Uncovering Financial Constraints

Abstract

We classify firms' financial constraints using a random forest model trained on the Hoberg and Maksimovic (2015) text-based constraint measures. Our model uses only financial variables to predict constraints, allowing us to significantly expand the cross-section and time-series of classified firms compared to the text-based measures. We conduct a number of tests to validate the informativeness of our measures. Using our classifications, we provide evidence that returns of debt (equity) constrained firms are more sensitive to shocks to the cost of equity (debt) financing. The results highlight the importance of financial flexibility in determining a firm's exposure to financing shocks.

Keywords: financial constraints, machine learning, random forests, financial flexibility

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