Personal Page

With Y. Feng and B. Xiao (2/7/2009), “Optimal Pricing for Sponsored Search Advertising”, First revision completed, resubmitted to “Management Science”.

Abstract: Sponsored search advertising has experienced a rapid growth in recent years and has become an efficient marketing tool for search engines. To increase advertising revenues, search engines influence advertisers’ bidding by setting fixed or variable reserve prices. However, it is not clear to them what the best reserve price should be. This paper presents a pricing model for the search engine and derives the optimal reserve price for sponsored search advertising. It focuses on the generalized second-price auction, which is widely used by major search engines including Google and Yahoo!. We show that if the value per click of bidders is a variable with increasing generalized failure rate, the expected revenue rate function is quasi-concave and there is a unique reserve price that maximizes the search engine’s revenue. The first-order optimality condition shows that the optimal reserve price in a sponsored search auction should be set higher compared to the one commonly defined in the literature. Sensitivity analysis is conducted to show how the optimal reserve price is affected by the population of advertisers, the number of ad links for sale and the distribution of advertisers’ willingness to pay.

With B. Xiao and J. Li, (2/18/2009), “Optimal Reserve Price for the Generalized Second-Price Auction in Sponsored Search Advertising”, First revision completed, resubmitted to “Journal of Electronic Commerce Research”.

Abstract: Sponsored search advertising has grown rapidly since the last decade and has become a significant revenue source for search engines. To increase revenues, search engines often use fixed or variable reserve price to influence advertisers’ bidding. It is not clear to them, however, how to determine the best reserve price. Much of the literature does not reflect the dynamic feature of the auction. This paper studies reserve price for the internet advertising auction in a dynamic framework. It focuses on the generalized second-price auction, which is widely used by major search engines (e.g., Google and Yahoo!). Unlike the assertion in the literature that the number of advertisers and the number of ad positions have no impact on the selection of reserve price, our result is noticeably different. We show that the optimal reserve price is affected by both factors. In particular, under a set of mild conditions, the optimal reserve price is equal to the expected value of some order statistic of advertiser’s per-click values. Simulations based on the continuous-time bidding process confirm our theoretical findings.

With V. Tilson and Y. Wang, (2008), “Channel Strategies for Durable Goods: Coexistence of Selling and Leasing to Individual and Corporate Consumers”, to appear in “Production and Operations Management”. Tentatively scheduled to appear in the final issue of 2009 (vol. 18, no. 6).

Abstract: In durable goods markets, such as those for automobiles or computers, the coexistence of selling and leasing is common as is the existence of both corporate and individual consumers. Leases to the corporate consumers affect the price of used goods on the second-hand market which in turn affect the buying and leasing behavior of individual consumers. The setting of prices (or volumes) for sale and lease to individual and corporate consumers is a complicated problem for manufacturers.

We consider a manufacturer who concurrently sells and leases a finitely durable good to both individual and corporate consumers. The interaction between the manufacturer and consumers is modeled as a dynamic sequential game, where each player seeks to maximize its own payoff over an infinite horizon. We study how the corporate channel, substitutability of new goods and used goods, and transaction costs in the second-hand market affect the manufacturer’s pricing decisions, consumer behavior and social welfare in the retail market.

Making a number of simplifying assumptions including two-period lifetime for the finitely durable goods we consider Markov Perfect Equilibrium as the solution concept. We show that the manufacturer can maximize her profit by segmenting consumers according to their willingness to pay. Selling and leasing are the mechanism used for price discrimination in the retail market. We show that as she leases a larger share of her production to the corporate consumer, 1) the manufacturer does not necessarily have to adjust the optimal selling price of new goods to individual consumers, and the volume of sales of new goods to individual consumers can stay the same; 2) the manufacturer does increase the retail lease price, and the number of individual leases decreases; 3) the net supply of used goods on the market increases, leading to a lower market price for used goods; and 4) more individual consumers are able to participate in the market and their collective welfare or net utility improves. We also show that as production costs increase the manufacturer increases prices reducing volumes across all channels. When transaction costs increase, the manufacturer reduces leasing in both corporate and retail channels.

With X. Shi, B. Xiao and Y. Feng, (2008), “Revenue Management in China: an industry and research overview”, to appear in “Journal of Revenue and Pricing Management.” Published online on August 15, 2008.

Abstract: To many researchers and practitioners in China, revenue management was not a familiar term until the beginning of the last decade. As the nation’s economy is booming at an unprecedented pace, its service sector has grown rapidly. Managing perishable products, which are commonly seen in service industries, is becoming a frequently faced issue in business practice. Research and applications of revenue management in China is drawing an increasing attention in both academia and industry. This paper surveys the latest development of revenue management in China. Although from the theoretical and practical point of view, research on revenue management is fairly new and much behind what is needed, the momentum and potential is visible. The broad spectrum of its application has generated various front-line research problems which will enrich the field.

