Takeaway:
Didi’s unique big data of travel scenarios has a very broad application prospect in the financial field. In the future, it can further cooperate with banks, insurance, payment and financial institutions to help traditional financial institutions improve the efficiency of resource allocation and reduce the cost of acquiring passengers and risk management. Big data of travel scenarios has important commercial value in transaction fraud identification, risk pricing, precision marketing, whole-life cycle risk management, growth operation and other aspects. The application and analysis ability of big data is becoming the core competitive factor for the future development of financial institutions. From the perspective of auto finance and auto loan products, this paper combines scene data with traditional credit risk control concepts to accurately identify credit risk changes in the process of business development, which plays a positive role in improving business model and reshaping user value.
Catalogue of 0.
- What is auto finance?
- What is Didi Auto Finance doing?
- Application of Didi Big Data in risk control of auto finance
- Problems and solutions from an asset side perspective
- Problems and solutions from a whole-process risk management perspective
- Three optimization points on data application
- Application prospect of Didi Big Data in the risk control scenario of auto finance
- Enterprise credit intelligent risk control
- Intelligent risk control for retail credit
1. What is auto finance?
Automobile finance mainly refers to the financial services related to the automobile industry, which is the financing method involved in the r&d, design, production, circulation, consumption and other links of the automobile. It mainly includes fund raising, credit installment, mortgage discount, financial leasing, and related insurance, investment and other activities.
Business Model
In the retail business, commercial banks and finance leasing companies serve as capital parties, dealers /4S stores/leasing companies serve as sales channels, and auto e-commerce platforms play a guiding role in providing financial products and services for installment purchase for individual consumers who need to purchase cars.
From the perspective of competition pattern, bank and manufacturer finance are the main players in the retail market, with absolute advantages in capital cost and channel acquisition. In addition, as an online diversion service provider, auto e-commerce platforms have also been active in the auto finance market in recent years to improve customer acquisition efficiency for traditional financial institutions. In terms of product types, sale-and-leaseback is the mainstream of the market, while direct leasing needs to develop rapidly.
2. What is Didi Auto Finance doing?
1) At the present stage, Didi auto finance is positioned to serve the travel ecology and provide low-cost financing solutions for drivers who want to purchase cars based on user value.
2) Build an internal risk control system for auto finance, continuously improve overall risk management ability, generate high-quality online car hailing financial assets and gradually form risk pricing ability through the accumulation and application of online car hailing scene data.
3) Provide high-quality financial assets and systematic risk control capabilities to traditional financial institutions, achieve efficient matching of capital and assets, and accumulate financial asset management capabilities. At the same time, as a bilateral platform connecting capital and assets, it will establish long-term partnership with mainstream financial institutions and continue to provide financial support for the ride-hailing system.
In the future, the business scope of Didi Auto Finance will continue to be enriched with the development of the travel industry ecology, extending to the entire travel industry chain, providing financial services for auto dealers, 4S shops, agents and other auto sellers to purchase cars and operation equipment, so as to meet the financial needs of all links in the upstream and downstream of the industry chain. Gradually form a new financial mode of automobile industry integrating information flow, capital flow and logistics.
3. Application of Didi Big Data in risk control of auto finance
Under the traditional credit framework, the risk control mode of determining repayment ability by the lender’s central bank credit investigation no longer meets the risk management needs of online car hailing finance. In the scenario of online car hailing, the risk control of auto finance puts forward higher requirements for the authenticity, stability and timeliness of risk warning of loan assets. It is particularly important to establish intelligent marketing and intelligent risk control decision-making system based on big data.
From the asset end:
C-end problem of auto loan: data in pre-loan access scenarios are not used as a supplement for personal credit investigation, data in loans are missing, there is no matching risk warning scheme, and the efficiency of post-loan collection is low, so dynamic credit score needs to be formed for online car-booking lenders.
Solution: Use Didi big data to supplement the traditional retail scorecard model, apply the data that can reflect the characteristics of individual credit risk in the scene to the field of auto finance, and formulate risk control policies and access standards. At the same time, the probability of default (PD) scoring model of the car-owning population in the system was established to pay attention to the significant changes of PD parameters and provide risk warning schemes under big data. Gradually build a comprehensive risk management system in the ride-hailing scenario, and improve the risk management capability of the whole process.
