摘要
logistic模型是研究企业财务失败和信用违约的主流方法,但它不能很好地控制模型之外的因素对违约事件的影响。针对这一问题,本文引入截面数据的随机效应logistic模型,并对2005-2010年我国沪深两市中小上市企业的财务失败风险进行实证检验。结果表明,随机效应logistic模型相比logistic模型具有更高的样本内判别效果和样本外预测效果,而且在财务指标基础上加入公司治理指标能进一步提升模型的预警能力。研究结论对中小企业信用风险评估等具有一定的参考价值。
Small-and-Medium Enterprises （SMEs） play an indispensable role in increasing employment,encouraging innovation,and promoting economic prosperity.However,appropriate models of measuring credit default risks for SMEs are not available to commercial banks,thus leading to restricting loans for SMEs.Although many domestic researchers have been modeling default prediction,their models mainly focus on large enterprises and are ineffective when applied to SMEs.Logisic model is currently the most popular method in modeling credit default risk.However,its ability to predict out-of-sample default events is limited because of its poor performance in controlling those factors in the model.To solve this problem,we employ a random effects logistic model to improve the estimation accuracy of parameters in logistic model.Besides,most domestic researches consider only financial ratios,but ignore non-financial factors such as corporate governance factors.Some researchers argue that corporate governance may affect the quality of financial information and the effectiveness of measuring credit default risk.In this study,we use financial distress events （Special Treatment events） as proxy indictors for credit default,and employ both financial ratios and corporate governance factors to model a random effects logistic model of credit default for SMEs.In Section I,we introduce the principle and estimation procedure of random effects logistic based on cross sectional data.Under the assumption that the default probability follows Beta distribution,a likelihood function of random effects logistic is derived and the maximum likelihood estimation （MLE） is employed to obtain parameters.In Section Ⅱ,we discuss data source and variables application.In Section Ⅲ,we establish an empirical work under the model specification of random effects logistic model,and compare its prediction power with logistic model.The results show that regardless whether corporate governance indictors are included the random effects logistic model performs better than the logistic model in both discrimination power within sample and prediction power out-of-sample.Robustness test under different critical values further suggests that random effects logistic model has a better prediction power and a higher stability than logistic model.Additionally,we find the prediction power is also improved when model specification considers corporate government indicators,which may eliminate the financial information distortion caused by potential earnings management behaviors.In sum,logistic model can control factors out of model specification due to random effect.Compared with logistic model,the parameter estimation is more stable and the prediction power is higher in random effects logistic model.Prediction power will achieve the highest when both financial ratios and corporate governance indictors are included in random effects logistic model,which also indicates that financial distress risk of SMEs depends on both financial and corporate governance conditions.Corporate governance information should not be ignored when modeling financial distress prediction model for SMEs.Although our study is based only on a limited number of SMEs,the findings can help model financial distress risk and measure credit risk for unlisted SMEs.Based on the estimation of the default probability with the internal rating-based approach （IRB） recommended by Basel Ⅱ,our findings may help commercial banks evaluate credit default risks of SMEs and promote commercial banks to augment loans for SMEs,and thus alleviating SME＇ s financing problems.
出处
《管理工程学报》
CSSCI
北大核心
2014年第3期126-134,共9页
Journal of Industrial Engineering and Engineering Management
基金
国家自然科学基金资助项目(71072097
71172066)
教育部人文社科重点研究基地重点资助项目(2009JJD630002)
