How Rewarding is the Reward? Demand Estimation of Crowdfunding Platforms

Abstract

Reward-based crowdfunding platforms are best known to facilitate the fund-raising process for entrepreneurs and small businesses. Entrepreneurs seek money from the backers to kick off their projects in exchange for rewards. While the competition among the projects to get more supporters grow, many projects fail to reach their fund-raising target. We use a structural demand estimation to understand the role of different aspects of reward scheme design and pricing. More specifically, we apply an aggregate level discrete choice demand model on Kickstarter projects. We characterize the features of the projects by utilizing the document embedding vectors from natural language processing methods. We treat the endogeneity of the price by applying latent instrument variables and finite mixture model for price. The results show that the price coefficient is biased towards zero in the absence of endogeneity treatment. On the reward scheme design, we show low-level rewards cannibalize the more expensive ones. We also find that low-level rewards have the least contribution in customer welfare. Campaign creators and online platforms benefit from the insights of the model for reward scheme design and pricing.

Keywords: crowdfunding, demand estimation, price sensitivity, aggregate data, latent instrument variable, mixture mode, document embeddings