2016 INFORMS Annual Meeting Program

SA31

INFORMS Nashville – 2016

SA29 202A-MCC Dynamic Mechanism Design Sponsored: Manufacturing & Service Oper Mgmt, Sustainable Operations Sponsored Session Chair: Peng Sun, Duke University, Durham, NC, United States, peng.sun@duke.edu 1 - Dynamic Mechanism Design With Budget Constrained Buyers Under Limited Commitment Omar Besbes, Columbia University, ob2105@columbia.edu, Santiago Balseiro, Gabriel Weintraub We study the dynamic mechanism design problem of a firm repeatedly selling items to budget-constrained buyers when the seller has limited commitment power. We argue that this problem is generally intractable. Thus motivated we introduce a fluid model that allows for a tractable characterization of the optimal mechanism. We leverage our characterization to provide insights into the dynamic structure of the optimal mechanism and show that the proposed mechanism is a good approximation in large markets. 2 - Dynamic Short-term Contracts Under Hidden Inventory And Backlog Hao Zhang, University of British Columbia, Vancouver, BC, We study a supply chain consisting of a supplier and a retailer faced with random demand over multiple periods. At the beginning of each period, the supplier offers a one-period contract and the retailer chooses order quantity before the demand is realized. The retailer carries leftover inventory or backlogs unmet demand, which is unobservable by the supplier. We find that in the infinite- horizon setting with exponentially distributed demand, for a large parameter set, the optimal sequence of short-term contracts is a generalized base-stock policy, where the base-stock level weakly increases with the beginning inventory. 3 - Dynamic Mechanism Design Without Money Huseyin Gurkan, Duke University, Duke University, Durham, NC, 27707, United States, huseyin.gurkan@duke.edu, Santiago Balseiro, Peng Sun We consider a principal repeatedly allocating a single resource in each period to one of multiple agents without relying on monetary payments over an infinite horizon. Agents’ private valuations are independent and identically distributed. We show that as the discount factor approaches 1, the optimal dynamic mechanism without money achieves the first-best allocation (the welfare maximizing allocation when valuations are public). As part of the proof, we provide an incentive compatible dynamic mechanism that asymptotically achieves the first-best. 4 - Optimal Contract To Induce Continued Effort Peng Sun, Duke University, Durham, NC, psun@duke.edu, Feng Tian We consider a principal incentivizing an agent to exert effort in order to raise the arrival rate of a Poisson process. The effort is costly to the agent, unobservable to the principal, and affects the instantaneous arrival rate. Each arrival yields a constant revenue to the principal. A contract involves payments and a potential stopping time in order to motivate the agent to always exert effort. Although payments can take general forms contingent upon past arrival times, the optimal contract has a simple and intuitive structure, which depends on whether the agent is less patient than the principal. SA30 202B-MCC Revenue Management: Algorithms and Applications Sponsored: Manufacturing & Service Oper Mgmt Sponsored Session Chair: Retsef Levi, MIT, Cambridge, MA, United States, retsef@mit.edu 1 - Assortment Optimization Under A Mallows Distribution Over Permutations Antoine Desir, Columbia University, 601 W 113th Street, Apt 3J, New York, NY, 10025, United States, ad2918@columbia.edu, Vineet Goyal, Srikanth Jagabathula, Danny Segev We study assortment optimization under Mallows distribution over permutations model that is specified by a central permutation and a decay parameter. The probability of any permutation decays exponentially in the (Kendall-Tau) distance from the central permutation. We present an efficient procedure to compute exact choice probabilities for any assortment even with exponential size distribution. Canada, hzhang01@sauder.ubc.ca Lifei Sheng, Mahesh Nagarajan

Our procedure crucially exploits the symmetries of the Mallows model and leads to a compact IP formulation for assortment optimization. We also give an efficient near-optimal approximation for the IP. 2 - Auctions In The Online Display Advertising Chain: A Case For Transparency Amine Allouah, Columbia University, 520 W 122nd Street Apt 24, New York, NY, 10027, United States, mallouah19@gsb.columbia.edu Omar Besbes In the online display advertising market in which auctions are run to sell impressions in real time, advertisers most often bid for impressions through intermediaries. We investigate the impact of the active role such intermediaries take on the selling mechanism that sellers should use and on the performance metrics of the different agents in the advertising chain. 3 - Fast Provably Near-optimal Algorithms For Dynamic Assortment Optimization Ali Aouad, Massachusetts Institute of Technology, aaouad@mit.edu Retsef Levi, Danny Segev We study the dynamic assortment planning problem, where the demand is stochastic, and retailers’ decisions need to be robust (revenue-wise) to stock-out events, elicited by the inventory limitations. While being key to revenue management, particularly in retailing and airlines, the computational aspects of such problems are still wide open. We devise the first efficient algorithms with provable performance guarantees, under several common modeling primitives, including the widespread Multinomial Logit choice model. In practical comparisons against incumbent heuristics, our algorithms improve the revenue by 9% to 35%, with better computational efficiency and robustness in most cases. SA31 202C-MCC Derivatives in the Operations/Finance Interface Area Sponsored: Manufacturing & Service Oper Mgmt, iFORM Sponsored Session Chair: Arun Chockalingam, Eindhoven University of Technology, Den Dolech 2, Eindhoven, 5612 AZ, Netherlands, a.chockalingam@tue.nl 1 - The Optimal Hedging Strategy In A Competitive Supply Chain With Substitutable Commodities Ehsan Bolandifar, Chinese University of Hong Kong, ehsan@baf.cuhk.edu.hk, Zhong Chen This paper studies two risk-neutral processors procure two substitutable commodities from spot markets to process and sell through a retailer. First, we characterize the optimal index-based contracts for processors that indicates the processor’s optimal contract consists of a processing margin which is independent of its financial hedging decisions and a hedge ratio which is a function of commodity price volatility. Next, we characterize conditions under which, the retailer prefers to be exposed to commodity price risks. We show that processors can benefit from market pricing, when these prices are linked to input commodity prices and index-based contracts are a means to achieve it. 2 - Integrated Risk Management In Commodity Markets Fehmi Tanrisever, Bilkent University, tanrisever@bilkent.edu.tr In this paper, we examine the integrated operating and financial hedging decisions of a value maximizing firm, in the presence of capital market imperfections. Our results show that the working capital and the hedging polices of the firm interact with each other in a multi-period dynamic inventory model. In particular, looser working capital policies lead the managers to take relatively more speculative positions in the market to maximize firm value. 3 - Production Planning With Shortfall Hedging Under Partial Information And Budget Constraint Liao Wang, Columbia University, lw2489@columbia.edu, David D Yao We study production planning integrated with risk hedging by considering shortfall (from a pre-specified target) as the risk measure. The optimal hedging strategy is identified via a dual lower-bound problem, and takes the form of a digital option combined with a put option; and optimizing the production quantity, given the optimal hedging, is shown to be a convex minimization problem. With both production and hedging optimized, we provide a complete characterization of the efficient frontier, and an explicit quantification of the shortfall reduction achieved by hedging.

27