Vijay Kamble is an Assistant Professor of Information and Decision Sciences in the College of Business Administration, and of Computer Science (by courtesy), at the University of Illinois at Chicago. He previously obtained his Ph.D. in Electrical Engineering and Computer Sciences from UC Berkeley (2015) and was a postdoc in the Social Algorithms lab at the Management Science and Engineering Dept. of Stanford University (2015-17).
His current research interests are in the areas of machine learning, statistical learning theory, market design, and optimization with applications to revenue management, pricing, and the design and optimization of online platforms and marketplaces.
| Name of Speaker | Dr Vijay Kamble |
| Schedule | 12 November 2021, 10am – 11.30am
(60 min talk + 30 min Q&A) |
| Link to Register | https://nus-sg.zoom.us/meeting/register/tZcsf-CqqT8rHt2NWfT9LgIr1u-ghDpdIlPc |
| Title | Pseudo-competitive games and algorithmic pricing |
| Abstract | With recent advances in artificial intelligence methodologies, algorithmic pricing in the face of unknown or uncertain demand has become ubiquitous in the practice of revenue management. While such algorithmic approaches are known to satisfy attractive revenue guarantees in well-behaved, non-strategic environments, the outcomes arising from such approaches in competitive settings remain poorly understood. Motivated by the goal of studying outcomes of algortihmic price competition in practical environments, we study a game of price competition amongst firms selling homogeneous goods, defined by the property that a firm’s revenue is independent of any competing prices that are strictly lower. We call this the pseudo-competitive property and the games of price competition induced by such revenue functions pseudo-competitive games.
We show that this property is induced by any customer choice model involving utility-maximizing choice from an adaptively determined consideration set, encompassing a variety of empirically validated choice models studied in the literature. For these games, we show a one-to-one correspondence between pure-strategy local Nash equilibria with distinct prices and the prices generated by the firms sequentially setting local best-response prices in different orders. In other words, despite being simultaneous-move games, they have a sequential-move equilibrium structure. Although this structure is attractive from a computational standpoint, we find that it makes these games particularly vulnerable to the existence of strictly-local Nash equilibria, in which the price of a firm is only a local best-response to competitors’ prices when a globally optimal response with a potentially unboundedly higher payoff is available. We moreover show, both theoretically and empirically, that price dynamics resulting from the firms utilizing gradient-based dynamic pricing algorithms to respond to competition may often converge to such an undesirable outcome. To address this concern, we finally propose an algorithmic approach that incorporates global experimentation under certain regularity assumptions on the revenue curves. This is joint work with Chamsi Hssaine and Sid Banerjee, both from Cornell ORIE. |
