Research Interests: Online Platforms, Rating System, Service Operations, Behavioral Operations
Chen Jin is a postdoctoral research fellow in Operations, Information and Decisions group. He is interested in service operations and information systems with application in e-commerce and online platforms.
He earned his Ph.D. degree in Industrial Engineering and Management Sciences from Northwestern University in 2016.
Abstract: Platforms in the sharing economy such as Uber and Lyft adopt a bilateral rating system (BRS) that allows service providers to rate customers and to make accepting/rejecting decisions based on the customers' ratings while in the traditional online platforms (e.g., eBay), only customers have the privilege to rate the other party, i.e., a unilateral rating system (URS) is implemented. This novel feature of the rating system in the sharing economy changes the service provider's effort structure in a fundamental way, which in turn affects the pricing strategy of the platform and the welfare of service providers as well as customers. With a stylized model, we compare URS and BRS in the context of ride-sharing service to study their impact on the decisions as well as revenue/welfare of all stakeholders. Our results show that being empowered to turn down customers at the service providers' discretion (in BRS) may not always improve the economic situation of service providers. Specically, the platform could squeeze the service providers' profit margin (per order) forcing them to serve only highly rated customers and hence reducing both their transaction volume and the profit margin. This leads to a decline in the service quality compared with URS. Meanwhile, the platform could also suffer a loss from BRS when the customers' valuation to the service is high (e.g., in a city with a less developed public transportation system), as the service providers' "cherry-picking" behavior in selecting customers is particularly costly to the platform in that case. This makes the platform have to give some of its revenue back to the service providers to mitigate their over-selection behavior, which can potentially reduce the transaction volume substantially. In practice, the platform could change the decision time of drivers (to reject customers' request) based on the estimation of the customers' valuation (in each city) to the ride-sharing service and hence, switching between bilateral and unilateral rating systems effectively.
Chen Jin and Chenguang (Allen) Wu, Pricing Service Systems When Customers Collude.
Abstract: Many service and manufacturing industries face two types of customers, individual and group customers. An individual customer procures a service or product based on his/her self-interest while group customers act collusively with decisions made by a group leader. Examples of the group leader are a travel agency who sells group tours, as well as an intermediary agency who collects and resends the orders of individual buyers to a manufacturer. Being aware of the disutility resulted from congestion, the group leader regulates the demand of group customers to maximize the aggregate surplus of the customer base. In a queuing framework, we characterize the interaction between individual and group customers and analyze its implication to the service provider's optimal pricing and priority decisions. We demonstrate that the service provider may choose to only target individual customers when their demand is sufficiently large. If both types of customers have to be served, we find that a policy which prioritizes group customers generates more revenue than the First Come First Served policy and a policy which prioritizes individual customers. Our results explain the operational incentive of a common practice in many service systems: group customers receive special group rates and higher priority in admission lines.
Chen Jin, Laurens Debo, Seyed Iravani (Under Review), Observational Learning in Large-Scale Congested Service Systems.
Abstract: We study the impact of observational learning in large scale congested service systems with servers having heterogenous quality levels and customers that are heterogonously informed about the server quality. Providing congestion information to all customers allows them to avoid congested servers, but, also implies that less informed customers learn about the quality from observing the choices of other customers. Due to an exponentially growing state space in the number of servers, identifying Bayesian equilibria is intractable with a large, discrete number of servers. In this paper, we develop a tractable model with a continuum of servers. We find that the impact of observational learning on the customers' choice behavior may lead to severe "imbalance" of server load in the system, such that a decentralized system significantly under-performs in terms of the social welfare, compared with a centralized system. The decentralized system performs well only when (a) either the congestion costs are high and there are sufficient informed customers, or (b) when the congestion costs are medium or low and the aggregate capacity of high-quality servers matches the aggregate demand of informed customers. We also find situations in which making more customers informed about service quality leads to a decrease in social welfare. Our paper highlights the tension between observational learning and social welfare maximization and thus observational learning in large-scale service systems might require intervention of the platform manager.
Chen Jin, Laurens Debo, Seyed Iravani, Mirko Kremer, Observational Learning in Environments with Multiple Choice Options: The Wisdom of Majorities and Minorities.
Description: We study observational learning in environments in which customers choose among multiple options with uncertain quality for which they observe the aggregate choices of previous customers (the sales of each option). When customers have heterogeneous knowledge about the quality of the options, the choices of the better informed customers turn sales into informative signals, allowing less informed customers to learn about the options' quality. We characterize the equilibrium choices for environments with any number of options. Although uninformed customers avoid options with no sales, they often prefer minority options with low sales over majority option with higher sales, and we characterize the conditions for this effect: "minority wisdom" tends to arise when the number of options is large, and when the fraction of informed customers in the market is low. We test the prediction from our observational learning model in the laboratory. The data shows that human subjects occasionally follow minorities, but signicantly less often than predicted by the full rationality paradigm of our equilibrium model. Noisy decision making provides a simple explanation for the observed pattern: when minority options may emerge because of a random error, rather than a rational choice, their value as a quality signal is diminished.