A model that assesses borrower creditworthiness based on population data, tested by financial companies and banks, can reduce lending risk by up to 20%.
This information was given by Colonel Vu Van Tan, Deputy Director of the Department of Administrative Police for Social Order (Ministry of Public Security, C06) at the Workshop on applying population data in assessing the creditworthiness of borrowers, on the afternoon of August 7.
According to Mr. Tan, this model is built according to FICO standards (the leading company in building customer creditworthiness assessment models, applied in more than 30 countries) of the US, and has now been basically completed with 18 residential information fields.
MB Banking and Finance Company (MCredit) tested 10,000 citizen data, PVcombank tested 20,000 data, Datanest 60,000 data. The results showed that it reduced the risk ratio when lending capital of banks and credit institutions by 7-20%.
"After testing, banks all want to officially deploy it in their processes," said Colonel Vu Van Tan.
The Ministry of Public Security's project to apply population data in assessing the creditworthiness of borrowers can help credit institutions reduce risks when lending. Photo: Giang Huy
The combination of the banking industry and the Ministry of Public Security in using data has brought many benefits, such as authenticating and synchronizing the management of personal identification codes with credit information of 41 million customers, deploying chip-embedded citizen identification cards to withdraw money at ATMs, and using electronic identification accounts for authentication.
According to the leader of the Ministry of Public Security, although modern technologies are applied, they are only used as tools, lacking information and data to support banks in making lending decisions. Borrowing capital for production and business still faces many difficulties, leading to the situation of black credit causing consequences.
According to Colonel Vu Van Tan, there are three main reasons: banks do not have a basis to evaluate and identify loan recipients; there is no policy to support the disadvantaged and there is a lack of a state management mechanism to control black credit.
Accordingly, C06 has coordinated with the School of Information Technology, Hanoi University of Science and Technology to implement a project to assess the creditworthiness of borrowers based on population data, using machine learning technology and artificial intelligence according to FICO credit reference standards in the US.
According to Deputy Governor of the State Bank of Vietnam Pham Tien Dung, credit scoring in Vietnam is an increasingly widespread and popular risk management tool in banking. For the model to operate effectively and predict future debt repayment ability, the accuracy of data plays an important role.
"To have a source of data to assess creditworthiness, it is necessary to share from alternative sources, especially the national population database," said the Deputy Governor.
Expanding data sources is also the first solution mentioned by Mr. Cao Van Binh, General Director of the National Credit Information Center (CIC) in improving the efficiency of assessing the creditworthiness of borrowers.
At CIC, this model was built in 2015. By 2019, due to expanding coverage, CIC had built a CB 2.0 model for assessing the creditworthiness of individual borrowers. The model was completed and the scoring results of the model were provided from April 2021.
According to Mr. Binh, CIC's information provision growth always reaches 15-20% per year, higher than the average credit growth of the economy. In the first 6 months of this year alone, CIC has provided more than 31 million information reports of all kinds.
However, for each bank, assessing the creditworthiness of customers still requires additional criteria.
BIDV representative said that the customer credit rating model uses statistical methods and sets principles and parameters, but users still have to collect information themselves, actively search for and verify information. However, when deploying retail credit products on digital channels, the existing internal credit rating system has many limitations in automatically collecting and verifying information and providing accurate results.
"Having information sources verified and authenticated by a third party, especially a competent state agency, is extremely important and meaningful in the bank's retail credit activities, especially with digital products," said a BIDV representative.
One of the solutions applied by this bank is to cooperate with the RAR Center - Ministry of Public Security to implement the Customer Rating Project based on citizen identification data. Based on the results of the backtest model, BIDV said it will research and propose the application of Credit score for some retail credit products.
Minh Son
Source link
Comment (0)