted Discovery And Automated Proofs Of Cutting Plane Theorems Yuan Zhou, University of California - Davis, One Shields Avenue, Davis, CA, 95616, United States, yzh@math.ucdavis.edu, Matthias Koeppe Inspired by the breakthroughs of the polyhedral method for combinatorial optimization in the 1980s, generations of researchers have studied the facet structure of convex hulls to develop strong cutting planes. We ask how much of this process can be automated: In particular, can we use algorithms to discover and prove theorems about cutting planes? We focus on general integer and mixed integer programming, and use the framework of cut-generating functions. Using a metaprogramming technique followed by practical computations with semialgebraic cell complexes, we provide computer-based proofs for old and new cutting-plane theorems in Gomory-Johnson’s model of cut generating functions.

Sam Kirshner, UNSW Business School, Level 2, West Wing, Quadrangle Building, UNSW, Sydney, 2052, Australia, s.kirshner@unsw.edu.au, Lusheng Shao

Prospect theory and reference points have recently been utilized to explain the behavioral ordering of human newsvendors. Adopting this approach to modeling newsvendor behavior, we analytically explore the implications of reference points in a two-tier supply chain. We show that reference points enable coordination in a wholesale contract setting, and demonstrate how the reference points alter coordinating contracts under buy-backs and revenue sharing agreements.

329