(1) Broadening the technical channels of credit rating will help to achieve accurate rating.
First of all, Internet credit reporting is the development of traditional credit reporting mode in terms of technical means. Based on the rapid development of big data and Internet technology, Internet credit investigation has broadened the technical channels of traditional credit rating. Most traditional credit ratings are based on certain mathematical models. On the basis of investigating the correlation between default rate of individuals or enterprises and other observed variables, the relationship between default rate and credit rating is determined according to historical data. Finally, through the investigation of the rated person, the credit rating which can roughly represent the default rate of the rated person is given.
(2) It enriches the data sources of credit rating and helps to make the rating more realistic.
In addition to technical means, the traditional rating method has certain limitations. First, it is necessary to forecast future data according to historical data, but there is uncertainty in forecasting future data according to historical data. Second, we can only rely on the financial data of individuals or enterprises for analysis. When rating individuals or enterprises, the dimension is relatively single, so it is impossible to get a comprehensive three-dimensional rating. Moreover, in many cases, financial data are more easily tampered with, lacking other data for cross-validation, which reduces the reliability of rating. Internet rating can largely avoid the above shortcomings. First, Internet credit information covers a wider range of data, further enriching the data used for rating, including not only the internal data of traditional financial institutions such as financial data, payment data and transaction data, but also the external data of non-financial institutions such as social data, purchase records and evaluation records, which is helpful to describe individuals more comprehensively and accurately. Second, the frequency of using data for Internet credit investigation is higher, and the frequency of data collection for Internet credit investigation is real-time, which records every transaction or social content of individuals, so it is possible for Internet credit investigation to adjust its rating in real time. Behind the Internet credit information is a huge amount of transaction data, which can continuously realize cross-validation between data to enhance the authenticity of rating results.
(3) Enriching the business entities of credit rating is conducive to strengthening market competition.
According to the latest notice of the central bank, it means that the central bank has been allowed to access the personal credit rating business of eight companies, which has laid a legal foundation for these eight credit rating companies to legally carry out personal credit rating work and provided a solid legal guarantee for the development of personal credit rating.
shortcomings
(a) Data quality needs to be improved.
The cornerstone of Internet credit investigation lies in having massive data, which can be used to screen out suitable data. However, there are still major defects in the quality of data held by Internet companies. First, Internet companies cannot collect personal sensitive information, and there are forbidden areas in law, such as personal payment records, deposits, etc., and these sensitive information is an important factor affecting the results of personal credit investigation. Second, the data accumulation of Internet companies is too short, resulting in a long data width, but not deep enough. Any model needs constant verification and trial and error within a certain period of time. As far as Internet credit reporting is concerned, the accumulated time can't even meet the requirements, so the credit reporting results of Internet companies may not be convincing.
(b) The reliability of the Internet credit model needs to be tested.
On the basis of solving the data problem, the reliability of the model needs to be solved. The reliability of the model comes from two aspects. Firstly, how to determine a reasonable model; Secondly, after determining a reasonable model, how to make the credit subjects and users of credit results accept it. From the first aspect, it is very complicated to determine a reasonable model. The success of the model depends on three factors: model design, data matching and application scenarios. Thanks to big data technology and cloud computing technology, the model of Internet credit reporting is becoming more and more complicated in design. The more complex the model is, the more it can truly reflect the credit situation in theory. However, the more complex the model, the higher the data quality. Internet companies should prevent the mismatch between model design and data. In addition, there may be matching problems between model design and application scenarios. For different application scenarios, the more complex the model, the better. It is necessary to continuously design models for application scenarios.