Uncovering Financial Constraints

Abstract

We classify firms’ financial constraints using a random forest model that uses only financial variables to predict constraints. Our methodology significantly expands the cross-section and time series of classified firms compared to text-based measures, while maintaining similar levels of informativeness. We construct two versions of the constraint measures. One version uses many firm characteristics to identify constraints. The other uses a small set of arguably exogenous characteristics. Using our measures, we provide two novel facts related to equity constraints: (1) institutional investors hold a relatively lower percentage of the shares of equity-focused constrained firms, while retail investors (Robinhood and 1990s discount brokerage account holders) display a relative preference for equity-focused constrained firms, and (2) the equity issuance and investment of equity-focused, constrained firms increases during periods of high investor sentiment.

Keywords: financial constraints, machine learning, random forests, institutional investors, retail investors

JEL Classification: G00, G30, G35, G1, G3