INFORMS 2021 Program Book

INFORMS Anaheim 2021

MC23

3 - The More Monitoring, the Better Quality? Empirical Evidence from the Generic Drug Industry Anqi Wu, University of Illinois at Urbana-Champaign, Urbana, IL, 61821, United States, Yixin Iris Wang With growing concerns on drug safety issues, FDA has implemented major moves to allocate more inspections to high-risk manufacturers. The actions of FDA implicitly assume that the frequent inspections at low-quality facilities could reduce the drug safety concerns. However, whether more monitoring guarantees better quality remains unclear. In this study, we directly examine the validity of this assumption in the generic drug industry and test the impact of inspection frequency on the manufacturing quality. Overall, we do not find evidence that supports the link between more frequent inspections and fewer recall events, suggesting that the current risk-based inspection model alone might not help bring down the quality failures in drug products. We propose two explanations for the results the organizational culture and the inherent quality levels. MC23 CC Room 204C In Person: Service Economics and Revenue Management General Session Chair: Andrew E. Frazelle, The University of Texas at Dallas, Dallas, TX, 75205-3685, United States Co-Chair: Pnina Feldman, Boston University, Boston, MA, 2215, United States Co-Chair: Ricky Roet-Green, University of Rochester, Pittsford, NY, 14534-2883, United States 1 - Ownership Utility of Rental Products in Rent-to-own Businesses Milad Armaghan, University of Texas at Dallas, Richardson, TX, 75080-3021, United States Rent-to-own (RTO) businesses offer products to renters in exchange for a periodic fee. Renters can also purchase the already-rented product at a dynamically determined buyout price set by the firm. To model renter decisions in this market, we develop a new utility framework that takes full advantage of the unique features of the RTO space, namely the repeated signals about each renter’s ownership utility provided by his acceptance or rejection of different buyout prices. Within our framework, we propose several different specifications of renter ownership utility, and we use maximum likelihood estimation to identify optimal parameter estimates analytically and, where necessary, numerically. In particular, we propose one structure in which utility is independent and identically distributed across periods and renters, and another in which utility is heterogeneous across renters but constant across periods for each renter. We also model the spectrum between these extremes, first by developing an algorithm to attribute different utility structures to different renters, and second by proposing a stochastic process with utilities in successive periods imperfectly correlated. Our methodology and results can be used by RTO firms to estimate renter willingness to pay when setting buyout prices. Furthermore, using transaction data from a prominent RTO firm, we test the estimation performance of the various utility specifications. While some specifications perform better than others on our data, we believe that all of the utility structures that we propose and analyze are of independent interest for estimating renter ownership utility in different settings. 2 - Cherry Picking and Service Segmentation on On-demand Service Platforms Qiaowen Guo, Washington University in St. Louis, St. Louis, MO, United States, Kaitlin M. Daniels, Panos Kouvelis We study an on-demand platform that facilitates a marketplace in which servers serve two types of customers patient and impatient. Like non-platform firms, the platform can construct a menu of prices and expected wait times tailored to each type of customer. Our platform, however, only has indirect control over the wait experienced by customers. Instead, customer wait depends on the decisions of individual servers, who choose for themselves which jobs to serve and may exhibit “cherry picking,” a behavior in which a server strategically declines a low- value job to be available to serve a high-value future job. We study the platform’s optimal incentive design. 3 - Strategic Behavior in Queues with Arrival Rate Uncertainty Binyamin Oz, Hebrew University of Jerusalem, Jerusalem, Israel, Refael Hassin, Moshe Haviv, Moshe Haviv We consider a general queueing model with a Poisson arrival process whose rate is random, and realized once for the entire process. We show that the distribution of the arrival rate at arrival instants is the size-biased counterpart of the original distribution. In particular, the ASTA (arrivals see time averages) property does not hold but rather a rate-biased version of it that we define and coin by the term RASTA (Rate-biased ASTA). We show that the RASTA phenomenon plays a crucial role in the analysis of strategic behavior of customers who evaluate the consequences of the actions they take upon arrival. We study such a system with a single server and strategic customers who decide whether to join or balk without observing the queue.

