IE Seminar: “Unlocking the Value in Product Return Data: Inventory Management with Sales Dependent Stochastic Product Return Flows From Multiple Periods”, Enis Kayış, 1:30PM December 8 (EN)

Speaker: Enis Kayış (Özyeğin University)

Date: December 8, 2023
Time: 13:30 – 14:30
Place: EA-409

Title: Unlocking the Value in Product Return Data: Inventory Management with Sales Dependent Stochastic Product Return Flows from Multiple Periods

Abtract: In the fast fashion retail sector, handling product returns has become a significant challenge due to rapidly changing consumer preferences, increased share of e-commerce sales and increasing product return rates. These retailers are now obliged to account for product return flows in managing inventories and addressing their environmental impacts. Our objective is to study a retailer’s optimal inventory control policy under product returns to maximize expected profit over a finite horizon. We quantify the value of using detailed return information and demonstrate that our proposed model increases retailer’s profit compared to a model aggregating the return information.
In the first part of the talk, we model a period’s returns to be stochastically dependent on the previous period’s sales quantity. Using dynamic programming formulation, we solve for the optimal periodic review inventory policy and provide structural results on the optimal policy. Through numerical studies, we show that incorporating detailed sales-dependent returns could increase a retailer’s expected profit by 23%. Ignoring this dependency in determining the optimal inventory policy results with increased order frequency, higher levels of backorders and more leftovers which could eventually end up in a landfill, but above all could lead to a significant overestimation of the resulting profit.
Next, we extend the problem and consider stochastic product returns over multiple periods under decreasing product prices over time and order capacity constraints. Due to the resulting computational complexity, we propose an Approximate Dynamic Programming value iteration algorithm using basis functions. Our proposed algorithm reduces the solution time drastically without a significant sacrifice from optimality and generates solutions for instances that are not solvable exactly, otherwise. Using an extensive computational study, we propose several managerial insights regarding the optimal inventory policy, the increase in the retailer’s profit owing to using detailed return information, and the reduced landfill usage due to unsold merchandise.

Bio: Enis Kayış is an Assistant Professor of Industrial Engineering at Ozyegin University. He received his PhD from Management Science and Engineering from Stanford University in 2009 and his MS in Statistics from the same university in 2007. During 2009-2012, he had worked at Hewlett-Packard Labs as a research scientist in several projects including demand estimation, product portfolio management and pricing, healthcare operations, and forecasting. His current research interests include optimization models for data science, data-driven decision making and business analytics with applications in supply chain management and healthcare operations. Prior to his graduate studies, he received BS in Industrial Engineering and Mathematics, both from Bogazici University.