Informs Annual Meeting 2017

WA10

INFORMS Houston – 2017

3 - Supply Constrained Location-distribution Problems in Public and Nonprofit Settings: A Fractional Programming Approach Chong Hyun Park, Southern New Hampshire University, Manchester, NH, United States, c.park1@snhu.edu, Gemma Berenguer There is a need to design location-distribution problems in public and nonprofit contexts where limited resources have to be allocated to different demand regions and, thus, a combination of efficiency and equity goals can rightfully be considered. We propose the use of a new inequity measure related to Gini coefficient that is based on the relative utility obtained by each demand region given a certain supply allocation. We suggest a novel bi-objective integer linear fractional programming approach to solve the suggested supply constrained location-distribution models and show the suggested algorithm can solve the problems in polynomial time. A food crisis in Angola is presented as a case study. 330B Auctions/Mechanism Design Contributed Session Chair: Rad Niazadeh, Stanford University, Palo Alto, CA, United States, radniazadeh@gmail.com 1 - Combinatorial Auction Mechanism for Freight Information Platform with Quoted Price Uncertainty Yuting Zheng, Huazhong University of Science and Technology (HUST), Wuhan, China, ytzheng@hust.edu.cn, Jianbin Li, Peng Hu We consider a combinatorial auction mechanism for O2O freight information platform with quoted price uncertainty. This research question derived from a O2O freight platform we field investigated for half a year. Meanwhile, a series of numerical examples with real data are enumerated in this paper with the different approaches to cope with the uncertainty of carriers’ quoted price which may enlighten the managers in related industry. 2 - Optimal Allocation of Budget for Incentive Structure Design in a Resource Sharing Network Pu Yang, Cornell university, 292 Rhodes Hall, Ithaca, NY, 14850, United States, py75@cornell.edu, Peter Frazier, Krishnamurthy Iyer We propose a spatial temporal model for network economic settings with heterogeneous locations, and characterize a mean field equilibrium in this system. We study the problem of how to optimally allocate a budget of incentives to maximize a system level objective, by alternating the equilibrium behavior of agents in the system through subsidies or penalties. We show that the intractable problem of finding the optimal payment structure can be reduced to a problem of finding the optimal equilibrium strategy achievable with the budget, and the latter problem is a standard constrained optimization problem with a low dimension domain that could be solved via mainstream constrained optimization methods. 3 - Preferences and Decision Support in Competitive Bidding Philippe Gillen, Center for European Economics (ZEW), L7,1, Mannheim, 68161, Germany, gillen@zew.de, Nicolas Fugger, Alexander Rasch, Christopher Zeppenfeld We examine bidding behavior in first-price sealed-bid and Dutch auctions, which are strategically equivalent under standard preferences. We investigate whether the empirical breakdown of this equivalence is due to non-standard preferences or due to the different complexity of the two formats (i.e. different levels of math./individual sophistication). We elicit measures of individual preferences and then manipulate complexity by offering various levels of decision support. Our results show that the strategic equivalence only breaks down in the absence of decision support. This indicates that the empirical breakdown is caused by differing complexity rather than non-standard preferences. 4 - Bernoulli Factories and Blackbox Reductions in Mechanism Design Rad Niazadeh, Postdoctoral Researcher, Stanford University, WA10 We provide a polynomial time reduction from Bayesian incentive compatible mechanism design to Bayesian algorithm design for welfare maximization problems. Unlike prior results, our reduction achieves exact incentive compatibility for problems with multi-dimensional and continuous type spaces. The key technical barrier preventing this in prior black-box reductions is that repairing violations of incentive constraints requires understanding the distribution of the mechanism’s output. We overcome this barrier by employing and generalizing a computational model in the literature called “Bernoulli Factories”, which is about simulating a new coin from old ones by only sampling. Palo Alto, CA, United States, radniazadeh@gmail.com, Shaddin Dughmi, Jason Hartline, Robert Kleinberg

