摘要
对于互联网金融企业而言,如何对借款者的信用进行准确评估,从而做好风险管控工作,并非易事。传统的金融信用研究一般都采用"硬信息"来评价借款者的信用;近年来,用户社交、消费记录、网络行为等各类"软信息"正在越来越多地被企业使用。但是,现有关于"软信息"在信用评估中的应用研究大多着眼于图像分析、借款描述等文本研究,而对短信文本的挖掘还不够。文章利用中国一家排名前列的金融科技公司的13 799位借款者发送的317 872条短信数据,基于自我建构理论,分析了"我"和"我们"两类词语的表达与违约之间的关系。研究发现:(1)独立自我建构倾向越高的借款者,违约率越高,而互依自我建构倾向越高的借款者则违约率更低。(2)相比于北方人,由于南方人显示出更强的互依自我建构倾向,其违约率相对较低。(3)文化教育水平不仅降低了借款者的违约率,而且显著调节了自我建构对信用风险的作用。文章既为互联网金融企业的贷款决策提供了依据,也对促进互联网金融市场的健康有序发展具有重要的理论意义和实践价值。
Although the internet financial market has risen rapidly recently in China, the high default rate has become a constraint to the healthy development of the industry. Because "soft information" can provide additional reliable signals about borrowers’ true credit risk, the related research in credit evaluation has been already the most cutting-edge trend. However, currently there is little research on the mining of SMS.Based on borrowers’ SMS, this research studies the impact of different self-construals on default by mining the quantities of "I" and "we". As a kind of self-cognition of consumers, self-construals directly affect consumers’ credit decision-making behavior by influencing their thinking mode, and indirectly influence consumers’ credit decision-making behavior by changing their resource constraints. We propose that borrowers with interdependent self-construals have lower default rates than those with independent self-construals.Self-construals are influenced by culture. The different tendency of collectivism and individualism in north and south of China leads to the difference of self-construals. We believe that the southerners are more interdependent and the northerners are more independent, which leads to a relatively low default rate in the south. The relationship between self-construals and credit risk is also influenced by education. Education will change people’s thinking mode and reduce resource constraints, so as to weaken the positive effect of independent self-construals and strengthen the negative effect of interdependent self-construals on default.Our data come from a well-known financial technology company in China. We receive a total of 13 799 samples from July to December 2017, including 317 872 text messages from users. We mainly use Logit regression and negative binomial regression to analyze the data, and find that there is a significant correlation between self-construals and credit risk. Borrowers with independent self-construals have higher default rates,while borrowers w
作者
张懿玮
高维和
Zhang Yiwei;Gao Weihe(College of Business,Shanghai University of Finance and Economics,Shanghai 200433,China)
出处
《财经研究》
CSSCI
北大核心
2020年第1期34-48,共15页
Journal of Finance and Economics
基金
国家自然科学基金面上项目(71572099)
国家自然科学基金海外及港澳学者合作研究基金(71728007)
关键词
自我建构
文化差异
信用风险
短信文本
软信息
self-construals
cultural difference
credit risk
SMS
soft information
