PhD Defense by Sungwoo Kim

Primary tabs

Thesis Title: Mechanism designs in online marketplaces



Dr. He Wang, School of Industrial and Systems Engineering, Georgia Institute of Technology


Thesis Committee:

Dr. Xuan Wang, Information Systems, Business Statistics and Operations Management, Hong Kong University of Science and Technology

Dr. Alan Erera, School of Industrial and Systems Engineering, Georgia Institute of Technology

Dr. Alejandro Toriello, School of Industrial and Systems Engineering, Georgia Institute of Technology

Dr. Benoit Montreuil, School of Industrial and Systems Engineering, Georgia Institute of Technology


Date and Time: Friday, April 15th, 2022, 09:00 AM (EST)


Meeting Link: https://gatech.zoom.us/j/4444829841?pwd=K1JXK3NkVXU0eWdlS3RrdWtXYndhUT09

Meeting ID: 444 482 9841

Passcode: 540855



Online marketplaces have grown fast in past years. An important issue of online marketplaces is how to design mechanisms (or policies) to handle uncertainties of supply and demand. For example, during the early stage of the COVID-19 outbreak, online platforms suffered from a sudden surge of demands. Even though the distributions of supply and demand are stable, designing a good mechanism is not easy to achieve. This thesis focuses on designing various policies and mechanisms to handle this issue. 


In Chapter 2, we focus on an order fulfillment problem with two warehouses. Three policies are considered: greedy, randomized-greedy, and continuous protection level policies. The competitive ratio analysis is used to measure performances of policies in the worst case. With two customer regions, we show the competitive ratios of three different policies, and then, prove that the proposed randomize-greedy policy performs best among all deterministic and randomized policies. We extend our analysis for the problem with multiple customer regions. The computational experiments are conducted to check the performances of policies in general instances. The results show that the CPL policy outperforms the others in general.


In Chapter 3, we consider a freight platform that serves as an intermediary between shippers and carriers in a truckload transportation network. The objective of the platform is to design a posted price policy that determines prices for shippers and payments to carriers, as well as how carriers are matched to loads to be transported, in order to maximize its long-run average profit. We formulate the platform’s optimization problem using dynamic programming. Since our formulation is intractable to solve, a fluid approximation of the problem is used. Based on the solution of the fluid approximation problem, we propose an asymptotically optimal static posted price policy.


In Chapter 4, we extend the scope of the previous freight marketplace problem to auction mechanisms. The Lagrangian decomposition based auction mechanism is proposed. Next, a hybrid mechanism which combines a posted price mechanism and an auction mechanism is introduced. This mechanism balances platform’s profit and carrier’s waiting time. We show that these auction mechanisms are asymptotically optimal and generate higher profits than the static posted price mechanism.


In Chapter 5, we cover multi-period dynamic mechanisms for a freight marketplace. We consider three types of mechanisms used in practice: posted price mechanisms, auction mechanisms, and hybrid mechanisms. Under any auction mechanism, carriers may find another booking option outside of the platform during the auction waiting time. We analyze the effect of this behavior in different mechanisms. To handle the complexity of our model, various asymptotically optimal mechanisms are proposed based on the fluid approximation. The computational results show that the static auction mechanism is outperformed by other mechanisms, while a dynamic hybrid mechanism outperforms the others.


  • Workflow Status: Published
  • Created By: Tatianna Richardson
  • Created: 04/05/2022
  • Modified By: Tatianna Richardson
  • Modified: 04/05/2022