The determinants of reward-based crowdfunding project delivery performance: A configurational model based on Latent Dirichlet Allocation

Abstract

Abstract Reward-based crowdfunding (BRCF) projects as the most popular crowdfunding model, the funders expect the project owners to deliver the promised rewards within the specified delivery time. Previous studies examined various factors such as level and amount of funding received, project category and size that might influence the reward delivery performance. However, textual information of projects has rarely been studied for analyzing on time or late delivery. The main contribution of our research is applying text analytical framework that can extract latent semantics from the textual descriptions of projects to predict the reward delivery performance. More specifically, we use the Latent Dirichlet Allocation (LDA) topic model for effective extraction of topical features to form a configuration model, and Qualitative Comparative Analysis (QCA) was conducted for the antecedent factors. The findings indicate that the configurational model has high consistency and coverage, which indicate sufficient antecedent conditions for on-time and late delivery. The asymmetric analysis of this study also supports the application of complexity theory in the field of crowdfunding, which offers insights into the setting of appropriate antecedent conditions to achieve on-time delivery. The managerial insights are also discussed in the paper.