With I. Karaesman, P. Keskinocak and S. Tayur. (2008), “Aircraft and Crew Scheduling for Fractional Ownership Programs”, “Annals of Operations Research” 159 (1): 415-431.

Abstract: Fractional aircraft ownership programs, where individuals or corporations own a fraction of an aircraft, have revolutionized the corporate aviation industry. Fractional management companies (FMC) manage all aspects of aircraft operations enabling the owners to enjoy the benefits of private aviation without the associated responsibilities. We describe here the development of a scheduling decision support tool for a leading FMC. We present mathematical models, exact and heuristic solution methods. Our computational results using real and randomly generated data indicate that these models are quite effective in finding optimal or near-optimal solutions. The first phase of the implementation of one of these models at the FMC led to a significant improvement in effective utilization of the aircraft, reduction of costs due to reduced empty moves, and hence increased profits.

With B. Xiao, (4/2008), “A Revenue Management Model for Products with Two Capacity Dimensions”. Under first revision for “Production and Operations Management”.

Abstract: Most revenue management models are developed based on the assumption that the perishable products under study are identical on the use of capacity, or the supply-mix is inflexible if the capacity usage differs among products. In reality, however, products or services often have multiple capacity attributes. Shipping capacity of container liners, for example, is characterized by both volume and weight. For a vessel, the mix of 20-foot and 40-foot containers is not fixed as long as capacity in each dimension is not exceeded. Restaurant revenue management aims to maximize the revenue per available seat-hour that captures both the number of dining tables and service time. Similar issues arise in the air cargo, trucking and health care industries.

We study the revenue management problem with two capacity features and formulate the problem as a continuous-time stochastic control model. With mild conditions we derive the optimal solution and explore its structural properties. We show that unlike the policy for single-dimensional revenue management problems, which depends only on price, the optimal policy for multi-dimensional capacity control considers both price and demand intensity. Moreover, the control policy is a threshold policy that displays a significant difference when the remaining supply-mix varies. Numerical examples are provided.

With I. Karaesman and P. Keskinocak, (12/2008), “Managing Uncertainty in On-demand Air Travel”. In preparation for resubmission.

Abstract: We study the scheduling problem for fractional management companies (FMCs) that provide on-demand air travel services. FMCs operate in a highly uncertain and dynamic environment where frequent changes in supply (e.g., due to aircraft break-downs) and demand (e.g., trip cancellations, new trip requests) occur throughout a scheduling horizon. Certain logistical factors require FMCs to prepare their schedules in advance without complete knowledge of the aircraft availability and the trips. These schedules are later updated as more information becomes available. However, modified schedules are desired to remain both cost-effective and persistent (i.e., close to the original schedule). We propose heuristics to create dynamic schedules for a FMC. The heuristics pro-actively solve the persistence problem by enforcing idleness of the aircraft in creating the original schedule. We also look at the effect of strategically positioning the aircraft during the scheduling horizon to serve yet-unknown demand.

We present results of computational experiments that quantify the values of our heuristics and the strategic positioning idea. The computational experiments are carried out in a simulator that mimics a FMCs day-to-day operations.

With I. Karaesman and P. Keskinocak, (1/2007), “Optimization vs. Persistence in Scheduling: Heuristics to Manage Uncertainty in On-Demand Air Travel”, peer reviewed, accepted in full paper in Proceedings of The 3rd Multidisciplinary International Conference on Scheduling: Theory and Applications. p242-251, Paris, France, summer 2007.
With I. Karaesman, P. Keskinocak and S. Tayur, (7/2005), “Scheduling Multiple Types of Time Shared Aircraft: Models and Methods for Practice”, Proceedings of The 2nd Multidisciplinary International Conference on Scheduling: Theory and Applications, p19-38, Stern School of Business, New York University, USA.
With K. Sycara, (2002) “Risk Management in Storable Energy Markets”, Proceedings of the First International Joint Conference on Autonomous Agents & Multi-agent Systems. Rome, Italy.
With S. Tayur, (2004), “Equilibrium Analysis of a Natural Gas Supply Chain”, Working Paper, Tepper School of Business, Carnegie Mellon.
With K. Sycara, (2003), “Decentralized Matching Schemes for the Navy Detailing Process”, CMU-RI-TR-03-51, Robotics Institute, Carnegie Mellon.
With K. Sycara, (2003), “Two-Sided Matching for the Navy Detailing Process with Market Complication”, CMU-RI-TR-03-49, Robotics Institute, Carnegie Mellon.