Entrepreneurial Learning and Disincentives in Crowdfunding Markets

Abstract

Reward-based crowdfunding has enabled entrepreneurs to interact with consumers even before product launches. However, this market persistently suffers from a high failure rate; that is, entrepreneurs fail to launch and deliver their products as promised. We investigate the extent to which this high failure rate is due to information distortion—entrepreneurs have uncertainty about consumers’ evaluation of the new products. We model the product-launch decisions of different types of entrepreneurs who raise funds through pre-selling on reward-based crowdfunding platforms and subsequently decide whether to continue with the product launch. To do so, we collect structured and unstructured data from Kickstarter’s digital video game category and classify attributes using supervised learning methods. We develop and estimate an integrated model of crowdfunding demand and entrepreneurs’ product-launch decisions. We find that the information entrepreneurs gather from crowdfunding sales has sizable impacts on the product-launch decisions of entrepreneurs with low managerial capital or new to the crowdfunding platform. Our counterfactual simulations suggest that platform policy regulating overfunded projects can
reduce the product-launch failure rate by about 13%.

Keywords: reward-based crowdfunding, entrepreneurship, pre-launch learning, supervised learning

JEL Classification: L26, M13, D82