We create stepwise logistic regression models to predict the probability of default for private nonfinancial firms under distressed financial and economic conditions in a developing economy. Our main aim is to identify and interpret the drivers of private firm probability of default. For applicability and efficacy purposes, we apply a unique real-world data set of Zimbabwean private firms. Our experimental results show that the ratios of earnings before interest and tax (EBIT) to total assets, bank debt to total assets, EBIT to total liabilities, accounts receivable to net sales and the ratio of current assets minus current liabilities to total assets, the age of the firm, the real gross domestic product growth rate and the inflation rate are all strong drivers of probability of default for Zimbabwean private firms. We conclude that accounting information is useful in differentiating between defaulted and non-defaulted private firms under downturn conditions in a developing economy. Moreover, we determine that the forecasting results of probability of default models are improved by incorporating macroeconomic variables.

中文翻译：

---

发展中经济体不良经济和金融条件下的企业违约风险建模

我们创建逐步逻辑回归模型来预测在发展中经济体陷入困境的金融和经济条件下私营非金融公司的违约概率。我们的主要目标是识别和解释私营公司违约概率的驱动因素。出于适用性和有效性的目的，我们应用了津巴布韦私营公司的独特真实世界数据集。我们的实验结果表明，息税前利润（EBIT）与总资产的比率、银行债务与总资产的比率、息税前利润与总负债的比率、应收账款与净销售额的比率以及流动资产减去流动负债与总资产的比率，公司年龄、实际国内生产总值增长率和通货膨胀率都是津巴布韦私营公司违约概率的重要驱动因素。我们得出的结论是，会计信息有助于区分发展中经济体低迷时期违约和未违约的私营公司。此外，我们确定通过纳入宏观经济变量可以改善违约概率模型的预测结果。