Author's School

Olin Business School

Author's Department

Supply Chain, Operations, and Technology


English (en)

Date of Award

Spring 5-15-2023

Degree Type


Degree Name

Doctor of Philosophy (PhD)

Chair and Committee

Lingxiu Dong Panos Kouvelis

Committee Members

Jacob Feldman, Rene Caldentey, Naveed Chehrazi,


The main purpose of this dissertation is to study the optimization problems and the value of information in various commercial settings, especially in the emerging platform economy.

Chapter 1, “Data-Driven Asset Selling”. Motivated by online asset selling marketplace business (e.g., used cars and real estate), we formulate a data-driven asset selling dynamic pricing framework which utilizes platforms’ access to customers’ online behavioral data. With mild assumptions on the demand model, careful characterization of the problem structure shows that the model admits some ideal properties that facilitate our regret analysis under our dynamic programming setting. Instead of studying the policy performance with a long horizon and large quantities of inventory, we study the asymptotic policy performance over a single unit of product as the demand rate grows. We propose a deterministic approximation policy (DA policy) and show that DA policy provides an upper bound for the original problem and its induced pricing policy achieves asymptotic optimality as the scale of the problem grows properly. Later we consider a dynamic pricing scenario where an idiosyncratic latent value for each asset is unknown. We propose a Thompson-Sampling-based and a MAP-based pricing and learning policy. Since the platform is restricted in an infrequent pricing environment, within each decision epoch, an adequate amount of customer online behavior records is available. Utilizing large-sample deviation properties, we are able to conduct regret analysis on the TS and MAP policies. Finally, we use numerical experiments to show that our proposed algorithms could improve the revenue performance significantly compared with an algorithm that is currently implemented by a leading used car platform. Besides, we find that using a simple deterministic proxy of demand forecast is mostly harmless, while accurate estimation of the idiosyncratic latent value can make significant differences. Simulations also reveal that in our problem setting, the exploration step in the TS policy may not help to outperform the MAP policy. This indicates that the effectiveness of exploration highly depends on the nature of the problem, which may be of independent interest.

Chapter 2, “Cash Hedging Motivates Information Sharing in Supply Chains”. Finance literature well documents that firms’ cash hedging strategies heavily depend on the market conditions. Unsurprisingly, such decisions could be challenging for an upstream firm in a supply chain where the end market conditions are not transparent to him. In this paper, we study the interplay between firms’ information sharing behaviors and cash hedging strategies in supply chains. First, we argue that the presence of a supplier’s cash hedging decision may motivate downstream retailers’ voluntary market information sharing with the supplier, since making the supplier more informed of the market conditions helps the retailer handle her risk in the wholesale price. This also forms a new reason why a supplier should consider hedging since the cash hedging decision itself can be used as a bargaining tool during the information sharing negotiation with his retailer. Then we find for homogeneous Cournot-competing retailers, asymmetric information-sharing outcomes could emerge as an equilibrium where publicly sharing information typically will not hurt, especially, sometimes it can achieve Pareto improvement of the supply chain and consumer welfare. Finally, when a single supplier serves multiple markets, the heterogeneity across market sizes and the correlation among market shocks play big roles in shaping the equilibrium. Especially in a simultaneous information-sharing game, greater market size heterogeneity and negatively correlated market shocks are more likely to result in the nonexistence of pure Nash equilibrium. When the Stackelberg sequence is introduced, greater market size heterogeneity and positively correlated market shocks are more likely to induce information sharing in the equilibrium. Furthermore, in the multi-market setting, the existence of an information-sharing channel may hurt retailers, the system as a whole, and consumer welfare.

Chapter 3, “Display Optimization under the Multinomial Logit Choice Model: Balancing Revenue and Customer Satisfaction”. In this paper, we consider an assortment optimization problem in which a platform must choose pairwise disjoint sets of assortments to offer across a series of T stages. Arriving customers begin their search process in the first stage and progress sequentially through the stages until their patience expires, at which point they make a multinomial-logit-based purchasing decision from among all products they have viewed throughout their search process. The goal is to choose the sequential displays of product offerings to maximize expected revenue. Additionally, we impose stage-specific constraints that ensure that as each customer progresses farther and farther through the T stages, there is a minimum level of “desirability” met by the collections of displayed products. We consider two related measures of desirability: purchase likelihood and expected utility derived from the offered assortments. In this way, the offered sequence of assortment must be both high earning and well-liked, which breaks from the traditional assortment setting, where customer considerations are generally not explicitly accounted for. We show that our assortment problem of interest is strongly NP-Hard, thus ruling out the existence of a fully polynomial-time approximation scheme (FPTAS). From an algorithmic standpoint, as a warm-up, we develop a simple constant factor approximation scheme in which we carefully stitch together myopically selected assortments for each stage. Our main algorithmic result consists of a polynomial-time approximation scheme (PTAS), which combines a handful of structural results related to the make-up of the optimal assortment sequence within an approximate dynamic programming framework.