Informs Annual Meeting 2017

TD28

INFORMS Houston – 2017

TD28

3 - The Effect of a Front-of-pack Nutrition Labeling System to Promote Healthier Purchase Decisions Jeremy C. Young, Professor, Pontificia Universidad Javeriana, Bogota, Colombia, jeremyyoung@javeriana.edu.co, Claudio Mora, Luisa Fernanda Tobar, Pamela Elizabeth Vallejo Our intervention aims to study if food purchasing decisions can be improved by using a simple, informative, front-of-package label. In contrast to previous studies, our design uses controlled randomization that can disentangle whether the label itself or the information on the label is causing behavior change in a real-world setting. Our paradigm utilizes the 5-Color Nutrition Label, which exploits the nutritional information of products to inform consumers about each products’ nutritional contents in an easy-to-understand manner that promotes product comparison. We utilize all sales data and intervene in the 21 cafeterias and shops run by Food Services at Pontificia Universidad Javeriana. 4 - All That Glitters is Not Gold: Extracting Product Information from Online Consumer Reviews This study shows empirically that Online Consumer Reviews (OCR) may be understood as composed of three main elements: information, emotions, and how accessible each OCR was for consumers. This approach allows academic researchers and practitioners to develop a better classification of OCRs regarding their information content -as this article test empirically- and gain a more accurate knowledge of consumers’ opinions in the electronic marketplace. But this approach can benefit consumers as well, since making more helpful information accessible can assist them better in making purchase decisions and overcome information overload, especially for very popular products with hundreds of reviews. 350F Tutorial Erlang Queueing Models Sponsored: Applied Probability Sponsored Session 1 - Tutorial on Erlang – A Queueing Dynamics Jamol Pender, Cornell University, 206 Rhodes Hall, Ithaca, NY, 14850, United States, jamol.pender@gmail.com, William A. Massey Since 2017 is the 100th anniversary for the Erlang-B formula, we consider the Erlang-A queueing model. It is defined to be a Markovian multi-agent, infinite buffer service system that permits customer abandonment. We also assume non- homogenous Poisson arrivals with exponentially distributed service and abandonment times. We use dynamical systems to analyze the transient dynamics of this queue. We obtain dynamic performance metrics for the queueing process and customer delays. Also, these differential equations motivate dynamic optimization methods for new service staffing control algorithms. TD30 Jorge Fresneda, Drexel University, 3220 Market Street, Philadelphia, PA, 19104, United States, jef78@drexel.edu 351A Risk Analysis Contributed Session Chair: Chanjoo Lee, InterContinental Hotels Group, Atlanta, GA, United States, chanjoo.lee@ihg.com 1 - Evolution of Revenue Management System at IHG Chanjoo Lee, Manager Revenue Optimization, InterContinental Hotels Group, 3 Ravinia Drive, Suite 100, Atlanta, GA, 30346, United States, chanjoo.lee@ihg.com IHG’s PERFORM Price Optimization project started in 2005 to drive key strategic priorities such as brand performance enhancement and excellent hotel returns. The project was the first large-scale enterprise implementation of price optimization in the hospitality industry and provided a 2.7% increase in revenue as verified in the IHG 2009 Annual Report. In this talk, we will discuss how the IHG Revenue Management Systems including PERFORM Price Optimization evolved to increase user acceptance from the hotels and drive revenue improvement over the years. TD31

350D Approximation in Auctions

Invited: Auctions Invited Session Chair: Rakesh Vohra, University of Pennsylvania, rvohra@seas.upenn.edu

1 - A Duality Based Framework for Bayesian Mechanism Design Yang Cai, 65 Sherbrooke Est, Apt. 1711, Montreal, QC, H2X1C4, Canada, cai@cs.mcgill.ca, S. Matthew Weinberg, Nikhil Devanur It is well-known by now that the simple mechanisms used in practice are rarely, if ever, optimal. Recent work has aimed to understand this through the lens of approximation and has successfully shown that while virtually never optimal, these simple mechanisms are often approximately optimal. Still, the techniques used to prove these claims were setting-specific and therefore limited their applicability to broad settings. We provide a principled approach to design and analyze these simple mechanisms based on duality theory. Our approach unifies and improves previous results, and greatly extend known settings where simple mechanisms are approximately optimal. 2 - Robust Mechanisms under Common Valuation Songzi Du, Simon Fraser University, Burnaby, BC, Canada, songzid@sfu.ca We study robust mechanisms to sell a common-value good. We assume that the mechanism designer knows the prior distribution of the buyers’ common value but is unsure of the buyers’ information structure about the common value. We use linear programming duality to derive mechanisms that guarantee a good expected revenue for all information structures and all equilibria. Our mechanism maximizes the revenue guarantee when there is one buyer. As the number of buyers tends to infinity, the revenue guarantee of our mechanism converges to the full surplus. We numerically demonstrate that the revenue guarantees of our mechanisms are generally close to optimal when there are two buyers. 3 - Informationally Robust Optimal Auction Design Benjamin Aaron Brooks, University of Chicago, Becker Friedman Economics, 1126 E.59th Street, Chicago, IL, 60637, United States, ppivoriu@uchicago.edu A single unit of a good is to be sold by auction to one of two buyers. The good has either a high value or a low value, with known prior probabilities. The designer of the auction knows the prior over values but is uncertain about the correct model of the buyers’ beliefs. The designer evaluates a given auction design by the lowest expected revenue that would be generated across all models of buyers’ information that are consistent with the common prior and across all Bayesian equilibria. An optimal auction for such a seller is constructed, as is a worst-case model of buyers’ information. The theory generates upper bounds on the seller’s optimal payoff for general manyplayer and common-value models. 350E Manufacturing Contributed Session Chair: Jorge Fresneda, Drexel University, Philadelphia, PA, United States, jef78@drexel.edu 1 - Online Reputation Analysis by Social Network Jing Luo, University of Pittsburgh, 5712 Beacon Street, Pittsburgh, PA, 15217, United States, JIL204@pitt.edu The goal of this study is to find a way to detect the real reputation of the online business roles. We will use network science methods to find the real reputation of the supplier and customer by their previous purchase history. From the network, we will see the change of strong links and weak links by each group. Then we will study the change of network structure to find the reputation of the roles. We could find how the previous data could work for the future purchase. Then this network result could help to set the current transaction cost. 2 - Predicting Fashion Branding Effects using User Generated Brand Network Social media increases the visibility of young designer fashion brands but makes the competition more fierce due to the proliferation of social media channels. However, the branding effects of fashion brands on social media are challenging to capture due to the domination of major national brands’ marketing campaigns. In order to understand the branding effects among various types of fashion brands, this study examines and predicts the branding effects of fashion brands on a user generated brand network by using network analysis. Our findings demonstrate that both network information and brand preferences can improve our understanding of the choice of posting outfits with different brands. Tawei Wang, DePaul University, 1 East Jackson Blvd, DePaul Center 6028, Chicago, IL, 60604, United States, wang131@gmail.com, Yusan Lin TD29

386