Revealed Wisdom of the Crowd: Informational Value of Bids
Crowds may be wise in general, but not all crowd members are equally wise. Identifying informative heterogeneities among a crowd’s members, especially their actions, should lead to better wisdom. We test this conjecture using data from debt-based crowdfunding where we observe not only investment decisions but also the ultimate repayment (or not) of those loans. We find significant informational value in the heterogeneity of the investors’ observable behaviors. Specifically, we characterize investor behaviors in two ways: bid size and bid timing. Loans funded with fewer but larger bids are less likely to default than those funded with a larger number of smaller bids. For loans with similar bid-amount distributions, those receiving larger bids earlier in the bidding process are less likely to default. These features also help improve the predictive performance of loan default models. Importantly, these features are easily accessible, and our method is easily scalable without significant privacy concerns.