Answers are grounded in this paper and cite their sources.
Most Brazilian insolvency prediction models were developed before accounting convergence to international standards and with samples from different sectors. This study proposed and validated a logistic regression model for predicting the insolvency of companies in the industrial goods sector listed on the B3 and publicly traded companies in segments related to the industrial goods sector, using economic and financial information from 2012 to 2024, in addition to exploratorily comparing its performance with models from the Brazilian literature. The sample comprised 23 insolvent and 59 solvent companies. Correlation analysis, likelihood analysis, Wald test, logistic regression, and validation by Jackknife and ROC curve were applied. The final model was composed of three statistically significant indicators: immediate liquidity (X4), asset immobilization (X37), and operating cash flow over total liabilities (X56). The coefficients showed a negative sign, indicating an inverse association with the chances of insolvency. With a classification threshold of 0.30, the model showed 82.6% accuracy in identifying insolvent companies and 84.1% overall accuracy. The area under the ROC curve was 0.856, decreasing to 0.845 during validation. It is concluded that the model presents relevant discriminatory capacity and potential sectoral utility, although its results are limited by the sample size and the absence of external temporal validation.
No takes yet. Share an insight, caveat, or question.
CAMPOS et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: