Sep 29, 2021
The basic model logic of any bank loan is the same – to look for the price difference between customers who can repay the loan at a higher interest rate and customers who can't repay the loan. In traditional banks, the work of screening the lender's credit will fall on the credit officer or credit manager. Not only will they judge whether there is financial risks in a person from rigid standards, but also they will be affected by intuition. Another criterion is credit score, which is similar to the information used by credit card companies to evaluate potential cardholders, but information viewing is labor-intensive.
Fintech loan business appeared in some Internet start-ups, which abandoned the traditional loan model and reduced the borrower's expenditure in a faster, more efficient and lower cost way. The core of financial technology is to regard technology as a substitute for traditional finance. It contains a wide range of businesses, and loan is only one of them.
Fintech loans inform the guarantee, interest rate and method of loan provision with various data. Generally, credit investigation institutions collect personal financial information, summarize it into a credit report and provide it to the lender, which includes historical loans, public records, credit query, income and working hours. However, the limitation of credit report is that there is limited or no records in Americans’ credit. If there is few or no credit cards usage records, it will be difficult to evaluate the credit rating.
The method to the problem of too few records to evaluate is to expand the data types, that is, alternative data, so that people without credit records can also have some credit scores. These alternative data are diversified, such as rent, water and electricity, job stability and so on and can be collected through ways such as browsing data, social media, shopping preferences and personal habits through mobile phones. The way of fintech loan is to use machine learning to intersperse loan evaluation to form an intelligent algorithm. Before using the intelligent algorithm, the weight of each factor is determined and evaluated by the loan model. Taking the larger sample group as comparison, the algorithm finally draws the report conclusion through independent execution.
The combination of financial technology and AI gives more people access to loans, yet 71% of consumers worry that AI will infringe on personal privacy to a certain extent.
The role of fintech loans in credit opportunities has greatly helped to alleviate the problem of low-income families and individuals. For some low-income families who are not qualified for bank loans, fintech loans eliminate the need for payday usury (with an annual interest rate of up to 400%) as the only option. Borrowers of fintech loans claim that their products increase opportunities to non-predatory loans so that everyone has the right to healthy finance.
First, in terms of interest rate, the borrower of fintech loan will provide loan products with an annual interest rate lower than 36%, and the borrower can obtain the loan with less payment. Secondly, the credit standard of fintech loans is lower than that of traditional loans, so fintech loans can occupy the market of consumers without credit score than traditional loans. In short, the use of alternative data in fintech loans not only benefits more borrowers, but also enhances the lenders' ability to identify the risks of borrowers.
In the context of fintech loans, lenders collect personal information and browse borrowers’ history for scoring estimation. Borrowers have to make a trade-off between borrowing and privacy protection. In today's frequent disclosure of privacy data in the credit industry, concerns about data transparency have gradually increased.
How to balance privacy and credit use in the context of financial technology? On the one hand, people believe that privacy protection is quite important that the government should restrict the development of fintech loans, or regulate fintech loans and ensure that only limited lenders can access data points. On the other hand, people believe that the goal is credit availability and privacy should not become an obstacle to obtaining credit information. In other words, the lender can rate the borrower's credit through various ways.
This paper does not advocate extremity and will not swing between the two views, but believes that the two are equally important. Financial inclusion will become the main direction in the future, meanwhile, it should pay attention to the privacy risk of data. The two aspects are not opposite. In terms of cost, although privacy is very important, it should not become the obstacle to Inclusive Finance. At present, a consent letter of binding forces is required for the collection of privacy data. Only with the agreement of the users can the third-party application be used to collect data. The logic is that the user voluntarily agrees and makes data disclosure. Critics worry that some underfunded people tend to ignore the importance of privacy.
However, this paper believes that the focusing on Inclusive Finance can go hand in hand with the privacy protection. The most important thing is to establish a clear access channel so that consumers can judge on the controversial data points, accordingly reducing risks and bringing more benefits while completing accurate assessment.
The risk of digital loan not only lies in privacy, but also algorithm discrimination is worthy of attention. Research shows that the discrimination of financial technology loans against ethnic minorities is obvious, and the interest on them has increased by 3%.
Algorithmic loans already exist and will continue to develop. In 2020, the launch amount of fintech consumer loans is expected to reach US $90 billion. Financial technology loans have been incorporated into the existing regulatory laws. This paper believes that regulatory policies should not restrict innovation, but there is a need for a plan to regulate financial technology loans.
The two laws regulate the consumer information protection in the loan industry. The Truth in Lending Act (TILA) focuses on the disclosure of information related to the financial terms of the loan itself, and will not assume too many obligations and responsibilities for privacy. The Fair Credit Billing Act focuses on the disclosure of substantive information, which is fair, just and respecting the privacy rights of consumers. It requires the lender to issue the reasons and considerations for credit evaluation, and indicate the reasons for rejecting the application to the consumer.
The case related to fintech loans is the case of inclusive communities of housing and community affairs of Texas Department in 2015. The plaintiff claimed that the defendant had class discrimination. The controversial point of the case lies in the determination of causality, which concludes in that there is no need to prove a specific discriminatory intention when filing a lawsuit for government acts that harm ethnic minorities. The court further ruled that the case supported information disclosure and could apply this rule to information disclosure cases.
Among many suggestions, the requirement for algorithm transparency is that the company disclose the algorithm code to the public, so that regulators and the public can determine whether the algorithm will harm people's interests. Algorithm transparency can also alleviate the privacy problem and solve algorithm discrimination to a certain extent.
Another solution is to allow users to limit the data collected by the company. Users can choose what information the company collects, they can alsoquit at any time when they realize that their privacy has been violated. However, at present, neither regulators nor the company itself are willing to return the discretion to users, but this paper believes that users are entitled to the operation logic and information collection of the algorithm.
The best solution is counterfactual explanations. Counterfactual explanation means that information can be provided without opening the black box. For example, a person's annual income is $30000, so he can't get a loan, yet he can if the annual income reaches $45000. In other words, counterfactual explanation is to make a hypothesis that can get the best result.
The combination of counterfactual explanation and fintech loan is reflected in that they allow the lender of fintech loan to comply with the provisions of FCRA. At the same time, they need to provide customers with their reasoning process and factors needed to be considered, so as to make customers realize their shortcomings and obtain loans after improvement. From lenders’ perspective, this is a mechanism to improve the risk characteristics, which enables customers to further reduce default; Borrowers can also make themselves meet the evaluation requirements by improving their own conditions. In other words, the implementation of counterfactual explanation is that customers can get loans by improving their own consumption behavior. It provides customers with standards for identifying and evaluating loan ability, and personalization provides improved solutions, so as to provide more benefits for both parties, and promote the development of Inclusive Finance while protecting privacy.