The COVID-19 pandemic has promoted the fintech industry to rethink its development goals and shifted its business priorities to the original focus – risk management. In the post-pandemic era, financial institutions must learn to become more experienced in dealing with emergencies and improve their risk control capability.
ADVANCE.AI devotes itself to the research and development of big data and artificial intelligence technology with deep thinking of the future industry's multi-scene integration requirements. For helping financial institutions and various enterprises implement comprehensive risk management, including identify market risks and customer risks effectively and deploy fully automated machine approval processes, ADVANCE.AI insists on in-depth research, walks in the front line of business with clients, and participates in the discussion of customer needs in order to help them achieve better user experience and more effective risk control practices.
The outbreak of the COVID-19 pandemic has led to a global economic downturn and a climbing demand for debt. Meanwhile, multi-platform loans have grown accordingly with the rising unemployment rates in many countries and regions.
Multi-platform loan (also abbreviated to "multi-loan") refers to the act of applicants applying for loans from two or more financial institutions. Generally speaking, each individual has limitations on his or her repayment capacity. Therefore, when a borrower has implemented loan applications on multiple lending platforms, his or her risk of late payment will increase.
The industry research has long pointed out the risk of late payment for multi-loan borrowers: three to four times higher than ordinary customers. Besides, the probability of default will rise by 20% when a loan borrower applies to one more financial institution.
As a result, once the economic downturn worsens and multi-loan applications are out of control, the pressure on the economy will become a trigger for the outbreak of a debt crisis.
Therefore, financial institutions need to pay more attention to multi-loan borrowers and precisely identify them by analysing the details of their loan demand, such as interest rates, debt amounts, cycles and application intervals, in order to discern which loan application belongs to a high credit-value customer's behaviour and which multi-loan application will lead to late payment and fraud risks.
Multi-platform loans are usually caused by the following reasons:
1) Instalment consumption
2) New loan application after repayment
3) Multiple failed loan applications (Basis for identification: loan application interval, which usually coincides with the loan approval cycle)
4) Malicious fraud (Basis for identification: lack of interval between loan applications, e.g., applying for multiple loans at the same time)
Through the mechanism of industry's joint prevention and control and big data laboratory method, ADVANCE Multi-platform Detection can effectively identify the risk of cross-platform loan applications by multi-dimensional data matching with relevant credit data based on the inputted applicant information such as ID card number and phone numbers and can display the results according to the occurrence time of borrowing and nature of the lending institution. The results are displayed to help financial institutions and lenders make credit decisions more efficiently, and the application scenarios are as follows:
Help financial institutions or lenders analyse users' credit based on historical multiple lending records and pre-screen high-risk customers based on recent multiple lending behaviour.
help financial institutions or lenders improve the accuracy of their models or strategy references, and enhance the effectiveness of their lending decisions. For example, it can determine the applicant's institution selection preference, interest rate preference, loan intention, loan purpose (analysing whether it is malicious or high-frequency borrowing) based on time slices
Help financial institutions or lenders to monitor their lending customers, identify "high co-debt risk", and take early risk management measures, such as adjusting risk exposures, updating monitoring strategies and implementing pre-collection.
In Indonesia, ADVANCE Multi-platform Detection can cover 90% of the local credit-active population; in Thailand and Vietnam, the coverage rate has reached 75% and 80%, respectively. Furthermore, in terms of borrowing occurrence time recognition, the smallest recognition unit can reach " the last 1 hour", and the most extended recognition period can reach 360 days. Besides, it also can identify mainstream types of licensed financial institutions or lenders around the world. For example, in the Indonesian market, ADVANCE Multi-platform Detection can identify banks, P2P lending, multi-finance institutions and other lending institutions that are regulated and licensed by the Indonesian Monetary Authority and the Central Bank of Indonesia, respectively. Overall, it features high coverage, accurate identification, good feature classification and strong robustness. In practical usage scenarios, such as reviewing individual cash loan applicants, the AUC value of the relevant OOT validation sample can reach 0.65.
[1] Data cited from People's Daily: How did cash loans become trap loans with ultra-high interest rates and violent collection; Data source: Zhaoyin Qianhai Financial Report, October 2017