B-end problem of auto loan: The lack of credit investigation data of CP (Car Partners) by traditional financial institutions leads to their inability to effectively identify channel risks. Especially for small and medium-sized CP, it is difficult to obtain credit from traditional financial institutions.
Solution: With the help of big data of Didi platform, support investors in the credit approval of CP. To be specific, the basic information of channels and data dimensions that can reflect their asset scale, asset use efficiency and driver management ability are systematically sorted out to form model variables. Meanwhile, bad samples in the system are continuously accumulated to establish a CP semi-supervised model. The output result of the model is the comprehensive score of CP credit rating, which intuitively reflects the risk grade of CP. At present, the CP rating of auto finance is monthly output, which can dynamically reflect the change of CP risk level.
From the whole process of risk management:
In the actual operation process, we found the following problems in the three stages of retail vehicle installment loan before, during and after the loan.
Pre-loan access risk: the loan applicant is not the driver of the actual operation of the vehicle after the loan, that is to say, A loan B return. This problem usually occurs in the channel into the link. Auto financing has certain operational risk in the process of product sales, offline sales channels in order to improve the single rate, find the credit quality is good, people are more likely to pass before the loan review instead of the driver to apply for a loan, but the actual running drops driver difference of credit assets, reimbursement ability is not enough to support for the month, PD default probability is higher. Then the credit risk of the installment loan will be released gradually during the asset performance period after the loan.
The first single pull, lenders and driver information is inconsistent:
Operation risk in the loan: The lender returns the car within the duration of the loan, and the car will be compensated by the leasing company. After the leasing company finds a new driver, the new driver will operate the car and continue to repay the loan. In this case, traditional risk control in pre-loan access to the judgment of the initial lender and vehicle GPS positioning can no longer effectively reflect the risk changes of post-loan operating vehicles. When the loaned vehicles are matched with multiple Didi drivers in succession during their lifetime, leasing companies face great challenges in vehicle operation management, cash flow management and driver management, and sometimes the centralized withdrawal of multiple drivers will cause channel concentration risks.
Operation in a car at different points to match multiple drivers:
Post-loan overdue collection: the traditional credit risk control for network about car loans after data is missing, the do not have access to the lender under the condition of income and operating data, not sure every overdue debts behind the lender’s reimbursement ability and willingness to reimbursement, so can’t do for income payments than high, repayment ability of borrowers to priority collection. In this case, it is necessary to develop a collection score card based on the data of the lender platform and the operation data of loan vehicles, and conduct classified management of collection.
Didi Big Data can solve:
The establishment of a comprehensive risk management system for ride-hailing finance.
In the preparation of retail data and the development of model variables, a long list of models covering four risk factors including city, channel and vehicle is formed from the credit base dimension of the lender to achieve dynamic monitoring covering the whole life cycle of loan assets. At the same time, by accumulating the dependent variables of the model (bad sample) in the asset performance of the investee enterprise, the risk level changes can be effectively grasped, the early warning and response mechanism can be established, and the loss rate can be reduced.
Each risk factor is drilled to form multiple risk indicators, which are combined to form a risk control strategy. Through the comprehensive application of single strategy and multi-strategy, the early warning and risk prevention can be realized in time.
Specifically, the optimization direction has the following points:
Optimization Point 1: Optimized from the traditional lender risk assessment at the time of loan to the whole process multidimensional dynamic risk monitoring.
Traditional credit risk control only focuses on the lender’s single dimension of credit risk measurement, while in the scenario of online car hailing, urban policy compliance, vehicle operation status and channel management ability all play a decisive role in the change of credit risk in the whole credit process. In this regard, we use the continuous accumulation of didi ride-hailing scene data and bad samples to supplement the dimension of traditional credit data and optimize A card and B card.
Early warning demand analysis:
Time of loan: verification of anti-fraud information, data dimension including but not limited to platform side verification of driver, vehicle, human-vehicle matching, channel basic information, and screening of channel entry risks.
After the loan is made, the changes of the lender’s credit risk are reflected in real time through monitoring in the loan, and a big data risk warning system is established.
