Informs Annual Meeting 2017

WC04

INFORMS Houston – 2017

WC04

2 - Pivotal Estimation and Confidence Bands for High-dimensional Linear Models with Error-in-variables Alexandre Belloni, Duke University, Fuqua School of Business,

320A Behavioral Research on Sustainability and Environmental Operations Sponsored: Manufacturing & Service Oper Mgmt, Sustainable Operations Sponsored Session Chair: Luyi Gui, The Paul Merage School of Business, UC Irvine, Irvine, CA, 92697-3125, United States, luyig@uci.edu Co-Chair: Mahdi Mahmoudzadeh, Scheller College of Business, Georgia Institute of Technology, Atlanta, GA, United States, mahdi.mzh@scheller.gatech.edu 1 - Framing Effects and the Non-equivalence of Trade-ins and Upgrades: Theory and Evidence Mahdi Mahmoudzadeh, Georgia Institute of Technology, Scheller College of Business, 800 West Peachtree Street NW, Atlanta, 30308, Georgia, mmahmoudzadeh3@gatech.edu Manufacturers of durable goods often buy back used products to return them back into remanufacturing systems, and to induce customers to switch to new versions. Trade-ins and upgrades are mechanisms that tie these together through replacement purchases. We provide experimental evidence that, in contrast to the classical theory’s view, these mechanisms are not equivalent, and that the framing effect induces consumers to change which prices they anchor to, as their reference points, in arriving at their willingness-to-accept for their used products. We then show how the existence of reference dependence among consumers overturns key predictions of the classical model of trade-ins and upgrades. 2 - The Impact of Environmental Certification on Customer Booking Behavior in Hospitality Matthew Walsman, Rutgers Business School, 28 Cambridge Dr, Berkeley Heights, NJ, 07922, United States, mwalsman@business.rutgers.edu There is some evidence that suggests that LEED certification boosts financial performance in the hospitality industry. The reason for this superior performance however, remains unclear. In this study we attempt to answer this question by investigating the behavioral decisions customers make when booking lodging and whether environmental certification influences customer booking decisions. 3 - Relative Performance Transparency: Effects on Sustainable Purchase and Consumption Behavior Yanchong Zheng, Massachusetts Institute of Technology, Cambridge, MA, United States, yanchong@mit.edu, Ryan Buell, Shwetha P. Mariadassou We study how transparency into the levels and changes of relative sustainability performance affects consumer behavior. In a series of online consumer choice experiments, we find that in the product purchase domain, transparency into a firm’s relative performance levels has a stronger impact on purchase behavior than transparency into the firm’s relative changes in performance. In the energy consumption domain, transparency into the consumer’s relative changes in performance is more dominant in motivating energy conservation than transparency into the consumer’s relative performance levels. We use structural equation models to identify the mechanisms underlying these results. 4 - How Does Precision Affect the Adoption of Energy Efficiency Practices? - Evidence From the Field and Laboratory Data Suresh Muthulingam, Assistant Professor, The Pennsylvania State University, Smeal College of Business, 460 Business Building, University Park, PA, 16803, United States, suresh@psu.edu, Saurabh Bansal We investigate whether precision affects the adoption or non-adoption of energy efficiency practices. An econometric analysis of data on energy efficiency recommendations made by the DOE shows that recommendations with precise costs and savings exhibit higher adoption rates than other recommendations. We use three laboratory experiments to isolate the precision effects and to identify the mechanism by which precision affects the adoption decisions. We find that credibility is an important moderator of the precision effect and that the impact of precision is more pronounced with budgetary constraints.

100 Fuqua Drive, Durham, NC, 27708, United States, abn5@duke.edu, Victor Chernozhukov, Abhishek Kaul, Mathieu Rosenbaum, Alexandre B. Tsybakov

We study high-dimensional linear models with error-in-variables. In addition to the high dimensionality, these models are challenging because of the need to account for measurement errors to avoid non-vanishing biases and its impact on regularization parameters. We propose a new estimator that is pivotal despite of possible heteroskedastic errors. It is defined as the solution of a convex program with second order cone constraints which allows the use of efficient algorithms. We also construct simultaneous confidence regions for the parameters in such models. We show its validity despite of model selection mistakes, and allowing for the number of parameters to be larger than the sample size. 3 - Matrix Completion Has No Spurious Local Minima Jason Lee, University of Southern California, Los Angeles, CA, United States, jasonlee@marshall.usc.edu Matrix completion is a fundamental machine learning problem with wide applications in collaborative filtering and recommender systems. Typically, matrix completion are solved by non-convex optimization procedures, which are empirically extremely successful. We prove that the symmetric matrix completion problem has no spurious local minima, meaning all local minima are also global. Thus the matrix completion objective has only saddlepoints an global minima. This is joint work with Rong Ge and Tengyu Ma. WC03C Grand Ballroom C Consumer Operations Sponsored: Manufacturing & Service Oper Mgmt Sponsored Session Chair: Vinayak Deshpande, University of North Carolina, Kenan Flagler Business School, Chapel Hill, NC, 27599, United States, Vinayak_Deshpande@kenan-flagler.unc.edu 1 - Price Competition in Differentiated Products Markets: When Do Structural Estimation Frameworks Work or Fail? Chungseung Lee, The University of Texas at Dallas, Richardson, TX, United States, cxl143430@utdallas.edu, Metin Cakanyildirim We address a conjecture underlying structural estimation frameworks that account for price endogeneity to analyze differentiated products markets. The conjecture claims that the model has a pure strategy Nash equilibrium(NE) and that observed prices constitute the NE. For demand models with a mixture of multinomial logits, we present conditions under which the NE-assumption holds and NE is unique. The price-sensitivity parameters of the demand model differ across products and market segments. Thus, our existence conditions are widely applicable, and yet are milder than those in existing studies. We conclude by discussing the implications of our results for the structural estimation methods. 2 - Exploiting Complementarities in an Online Advertising Supply Chain Changseung Yoo, The University of Texas at Austin, 2110 Speedway Stop B6500, CBA 5.202, Austin, TX, 78712, United States, csyoo@utexas.edu, Anitesh Barua, Genaro J. Gutierrez We examine channel structures and pricing models in an online advertising supply chain using a proprietary dataset. We develop analytic as well as structural econometric models that enable us to quantify synergy effects between them. While the extant literature emphasizes choosing between pricing models, we show that using multiple models in concert yields higher overall profitability due to strategic complementarities among the pricing schemes. We then explore the operating implications of the complementarities and their impact on profits and supply chain efficiency, and devise information/profit sharing contracts that boost the supply chain profit towards the benchmark scenario. 3 - Don’t Call Us, We’ll Call You: An Empirical Study of a Call Center that Offers Automated Callbacks Vinayak Deshpande, University of North Carolina, Chapel Hill, NC, United States, Vinayak_Deshpande@kenan-flagler.unc.edu, Brett Hathaway, Seyed Emadi We empirically study the behaviors of callers in a banking call center that offers callers the opportunity to wait offline and receive an automated callback if current waiting times are sufficiently long. Using a structural estimation approach, we model callers’ callback decisions. We use the recovered parameters of our model to run a series of counterfactual routing and staffing experiments, which provide managers guidance in determining when to offer automated callbacks.

505