Clearing Pricing Optimization at Zara
<-- Return to the list
Date: 11-04-2008
Start Time:
1:00pm
End Time: 2:00pm
Speaker: Felipe Caro, University of California - Los Angeles
Location: Uris 333
ABSTRACT
As part of an ongoing collaboration with Spain-based retailer Zara, we address the problem of deciding the price markdowns during clearance sales. As it has been reported in several case studies, Zara mostly produces in small batches and tries to avoid price changes during the regular season. In fact, items that are not selling well are typically replaced by new and more fashionable products. This is part of the fast-fashion strategy, which Zara has implemented effectively, and as a consequence, the leftover stock at the end of the season is significantly less compared to a traditional retail. However, once the season is over, price markdowns are still required in order to liquidate unsold stock and free up space for the new season.
Currently, these decisions are made based on past experience, and its outcome can have a considerable impact on the total revenue of the season that is ending. We have approached the pricing problem by formulating two models, one to forecast demand during clearance sales, and the other to optimize the markdown decisions with a finite horizon in mind. The main challenge in the forecasting model has been to overcome the lack of pricing data due to Zara's single-price policy during the regular season. For the optimization model, we have built approximate dynamic programming policies based on the existing pricing literature. We have incorporated store-level operational constraints in order to make the pricing policy implementable under Zara's current conditions. This talk describes the details of both models and its implementation at Zara. A controlled field experiment is planned to take place during the summer clearance sales of 2008. If available, the results of this pilot will also be discussed in the presentation.
This is joint work with Jérémie Gallien (MIT Sloan).
BIO
Professor Caro joined the UCLA Anderson School of Management in 2005. His main research interests are related to decisions made under uncertainty with a strong emphasis on practical applications. In his PhD dissertation at MIT, he developed analytical models with demand learning for the dynamic assortment problem faced by fast-fashion retailers. This work made him one of the selected winners in the MSOM Student Paper Competition. His current research projects also include supply chain competition, inventory management for low volume seasonal goods, forecasting and exploration versus exploitation problems.
Prior to receiving his PhD, Professor Caro worked as an instructor at the Industrial Engineering Department of the University of Chile. He taught courses in Optimization, Dynamic Programming and Stochastic Processes, and worked on research projects involving natural resources, mostly in the forestry and copper industries. He continues to collaborate on a permanent basis with his colleagues in Chile.
At the UCLA Anderson School, Professor Caro teaches the MBA core course on operations and technology management, and doctoral level courses on stochastic models in operations management. In 2007, he was awarded the George Robbins Assistant Professor Award that distinguishes excellence in teaching at the UCLA Anderson School.