INFORMS 2021 Program Book

INFORMS Anaheim 2021

TC34

4 - Information Design for Congested Social Services: Op timal Need-based Persuasion Jerry Anunrojwong, Columbia Business School, New York, NY, United States, Krishnamurthy Iyer, Vahideh Manshadi We study the effectiveness of information design in reducing congestion in social services catering to users with varied levels of need. We consider a model where an arriving user decides either to wait for the service by joining an unobservable FCFS queue, or to seek her outside option. To reduce congestion, the service provider seeks to persuade more low-need users to avail their outside option, thereby better serving high-need users. We show that with enough heterogeneity in need, information design not only Pareto dominates full-information and no- information mechanisms, in some regimes it also achieves the same welfare outcomes as the “first-best”, i.e., the optimal admission policy. In Person: Economics of Retail Distribution Services/Healthcare and Service Operations Management General Session Chair: Soraya Fatehi, University of Texas at Dallas, Richardson, TX, 75080, United States 1 - Analysis on Costs and Benefits of Machine Learning-based Early Hospitalization Prediction Overcrowding is often derived from boarding time delays of ED patients to wards. If we predict a patient’s hospitalization early enough and accurately in the ED, an inpatient bed for the patient can be prepared in advance. We predict an ED patient’s hospitalization and compare the performances between models. Based on the prediction results, we estimate how much time is saved in an ED. We also estimate time costs of beds to be kept as being empty for patients to be hospitalized in wards. According to the analysis, we provide the linkage between prediction performances, costs, and benefits. 2 - Contingent Free Shipping: Drivers of Bubble Purchases Sahar Hemmati, University of Maryland, College Park, MD, United States, Wedad Elmaghraby, Ashish Kabra, Nitish Jain Retailers often offer free shipping on orders above a pre-specified threshold (Contingent Free Shipping, CFS). In response, customers may pad below- threshold orders to avoid shipping fees. This behavior can economize logistics costs if customers do not engage in bubble purchases, padded orders with above- par return propensity. In this study, we empirically examine how the customers’ engagement in bubble purchases relates to: 1. CFS policy’s threshold and shipping fee and 2. ease of product return. We find that, in response to CFS policies, customers pad 12.4% to 28.4% of below-threshold orders. Both policy levers considerably affect their order padding and bubble purchase propensity. In markets with a customer-friendly return process, share of bubble purchases varies from 8.4% to 14.7% and is altogether eliminated in markets with inconveniences in the return process. 3 - Capacity Flexibility via On-demand Warehousing Soraya Fatehi, University of Texas at Dallas, Richardson, TX, United States, Leela Aarthy Nageswaran, Michael R. Wagner We study the on-demand warehousing business practice where firms who seek capacity for short term needs match with third-party warehouse providers who have excess capacity. We derive a firm’s optimal capacity investment decision in the presence of the on-demand warehousing option. Motivated by the practice of on-demand warehousing players and prior work on sharing platforms, we also investigate the implications of two pricing strategies, surge pricing and bid pricing. Our results indicate that on-demand warehousing allows firms to absorb their demand fluctuations and increase their profit. Our findings provide valuable insight into when the different pricing strategies may be used by an on-demand warehousing platform. We also study whether platforms should share information regarding available capacity of providers and how it impacts the system performance. Eunbi Kim, Korea University, Seoul, Korea, Republic of, Kap-su Han, Taesu Cheong, Sung-Woo Lee, Sujin Kim, Joonyup Eun TC34 CC Room 209B

with potentially unbounded loss for one of the firms. We numerically find that price dynamics resulting from gradient-based algorithms often converge to this undesirable outcome. We finally discuss extensions of our insights to a general model of “pseudo-competitive” pricing games with multiple firms, allowing for a mixture of loyal, satisficing, and opportunistic customers. 2 - How Does Competition Affect Exploration vs. Exploitation? A Tale of Two Recommendation Algorithms Z. Eddie Ning, CKGSB, Beijing, China In this paper, we use a continuous-time multi-agent bandit model to analyze firms that supply content to consumers. We compare a forward-looking recommendation algorithm that balances exploration and exploitation to a myopic algorithm that only maximizes the current quality of the recommendation in both monopoly and duopoly settings. Our analysis shows that competition can discourage learning. In duopoly, firms focus more on exploitation than exploration in their recommendations than a monopoly would. Competition decreases firms’ incentives to develop forward-looking algorithms when users are impatient. Development of the optimal forward-looking algorithm may hurt users under monopoly but benefits users under competition. TC33 CC Room 209A In Person: Operations Research at Facebook/Ride Sharing Operations General Session Chair: Amine Allouah, Columbia University, New York, NY, 10027, United States 1 - The Parity Ray Regularizer for Pacing in Auction Markets Andrea Celli, Bocconi University, Milano, Italy Internet advertising platforms typically offer advertisers the possibility to pace the rate at which their budget is depleted, through budget-pacing mechanisms. We focus on multiplicative pacing mechanisms in an online setting in which a bidder is repeatedly confronted with a series of advertising opportunities. Building on recent work, we study the frequent case in which advertisers seek to reach a certain distribution of impressions over a target population of users. We introduce a novel regularizer to achieve this desideratum, and show how to integrate it into an online mirror descent scheme with optimal order of sub-linear regret when inputs are drawn independently, from an unknown distribution. We demonstrate its effectiveness through numerical experiments on real-world data. 2 - Labor Cost Free-Riding in the Gig Economy Zhen Lian, Cornell University, Cornell Tech, New York, NY, 10044, United States, Sebastien Martin, Garrett J. van Ryzin We propose a theory of gig economies in which workers participate in a shared labor pool utilized by multiple firms. Since firms share the same pool of workers, they face a trade-off in setting pay rates; high pay rates are necessary to maintain a large worker pool and thus reduce the likelihood of lost demand, but they also lower a firm’s profit margin. We prove that larger firms pay more than smaller firms in the resulting pay equilibrium. These diseconomies of scale are strong too; firms smaller than a critical size pay the minimal rate possible (the workers’ reservation wage), while all firms larger than the critical size earn the same total profit regardless of size. This scale disadvantage in labor costs contradicts the conventional wisdom that gig companies enjoy strong network effects and suggests that small firms have significant incentives to join an existing gig economy, implying gig markets are highly contestable. Yet we also show that the formation of a gig economy requires the existence of a large firm, in the sense that an equilibrium without any firms participating only exists when no single firm has enough demand to form a gig economy on its own. The findings are consistent with stylized facts about the evolution of gig markets such as ridesharing. 3 - Matching Drivers to Riders: A Two-stage Robust Approach Oussama Hanguir, Columbia University, New York, NY, 10025, United States, Omar El Housni, Vineet Goyal, Clifford Stein Matching riders to drivers efficiently is a fundamental problem for ridesharing platforms who need to match the riders as soon as the request arrives with partial knowledge about future requests. A myopic approach that computes an optimal matching for current requests ignoring future uncertainty can be highly sub- optimal. In this paper, we consider a two-stage robust optimization framework for this matching problem where future demand uncertainty is modeled using a set of demand scenarios (specified explicitly or implicitly). The goal is to match the current request to drivers (in the first stage) so that the cost of first stage matching and the worst case cost over all scenarios for the second stage matching is minimized. We show that the two-stage robust matching is NP-hard under various cost functions and present constant approximation algorithms for different settings.

121