5 - Ants for Auctions – Solving the Winner Determination Problem using Ant Colonies Abhishek Ray, PhD Scholar, Purdue University, 401 S Grant Street, West Lafayette, IN, 47907, United States, ray52@purdue.edu, Mario Ventresca At the fundamental level, any combinatorial auction can be treated as an efficient allocation problem widely studied in integer programming. Determining winner in such auctions is demonstrably NP-Complete. Previous work in estimating a winner in combinatorial auctions employs a depth-first branch and bound approach that imposes upper bounds on contribution of unallocated items to the maximum revenue for the auctioneer. Our approach differs drastically since we employ a probabilistic nature-inspired search heuristic that uses the concept of stigmergy to explore and exploit the search landscape. 6 - Online Auctions and Multiscale Learning Rad Niazadeh, Postdoctoral Researcher, Stanford University, Palo Alto, CA, United States, radniazadeh@gmail.com, Sebastien Bubeck, Nikhil Devanur, Zhiyi Huang We consider revenue maximization in online auctions and pricing. A seller sells an identical item in each period to a new buyer, or a new set of buyers. For the online posted pricing problem, we show regret bounds that scale with the best fixed price, rather than the range of the values. We also show regret bounds that are scale free, and match the offline sample complexity, when comparing to a benchmark that requires a lower bound on the market share. To obtain this, we generalize the learning from experts and multi-armed bandit problems to their “multi-scale” versions, where the reward of each action is in a different range, and the regret w.r.t. a given action scales with its own range, rather than the maximum. 332A Incentives in Queueing and Sharing Platforms II Sponsored: Manufacturing & Service Oper Mgmt, Service Operations Sponsored Session Chair: Rouba Ibrahim, University College London, London, WC1E 6BT, United Kingdom, rouba.ibrahim@ucl.ac.uk Co-Chair: Amy R Ward, University of Southern California, Los Angeles, CA, 90089-0809, United States, amyward@marshall.usc.edu 1 - Labor Platforms for On-demand Services Saif Benjaafar, University of Minnesota, 111 Church Street SE, Department of Industrial and Systems Engr, Minneapolis, MN, 55419, United States, saif@umn.edu, Jian-Ya Ding, Guangwen Kong, Terry Taylor We study consumer and labor welfare in on-demand service platforms that rely on self-scheduled workers. The platform decides on wages and prices knowing that customers are price- and delay- sensitive and agents have reservation wage rates that depend on their availability. 2 - Designing Efficient Budget-constrained Incentive Schemes: To Compete or to Cooperate? Ragavendran Gopalakrishnan, Research Scientist, Conduent Labs India, Etamin Block, 4th Floor, Wing-A, Prestige Technology Park II, Bangalore, 560103, India, ragavendran.gopalakrishnan@conduent.com, Sabitha Devarajulu, Rahul Ghosh We study the design of real-world performance-based incentive schemes, which are necessarily budget-constrained, must be simple to understand and implement, and satisfy some fairness criteria. The resulting competition among agents for incentives is modeled as a noncooperative game. We characterize budget- constrained performance maximizing incentive schemes, and find that the maximum performance is a concave function of the available budget. We also study competitive vs. cooperative fairness among high and low performing agents to quantify the efficiency loss in moving from Nash to dominant strategy equilibrium. We conclude with additional insights obtained from numerical experiments. 3 - Pricing in a Two-sided Market with Time-sensitive Customers and Suppliers Philipp Afeche, University of Toronto, Rotman School of Management, 105 St. George Street, Toronto, ON, M5S.3E6, Canada, afeche@rotman.utoronto.ca, Mustafa Akan We consider a firm that matches stochastically arriving and time-sensitive customers and suppliers. Agents are rational and choose whether to enter the market upon arrival. Customers and suppliers differ in their arrivals rates, rewards from matching, and waiting costs. The matching is instantaneous and frictionless. We characterize the structure and performance of the profit- maximizing and socially optimal pricing policies. WA11

444