Informs 2015

389 / 552

INFORMS Philadelphia – 2015
387
4 - A Dynamic Learning Approach for Personalized
Promotion Recommendations
Adam Elmachtoub, Assistant Professor, Columbia IEOR, 500 West
120th St, New York, NY, United States of America,
Markus Ettl, Sechan Oh, Marek Petrik
Many companies are aiming to offer real-time personalized promotions to online
shoppers with the goal of increasing conversion rates and revenue. In this work,
we provide a dynamic learning model and algorithm that simultaneously
maximizes revenue while learning how customers choose based on their
attributes and the promotions they receive. We provide theoretical bounds on the
regret as well as new analytical tools to determine feature importance in the
context of promotion recommendations.
WA44
44-Room 103B, CC
Dynamic Pricing
Sponsor: Revenue Management and Pricing
Sponsored Session
Chair: Candace Yano, University of California, Berkeley,
IEOR Dept.and Haas School of Business, Berkeley, CA, 94720,
United States of America,
1 - Optimal Dynamic Pricing for Trade-in Programs
Mohammad Ghuloum, Doctoral Student, Indiana University,
1309 E 10th St, Bloomington, IN, 47405, United States of
America, , Goker Aydin,
Gilvan (Gil) Souza
Trade-in managers continuously monitor their inventory of used products, and
adjust the acquisition and selling prices accordingly. Considering such a firm, we
study a novel dynamic pricing problem, where not only the demand of the
product is random and sensitive to the selling price, but also its supply is random
and sensitive to the acquisition price.
2 - Pricing in Crowdfunding
Ming Hu, Associate Professor, University of Toronto,
105 St. George Street, Toronto, Canada,
, Mengze Shi, Xi Li, Longyuan Du
We study the pricing decisions under an all-or-nothing crowdfunding scheme.
First, menu or intertemporal pricing is more likely than a single price to be
optimal. Second, dynamic pricing (contingent on the pledge amount) can help
the creator to stay over the funding tipping point over time, increasing success
rate and profitability.
3 - Dynamic Competition under Market Size Dynamics:
Balancing the Exploitation-induction Trade-off
Nan Yang, Assistant Professor, University of Washington at
St. Louis, St. Louis, MO, 63130, United States of America,
Renyu Zhang
We study a dynamic competition model, in which retail firms periodically
compete on promotional effort, sales price, and service level over a finite planning
horizon. The key feature of our model is that the current decisions influence the
future market sizes through the service effect and the network effect. Using the
linear separability approach, we characterize the pure strategy Markov perfect
equilibrium in both the simultaneous competition and the promotion-first
competition.
4 - Optimizing Pre-season Order Quantities in the Presence of
Planned Promotions
Dimin Xu, UC Berkeley, Haas School of Business, Berkeley, CA,
United States of America, ,
Candace Yano
Most retailers plan major promotions well before a product’s selling season,
possibly to coincide with storewide sales events. We optimize the pre-season
order quantity for a product considering planned promotions (and consequent
time-varying prices), when demand is price- and time-sensitive and stochastic.
Our approach accounts for both systematic fluctuations and uncertainty in the
implied salvage value over the season. We present structural results and
managerial insights.
WA45
45-Room 103C, CC
Topics in Revenue Management
Sponsor: Revenue Management and Pricing
Sponsored Session
Chair: Florin Ciocan, INSEAD, Boulevard de Constance 77305,
Fontainebleu, France,
1 - When Fixed Pricing Meets Priority Auctions: Service Systems
with Dual Modes
Jiayang Gao, PhD Candidate, Cornell University, 507 Hasbrouck
Apts, Ithaca, NY, 14850, United States of America,
, Huseyin Topaloglu, Krishnamurthy Iyer
Suppose a firm offers two modes of service: a fixed price, FIFO queue, and a
priority queue. Customers choose a mode to participate, as well as their bids if
they join the priority queue. We prove that in the unique symmetric equilibrium,
customer behavior has a threshold structure, in which customers with very high
and very low patience levels join the priority queue, whereas those with
intermediate patience levels join the FIFO queue. We then discuss the firm’s
server allocation problem.
2 - Product Support Forum: Customers as Partners in
Service Delivery
Konstantinos Stouras, PhD Candidate, INSEAD,
Bd. de Constance, Fontainebleau, 77305, France,
, Serguei Netessine,
Karan Girotra
Online product support forums where customers can post complaints and
questions, or report issues about a product or service abound. More and more
companies crowdsource their product and service support back to their
customers, employing a few dedicated service operators.Through an analytical
model, we characterize the equilibrium behavior of such a service system and
compare it with a call center model.
3 - Econometrics for Learning Agents
Vasilis Syrgkanis, Microsoft Research, 641 Avenue of the
Americas, New York, United States of America,
, Eva Tardos, Denis Nekipelov
The goal of this paper is to develop a theory of inference of player valuations from
observed data in the generalized second price auction without relying on the
Nash equilibrium assumption. Existing work assumes that each player’s strategies
are best responses to the observed play of others. We show how to perform
inference relying on the weaker assumption that players use some form of no-
regret learning. We apply our techniques to a dataset from Microsoft’s sponsored
search auction system.
4 - Adwords Equilibria with Budgeted Bidders
Florin Ciocan, INSEAD, Boulevard de Constance 77305,
Fontainebleu, France, ,
Krishnamurthy Iyer
We examine a model of the AdWords market where bidders strategically choose
their budgets and bids, while the network can throttle bidders to optimize its own
revenues. We show the equilibria in this market take a simple form and that for
these equilibria the network’s optimal throttling policy is greedy.
WA46
46-Room 104A, CC
Empirical Studies in Public Services:
Health Care and Education
Sponsor: Manufacturing & Service Oper
Mgmt/Service Operations
Sponsored Session
Chair: Jun Li, Assistant Professor, Ross School of Business, University of
Michigan, 701 Tappan St, Ann Arbor, 48103, United States of America,
1 - A Multiple Case Study of Resource Flow in Education Systems
Samantha Meyer, Research Fellow, University of Michigan, Ross
School of Business, R5340, Ann Arbor, MI, 48109, United States
of America, Karen Smilowitz
The US spends more to educate its children than nearly every other developed
nation, but scores near the bottom on international tests. Yet, how the US could
better use its resources is hard to know. Social scientists focus on the way
resources influence power, trust, and competition, whereas operations scholars
focus on technical problems of resource distribution and use. The reality is that
both matter. In this study we examine the way social and technical issues interact
in education systems.
WA46