MC24 CC Room 205A In Person: Emerging Pricing and Sales Promotion Strategies General Session Chair: Rim Hariss, McGill University, Montreal, QC, H3A 1G5, Canada 1 - A Near-optimal Algorithm for Real-time Order Acceptance: An Application in Post-acute Healthcare Services We study a joint capacity investment and real-time accept/reject optimization problem in an infinite horizon with an application in post-acute care. To maximize the average profit per period, the firm accepts/rejects stochastic referral arrivals in real time. Accepted referrals require different resources over an episode. A referral differs in the revenue, the resource requirement, the frequency of resource usage, and the stochastic duration of the episode. Using a simple policy, we derive a worst-case guarantee on its optimality gap, and show that our policy is asymptotically optimal. We also illustrate the impressive numerical performance of our policy using public healthcare data. 2 - Pay-to-win in Video Games: Microtransactions and Fairness Concerns Duc Vu, The University of Texas at Dallas, Richardson, TX, 75080- 3021, United States, Xuying Zhao, Kathryn E. Stecke Microtransaction (selling add-on items in games) is a commonly observed in video games. Although microtransaction generates extra revenue, it also leads to fairness concerns from players who do not buy the add-on. A publisher needs to consider the cross-externalities between game players and add-on buyers, when deciding whether to adopt a microtransaction strategy, or a bundling strategy. We identify two determinant ratios of the optimal strategy: the market size ratio and valuation ratio of high-type (hardcore players) to low-type (casual players) game players. Although microtransaction may create fairness concerns, it actually generates higher consumer surplus than bundling if properly implemented. Our results provide plausible explanations for successes and failures of a microtransaction strategy in different games and guidance for future publishers. 3 - Optimal Promotional Budget Allocation for Customer Retention In Subscription Retailing Rim Hariss, Assistant Professor, McGill University, Montreal, QC, 02139, Canada, Yu Ma Many consumer-facing companies often adopt promotions to attract customer demand. In doing so, they have limited promotions budgets that they need to allocate strategically across different products categories in order to increase sales. Other than sales, subscription-based retailers have an additional revenue stream from memberships or subscriptions fees. To ensure the long-term profitability of these programs, these companies need to develop capabilities to accurately assess the demand and revenue implications of their promotion strategy as well as to how it affects customer churn. In particular, types, volumes and frequency of promotions for different product assortments could affect subscribed customers’willingness to extend their membership, which in turn affect these companies’ total revenue. MC25 CC Room 205B In Person: Circular Economy Strategies and Products’ End of Life Cycle/Socially Responsible and Sustainable Operations General Session Chair: Yen-Ting (Daniel) Lin, University of San Diego, San Diego, CA, 92110, United States 1 - Show, Don’t Tell: Education and Physical Exposure Effects in Remanufactured Product Markets Huseyn Abdulla, Ph.D. Candidate, Mays Business School, Texas A&M University, 301A Wehner Building, TAMU 4127, College Station, TX, 77840, United States, James Duane Abbey, Selin Atalay, Margaret Meloy We empirically examine the effectiveness of managerially-relevant, process- and product-related interventions to increase the appeal of and willingness-to-pay (wtp) for remanufactured consumer products: educating consumers about remanufacturing processes and providing physical exposure to a remanufactured product. We find that education does not cause a significant increase in the appeal of and wtp for toward remanufactured products. However, providing physical exposure to remanufactured products results in a significant increase in both the appeal of and willingness-to-pay for the remanufactured products. We discuss the practical implications of our findings for sellers of remanufactured products and circular economy policy-makers. Zihao Qu, University of Texas at Dallas, Richardson, TX, United States, Milind Dawande, Ganesh Janakiraman

64