Risk Controllers have to notice this: Is your risk control strategy a lossy closed loop or loss-free closed loop?

How to close loop of before-loan risk control strategy by GPS Asset Management System in Auto Financial Industry?

Author: Tony Yu

All Risk Controller should be concerned about one issue: how to verify the effectiveness of risk control strategies of auto consume financial corporation (including financial lease)?

In general, risk control strategy had formed before the application, and at this time, Risk Control Strategy, in a sense, is just an assumption made by directors for the risk situation and risk management. It needs data of actual business to verify whether it is objective and real. A risk control strategy with non-closed loop verification is a unprogressive and motionless .Such strategy will cause loss at the end or be weeded out sooner or later. However, even if we would like to verify the effectiveness of risk control strategy by closed loop, it will be verified after the issues happened in traditional risk control management. At that time, real loss had been caused. Well ,you have to say that it is a costly learning experience.

So can this cost be avoided or at least minimized? With the continuous development and evolution of digital technology, solutions can be found with the help of IoT (Internet of things) and AIT (Artificial intelligence technology).

Let's first look at how the current risk control strategy is designed. In general, risk control strategy of auto consumer financial is designed around the following factors:

I. Admission Strategy: factor determines the scene and scope of Customer Acquisition;

II. Anti-fraud Strategy: factor determines which strategy to be used to identify the fraudsters who have admittance of customer group;

III. Credit Evaluating Model: the basis of pricing, which determines how to confirm the risk or cost of customer's credit default;

IV. Quota Strategy: factor determines the risk range of exposure to a certain customer;

V. Interest Rate Strategy: factor determines the profitability and price competitiveness of the business.

Each factor in the design of risk control strategy in the current operational practice of auto consumer finance can be summarized as follows:

I. Admission Strategy

Admission Strategy usually provide the basic admission to the product scene design, attribute design, channel design and customer attribute requirements.

II. Anti-fraud Rules

Some common rules are as followed:

lBlacklists: Financial clients often have their own filled-in blacklists and may also share with peers, other financial institutions. It is an important part.

lApplication Behavior: Abnormal application behavior, which is usually accumulated according to their own data, but also the data shared with other institutions.

lBad Information: whether there is a bad record.

l Real-name Information: Refers to whether the customer used more than one identity information in the historical application. 

lConsumer behavior: Some companies require customers to submit data attached bank statements and consumption related information.

l Fraudulent group: The above steps are to evaluate the individual user's behaviors, while fraudulent group is to evaluate the relationship, usually through the three factors and other dimensions, to associate, and find the associated high-risk customers or customers with similar behavior in a short time.

III. Credit Assessment

Credit assessment is to evaluate users' capacity and willingness to repay, which is usually realized by establishing a credit evaluation model to calculate the credit score. Credit assessment comes from user information, which can be roughly divided into basic information, behavioral information, credit information, social information and consumption information, etc. The rule to establish model is to determine the credit evaluationd and variable weights. After the user information is input into the credit evaluation model, the credit score of the user can be obtained.

IV. Quota & Interest Rate

The strategy of quota and interest rate is not just a key factor to control and cover risks, but also determines the competitiveness of products in the market. In practice, the design of quota and interest rate are greatly influenced by market competition, thus brings higher requirements for the admission, anti-fraud and credit assessment strategies of the risk control strategy. Therefore, the quota and interest rate need risk control strategy to be more strict and precise in the fierce market competition.

Traditional risk control strategy design mainly relies on limited data and individual experience, and the recently arisen big data of risk control enriches the dimensions and data of risk identification, whilst cloud computing and artificial intelligence technology bring new insights and rational results of data in order to effectively prevent risk controller to be surrounded by the data. In brief, risk control ability get a great improvement. However, the core of risk control strategies is that even if with the help of big data and artificial intelligence technology, risk controllers cannot be omnipotent. Regardless of judgments made by people or machine, because they are both speculation made by limited information. This speculation can be infinitely close to the truth, but it is not reality. It needs real verification.

Now we know the basic principle and practice of risk control strategy design, and also know the risk control strategy is a set of methods of speculation by collecting data, judgment, and risk control under the risk controller’s risk perception and risk control assumptions. Assumptions, speculation and judgment requires verification, so how could we verify risks before the actual risk occurs?

The core nature of the risk control strategy is to prejudge the customer's risk and credit according to the customer information, however, one of the biggest problem is that a lot of information is provided by the client themselves and auto financing companies are unable to accurately and carefully verify the information due to manpower and cost, which can lead to the risk occurred, such as false information, stolen ID, lack of credit data, etc. and this kind of risk can be identified and validated through the digital management of loans or rental vehicles.

Wanway Digital Technology realizes the digital management of loan vehicle assets through the use of Internet of Things, Cloud Computing and Machine Learning Technology for auto consumer finance companies (including finance lease). Real-time online, visualization, real-time interaction and whole journey big data of Vehicle Asset can dynamically verify the effectiveness and accuracy of risk control strategy, promote optimization of the risk control model, and reduce actual risk occurred. The principle is as followed :

The digitized auto financial asset management scheme can verify and warn the risk control strategy before the risk actually occurs, as shown in the figure below:

From the attention only on before-loan risk control strategy to the attention of asset management of both before-loan and the post-loan, Digital management of assets run by advanced digital technology can enable the risk control strategy to be verified and adjusted before the actual risks occur, which will be the inevitable result of the depth development of digital technology in the field of auto finance.

[Tony Yu: COO of Shanghai Wanway Digital Technology Co., Ltd., early promoter and practitioner of digital transformation and intelligent upgrade of asset management in auto consumer finance.]