Abstract: The credit rating is a measure of the likelihood for a debt default, and an assessment of the debt default risk. This paper uses the information gain to establish a credit rating model, and conducts an empirical analysis of the loan data of small industrial enterprises. Our contribution has three folds. First, according to the idea that the larger information gain, the greater performance of the default identification, we select the indicators that have a greater impact on the default status. The shortcomings of the existing research that does not take the default discrimination ability as indicator selection criterion are changed. Second, for a pair of indicators with a high correlation, we delete the indicator with a small information gain, so as to not only avoid the information redundancy between indicators, but also deleting indicators with greater default discrimination ability. Third, we use the information gain value to weigh the indicators to ensure that the indicators with greater default discrimination capabilities have greater weights. It has changed the disadvantages of the existing research in weighting for indicator. The empirical results show that the 31 indicators selected in this paper, including asset-liability ratio, industry prosperity index, and pledge guarantee, have significant default discriminant ability. Solvency is a key factor affecting the credit rating of small industrial enterprises.

Key words: credit rating, small business rating, default discriminant, information gain

摘要： 信用评级就是衡量一笔债务违约的可能性,评价债务违约风险的大小。本文利用信息增益方法建立了信用评级模型,并以小型工业企业贷款数据为对象进行了实证分析。本文的创新与特色:一是按照指标的信息增益越大、越能将违约与非违约企业区分出来的思路,筛选出对违约状态有较大影响的指标。改变了现有研究不以违约鉴别力作为指标遴选标准的不足。二是在相关程度高的一对冗余指标中,删除信息增益小、即违约鉴别能力差的指标,既避免指标间反映信息重复,又避免误删违约鉴别能力强的指标。三是利用信息增益值对指标进行赋权,保证违约鉴别能力越大的指标赋予的权重越大。改变了现有研究赋权不反映指标的违约鉴别能力大小的弊端。实证结果表明:本文遴选的包括资产负债率、行业景气指数、抵质押担保等31个指标对违约状态有显著的鉴别能力,且反映信息不重复。偿债能力是影响小型工业企业信用评级的关键要素。

关键词: 信用评级, 小企业评级, 违约鉴别, 信息增益