Modeling the demand and supply of product related information; using evidence from YouTube

Abstract

Despite the role of product related information in new product launches, our knowledge about its demand and supply in the product reviewer market is very limited. The dissertation aims to fill this gap in the literature by modeling the economy of product information using data from YouTube. The main objective prior to the hypothesis formulation was to explore the product reviewer market on YouTube and identify the role, the demand, and the supply of product related information. We found that based on the products the videos are reviewing, we can identify different information markets on the platform, and this segmentation significantly differentiates the performance of the videos posted on them as well. However, the topics’ effect on the videos is diminishing over time. Then, we were able to formulate hypotheses regarding the characteristics of the demand and supply on the market and build model extensions aimed to answer them. First, related to the demand, we endogenized the overall interest towards the topic into a current state of satiation and topic awareness. The results indicate that both measures have a significant relationship with the performance of the videos, having positive and negative coefficients, respectively. Second, we also aimed to unfold the supply on the market and move away from the homogenous channels’ assumption. We considered two factors that can differentiate these channels, their sizes and their unobserved brand images. We found that the size of the channels has a significant positive impact on the performance of the videos, while it has a negative effect on the above-defined satiation and topic awareness. Our results suggest that the unobserved factors related to the image of the brand also significantly differentiate both the response variable and the topic effects. Finally, accounting for the long-term incentives of the channels, we aimed to derive a set of models examining their growth. The main hypothesis regarding these models was whether the performance of the videos translates into subscriber counts. We found that the performance positively influences the subscriber gaining process, and outstanding videos provide extra effects for this growth. In addition, we tested if the video-level reactions from the audiences can be related to the process and found that the average ratio of likes to views and dislikes to views are significant predictors of the subscriber count changes over time.