Economic Policy Uncertainty and Bankruptcy Prediction
Economic policy uncertainty; bankruptcy; financial distress prediction; econometric models.
The goal of this research is to verify that the economic policy uncertainty index increases predictive accuracy in multinomial mixed logit model that use accounting ratios, market and macroeconomic explanatory variables. This work starts with a 3-state formulation for nonfailed, insolvent and failed firms. The sample, with 43,608 observations from the United States of America’s market, with a time frame of 8 years (2012 to 2019). Base on the data of 5,451 firms listed, classification adequacy and predictive accuracy are compared using two models whose only difference is the American economic policy uncertainty index. The variable’s inclusion show that the model does not increase its predictive power and neither its significance to forecast the distress state of firms. Although the parameter’s signs are confirmed by the adopted theory. Tests were implemented to confirm indicator’s coherence and to classify firms as nonfailed, insolvent or failed with up to three years of anticipation. The results obtained were aligned with previous studies, alongside statistically significant variables, and similar coeficients’ signs as to those expected according to the literature. The prediction models developed have a public data foundation and indicate utility for administrators, investors, competitors, providers, employees, auditors, auditing firms and regulators.