Establish big data internal evaluation and verification governance framework, internal evaluation and verification process methods, and provide optimization strategies and real-time processes at different levels. In the early-warning model, typical early-warning strategies in loans are as follows:
Driver dimension strategy: water stability, earning ability, whether has handled the identity card, etc. Vehicle dimension strategy: vehicle operation situation on the platform, matching situation of vehicle and driver, vehicle mileage, whether a vehicle license has been obtained, etc. CP channel strategy: channel negative information scanning, channel concentration risk events, compliance ratio, channel concentration overdue, etc. City compliance strategy: whether the online ride-hailing platform certificate has been obtained, the progress of the city compliance card handling, classification management, etc.
As the data dimensions continue to enrich, the dimensions of the four risk factors will gradually increase. At the same time, we also verified one by one in the actual business, and carried out strategy iteration through the results of driver A card B card model.
Post-loan collection: optimize collection scoring model. Analyze and monitor the overdue days, order pulling behavior and monthly average income of overdue drivers in real time, and obtain the comprehensive score list of repayment ability and repayment willingness corresponding to each overdue debt, so as to help improve the efficiency of post-loan collection.
Optimization point 2: Increase the time width and time point observation depth of data observation, and introduce foresight on this basis.
Through the long-term observation data, the single risk strategy more iterative and strategy application for verification, we will get the driver the historical average and regularities of the variation of the credit risk, combined with the business at this stage and future development trends, on this basis to get prospective adjusted PD (probability of default), significant changes of credit risk quantitative and qualitative evaluation.
Optimization point 3: Rely on big data analysis ability to form a comprehensive judgment on the change of business global risk and return.
Through the risk management of the whole process of c-end leasing vehicles, the driver credit portrait and CP channel portrait under the form of leasing products are gradually outlined, and the operation risks of auto finance in business models and products are quickly identified, such as leasing of leasing packages, CP compensation, and centralized default risks. Then the auto financial asset quality can be clearly and accurately measured to achieve the balance of risk and return on the asset end and capital end.
4. Wide application prospect of Didi Big Data in auto finance scenarios
Enterprise credit intelligent risk control
Direction: In the whole travel industry ecosystem, there are a large number of scattered service providers/channel providers of small and medium-sized enterprises. The daily operation data of these small and medium-sized enterprises on Didi platform reflects their operation ability, capital liquidity management and driver management ability. Multi-dimensional business data can fully support the data risk control method to obtain funds, and provide innovative decision-making solutions for businesses, including identification of customer abnormal behavior, differentiated credit approval, risk control and early warning of the whole process, quota setting, etc.
Progress: at present, some car drops platform partners have a business with a licensed financial institutions has been with us data in-depth discussion in the form of risk control of credit under the condition of the platform does not provide guarantees, by withholding driver balance and platform multidimensional data to establish a risk control model, to provide high quality car rental company performs credit financing.
Retail credit intelligent risk control
Didi platform has an obvious bilateral effect, that is, both the supply side and the demand side complete transactions through the platform, so a large amount of transaction and operation data will be deposited on the platform. When the automobile financial service object is a system inside a car population, by drops big data supplement the deficiency of the traditional retail scorecard, to the credit system data is applied to the auto finance business scenarios, such as for the risk control policies and access standards for product level, the output automated credit scoring, fraud, exposure to risk management, risk pricing, etc.
Gradually establish the risk management system in the ride-hailing scenario, and realize the innovation of internal evaluation model in data, decision-making and algorithm levels.
Including: pre-screening customer group, characteristic model building and training, anti-fraud rule design, online strategy verification, joint modeling with partners, online post-loan overdue management, etc.
With the accumulation of big data risk control capabilities, intelligent risk control systems can be established for different business types, no matter the product form is new car finance lease or vehicle mortgage loan. On this basis, the dynamic monitoring of platform data can help screen personal credit users with good asset performance, form a whitelist, automate approval and loan, and improve the efficiency of asset matching.
The author:
Tang Pei drops | auto financing business analysts
An engineering student with financial industry management consulting background believes that the meaning of life is strongly related to valuable work, and has been looking for smart, thoughtful, highly sensitive to business, and broad vision partners to join the team.
At the same time, you are also welcome to pay attention to didi technology official account, we will bring you the latest open source information and technology dry goods!
Didi Technology Official account:
