Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

SC19

n SC19 North Bldg 128B Joint Session RMP/Practice Curated: Empirical Research in Revenue Management Sponsored: Revenue Management & Pricing Sponsored Session Chair: Dan Zhang, University of Colorado, Leeds School of Business, University of Colorado, Boulder, CO Co-Chair: Xiao Huang, Concordia University, Montreal, Quebec 1 - How Does Price Volatility Influence Demand of Revenue Managed Goods? Benny Mantin, University of Luxembourg, Luxembourg, L-1511, Luxembourg, Chiara Morlotti We estimate the impact of price fluctuations on demand in the context of revenue-managed goods, first by introducing a new measure of price volatility and second by quantifying the effect on demand elasticity. Using airfare and sales data collected for 21 European route-pairs over five months, we carry out our empirical analyses (IV 2SLS regressions) revealing that as price volatility increases, consumers tend to buy less. We demonstrate how this can be integrated into the classical revenue management model (Expected Marginal Seat Revenue) to yield higher revenues. 2 - Product Quality, Service and Pricing Ruxian Wang, Johns Hopkins University, Carey Business School, 100 International Dr, Baltimore, MD, 21202, United States, Shiliang Cui We develop a unified framework to investigate firms’ joint decisions on product quality, price and ancillary service, when they offer a line of products to consumers in a variety of monopolistic and competitive environments. 3 - Psychological Overage and Underage Costs in Three-part Tariff Plans: Evidence from Bike-sharing Economy Necati Tereyagoglu, Georgia Institute of Technology, Scheller College of Business, 800 West Peachtree Street NW, Atlanta, GA, 30308, United States, Brian S. Park, Eunhee Sohn Three-part tariff plans are often the pricing choice for service providers. These plans involve a fixed subscription fee, which provides an allowance of free units, and a penalty fee for each unit above the allowance. We hypothesize that consumers’ usage is determined not only by a psychological cost for passing the allowance (overage), but also by a psychological cost from ending the usage earlier than what is allowed after the paid penalty (underage). We test both costs using individual-level riding data from a bike-sharing company in New York City, and show that riding decisions are driven by the two psychological costs. We also show that the revenue effects of the psychological costs are significant. 4 - Price to Gain or Price to Retain? An Empirical Study of Hotel Pricing and Customer Cancellation Behavior Dan Zhang, University of Colorado, Leeds School of Business, University of Colorado, Boulder, CO, 80309, United States, Xiao Huang, Jian Wang We use hotel transaction data to investigate the customer cancellation behavior. We show that cancellation rates are highly correlated with booking rates and the posted rates at the time of cancellation. We discuss the implications of this finding on hotel overbooking. Our resultscomplement the empirical research on strategic customer behavior in revenue management and highlight the importance of accounting for the behavioral impact of pricing. 5 - Payment Modes and Order Cancellation: Empirical Evidence from an Online Travel Agent in China Huan Zheng, Shanghai Jiao Tong University, Management Science Department,, 535 Fa Hua Zhen Rd, Shanghai, China, Zilin Hao, Junxiong Yin Dynamic pricing is a widely adopted tool for Online Traveling Agencies. Consumers strategically respond to price fluctuations in various ways to maximize their utilities. In this study, we empirically explore how consumers choose different payment modes when booking hotel rooms, and strategically cancel their bookings from transaction data of a large Online Traveling Agency in China. Based on our two-period econometric model, we find that strategic consumers will keep monitoring price fluctuations and then decide whether they should cancel existing orders.

n SC20 North Bldg 129A Design and Operations of Marketplaces Sponsored: Revenue Management & Pricing

Sponsored Session Chair: Yash Kanoria Co-Chair: Abhinav Sinha, Columbia Business School, 3022 Broadway, New York, NY, 10027, United States Co-Chair: Fanyin Zheng, Columbia University, New York, NY, 10027, United States 1 - Engineering a Separating Equilibrium John Joseph Horton, New York University-Stern, New York, NY, 10012, United States, Ramesh Johari This paper explores whether platform-created signaling opportunities can move designed markets to more desirable equilibria. In a large online labor market, buyers were given the opportunity to signal their relative preferences over price and quality. The intervention caused substantial sorting by sellers to buyers of the right “type.” However, sellers clearly tailored their bids to the type of buyer they faced, bidding up against buyers with a high revealed willingness to pay. Despite this “markup,” a separating equilibrium was sustained over time, suggesting buyers found revelation incentive compatible. We find evidence that informative signaling improved both matching efficiency and match quality. 2 - Dog Eat Dog: Measuring Returns to Scale using a Digital Platform Merger We study returns to scale in digital marketplaces by using proprietary data from the merger of the two largest peer-to-peer platforms for pet sitting services. The merger offers a unique opportunity to study this question because it is one of the few cases where platform size changes in a sudden and exogenous way, not caused by incremental adoption. In addition, the local nature of the services exchanged provide us with multiple city-level markets that are differentially affected by the merger. We quantify changes in the aggregate number of matches and their quality after the platform merger, and how efficiency gains are differentially distributed across platform users. 3 - Optimal Growth in Two-sided Markets Garrett Van Ryzin, Cornell University, Ithaca, NY, United States, Zhen Lian We consider the problem of optimal growth in a two-sided market with both increasing and decreasing returns to scale. We show that in both cases faster growth is better and optimal policies involve large “subsidy shocks”. Our results also show the optimal balance between supply and demand as markets grow. 4 - Match Efficiency in Two-sided Platforms Abhinav Sinha, Columbia Business School, New York, NY, United States, Yash Kanoria, Fanyin Zheng We study the design of two-sided online marketplaces using recent transaction data from a large online platform. We are interested in estimating (i) the unobservable demand and (ii) the dependence of aggregate number of matches on macro-level market characteristics such as the total number of sellers and buyers in the market. Further, our aim is to run counterfactual analysis on platform design and recommend the best ways to improve matching efficiency and/or net profit. We introduce a matching function approach to capture search frictions in the market. n SC21 North Bldg 129B Machine Learning in Demand Modeling Sponsored: Revenue Management & Pricing Sponsored Session Chair: Andrew Vakhutinsky, Oracle 1 - Using Monte Carlo Simulations to Balance Supply and Demand in an On-demand Grocery Delivery Marketplace Jagannath Putrevu, Instacart, 50 Beale St, 11th Floor, Instacart, San Francisco, CA, 94105, United States Instacart uses a network of personal shoppers and drivers to shop for groceries and deliver them to our customers. For Instacart to continue growing rapidly, we want to capture as much demand as possible while keeping our costs low and our personal shoppers and drivers engaged on the marketplace. But this is easier said than done. On any given day, there are multiple factors that contribute to a large amount of variance, which makes this an extremely hard problem. This talk explores a simulation-based optimization methodology that we employed to Andrey Fradkin, Massachusetts Institute of Technology, 179 Prospect Street, Cambridge, MA, 02139, United States, Chiara Farronato

68