Lemon or Cherry? The Value of Texts in Debt Crowdfunding

Peer-to-peer (P2P) lending, as a form of debt-based crowdfunding, has received much attention in the past several years. Text, in particular, is a prevalent feature but much less understood. While there have been some studies on the role of text in this context, these studies typically consider text as a control variable, and use small, manually coded samples. Our study is one of the first to use a scalable approach to examine the informational value of texts borrowers write when requesting funds in debt crowdfunding. We first examine data from exogenous events on Prosper.com, and show that investors indeed consider textual descriptions when investing. Then, we show that text features can indeed explain and predict loan default, using both explanatory and predictive models. Finally, we show that investors correctly interpret the informational value of some, but not all, features of texts. Our study points to opportunities for efficiency improvement not yet documented in the crowdfunding literature, and has implications for researchers, platform owners, and regulators.