消费行为与信贷违约——基于自我调节理论的实证分析

Consumption Behavior and Credit Default——Empirical Analyses Based on Self⁃Regulation Theory

在金融科技快速发展的今天,信贷风险是金融行业关注的重心。传统研究在分析用户信贷违约行为时,主要集中于用户一些基本的、静态的信息。结合近年来日益普及的数字化支付交易,本文研究了消费行为与信贷违约的关系。本文首先从自我调节理论出发,对品类级消费倾向的价值进行了理论分析。在该理论的指导下,本文重点关注人们在医疗健康和教育方面的消费行为。通过运用问卷调查和计量经济学方法,本文实证验证了品类级消费倾向对信用风险评估的解释价值。值得一提的是,在实证中,本文发现虽然医疗健康和教育方面的消费频率不高,但对于了解借款人的需求、动态监测和评估其信用状况有很强的指示作用。进一步地,本文考虑了未来导向消费倾向对信贷违约的作用机制,分析了未来导向对用户违约的影响。结果表明,重视未来导向相关产品消费的用户其违约的可能性更低。本文为行为理论指导金融风险管控提供了有价值的思路。

With the rapid development of financial technology(i.e.,FinTech)today,credit risk has always been the focus of the financial industry.When analyzing individual users'credit behav-ior or assessing their credit risks,existing studies mainly focus on some basic or static information about users.Combined with the increasing popularity of digital payment transactions in recent years,this paper uncovers the relationship between consumer behavior and credit default.To under-stand the relationship between consumption and credit behavior,inspired by the self-regulation the-ory,we focus on two product categories:health and education.Self-regulation theory suggests that people engage in deviant behaviors due to self-regulation failure.Such failure is rooted in exposure to excessive impulses,as the energy resources in the human body can inhibit only a limited number of impulses.Consequently,failure to control oneself regarding one thing is more likely to be ac-companied by self.regulation failure respecting another thing.A financial credit deviation is a form of self-regulation failure,as it runs counter to one's fulillment of financial obligations and is al-ways caused by overspending and limited savings.Conversely,consumption in health(especially physical exercise and health products)and education(especially books and self-development)may indicate successful self-regulation,because people need to effectively manage and control mo-tivations and behaviors to plan for learning or exercise,in order to improve themselves in positive ways.Therefore,we postulate that the consumption tendencies within these two categories are asso-ciated with people's credit bchavior,which postulates,of course,requires empirical validation.Empirically speaking,we collaborated with Tencent.Recently,in 2020,Tencent released a new consumer loan product,called Fenfu,which allows users to spend money first and then apply for a loan later.With such a loan product,users can enjoy,in effect,a post-payment shopping experi-ence.For those who choose to adopt Fenfu,they are first assigned with a credit line,which is set by Fenfu's experts based on demographics.Users can repay at any time after consumption and Fen-fu charges its users a fixed daily interest rate according to the daily unpaid amount.For the purpose of this research,we analyze 22000 Fenfu users randomly selected from its nationwide pool.Fenfu assigned a label for each user to indicate whether he/she had ever defaulted up to the time that we collected data,in June 2021.We next obtain granular information on each user's historical transac-tion records that had occurred within the 6 months prior to their Funfu activation dates.We obtain 4187445 records in total.We adopt a standard econometric approach combining propensity score matching and logit model estimations to verify the relationships between consumption tendency and credit risk.Our empirical analyses provide strong evidence supporting our theoretical justifica-tions:that is,users with a higher consumption tendency toward health-and education-related prod-ucts present a lower default probability.Statistically speaking,we observe that the default probability decreased by 3.2%for health type users as compared with the control group.Meanwhile,the probability of default decreased by 2.0%for education-type users as compared with the control group.Furthermore,we analyze the impact of future orientation on user default.The results show that users who value future-oriented consumption of related products are less likely to default.Our empirical results show that future-o-riented consumption in the health or education category had significant negative effects on credit default behavior.Interestingly,we notice that the effect sizes were larger than the average con-sumption pattern in the health or education catcgory,indicating that future-oriented consumption tendency could better rflect an individual's self-regulatory behavior,thus resulting in a lower de-fault probability.Our work contributes to the existing literature in multiple ways.In particular,we expand the scope of the application of self regulation theory to the analysis of individuals'financial behavior.Based on this stream of literature,we disentangle both theoretically and empirically the relationships be-tween an individual user's consumption tendency and financial credit risks.That is,we bridge the gap between the two types of individual behavior,which,in turm,provides the theoretical founda-tions and practical guidance for the usage of consumption records in the FinTech industry.The aforementioned theory-guided feature construction enables us to build sound relationships between the constructed consumption features and individual credit behavior,which boosts our understand-ing of the underlying psychological mechanisms of financial deviation.Whereas domain theories are applied to guide the choices of possible features and relational hypotheses to explain outcomes,it would be even more intriguing if the theory-driven features can be employed in a prediction model to achieve satisfactory prediction performance,which by implication,has the essential ad-vantage of both generalizability and explainability.Regarding managerial implications,we unravel the relationships between consumption and credit behavior.In particular,we identify three product categories and proved their value,both theoretically and empirically,to any examination of a target user's credit risk.Financial professionals will be able to leverage our empirical findings in order to augment their decision-making;they will also be eager to scrutinize the product categories highlighted by our study for a deeper understanding of consumer bchavioral tendencies.