Online Information-aided Pricing: Toward Deeper Understanding of Signals in Unstructured Data
- Xin LI (Principal Investigator / Project Coordinator)Department of Information Systems
- Victor BENJAMIN (Co-Investigator)
- Yinghui, Catherine YANG (Co-Investigator)
DescriptionPricing is a critical part of business operations. By manipulating prices, a firm can influence the market for its long-term or short-term benefit. In the e-commerce era, the ability to change the price in real-time makes it easier to take advantage of dynamic pricing, where managers can monitor and adjust price continuously. While there exists a large body of literature on dynamic pricing (based on demand, inventory, etc.), this project takes a different angle to study how online information, including social media and mass media, affects pricing.In our previous research, we found that the accumulation of online product reviews affects e-tail price, which may occur due to firms’ anticipation of the outcomes of the reviews. This research plans to study further which signals embedded in the online textual information matter. In the process, we will advance text mining techniques to capture pricing signals from online textual information.The study plans to address two major issues. First, we want to examine whether online information provides additional signals for pricing. To address this problem, we will need to conduct rigorous econometric analyses for causality identification. In addition to the amount of online information, we will exploit the interactions between different aspects of online information as antecedents of price change. We also want to examine if such an impact changes over time due to the adoption of new information technologies by business managers.Second, we aim to improve the methods to detect semantic and sentiment signals from online textual information. Although there exist efforts in text mining on topic detection and sentiment classification, such studies were less aligned with the purpose of pricing. This research aims to develop novel methods that can be used by business managers for pricing. We will also explore methods that involve less human coding efforts, leading to lower application cost.The two issues to be examined in the study are interrelated. Ineffective signal capturing will cause difficulties in identifying pricing signals’ impact. It may take iterative efforts to address both issues in this research.This research is of both theoretical and practical value. Theoretically, it has the potential to deepen our understanding of online information’s impact on pricing in the context of evolving business environments. Practically, the text mining techniques developed in this research can be used by managers for their daily pricing decisions. The study has the potential to lead to a more agile pricing practice. The techniques can also be used in a wider range of business problems such as asset pricing.
|Effective start/end date||1/01/20 → …|