Email to a Friend

Depending on the size of your financial institution, they could either be using a custom made credit model or a cheaper credit bureau version. The cheaper version is just a generic model that may not appropriately fit the intended use. This is why larger firms use customized scoring models for each of their specific products.

Types of models

There are a range of scoring models available: application scoring models, behavioural scoring models, collection scoring models, fraud detection scoring models and bankruptcy scoring models. One industry expert cited that they have found that all types of models are used, as the nature of analysis differs.

Pitfalls with Current Models

A source from one financial institution cited that a key issue that is seen to be limiting the industry is that not enough attention is being given to how products get matched with different market segments.

Suitability of models

Another issue highlighted was the traditional problems associated with validation of model suitability. Decisions based solely on credit scores. One other key issue highlighted was that lenders cannot use credit scores alone to ensure that credit remains available to customers who would qualify for a low-interest loan. Banks should maintain multiple criteria to assist in balancing deficits in an individual’s credit score, preferably based on factors outside the scorecard characteristics.

Concerns with application of credit scoring models

Some industry experts believe that nothing is actually wrong with employing credit scoring models on product(s) that they were not initially designed for as long as they meet the test of validity for that product. Nonetheless, banks do need to be aware of the impact this could potentially have. It would be advisable to track performance and re-estimate the model if necessary in such instances.

Excessive overrides

Users are often reluctant to rely on a credit scoring model as they are in the habit of trusting their own judgements. Nevertheless, overrides can potentially result in problems. Regulators take a dim view of overrides above a certain threshold, typically 5% on the low side and 10% on the high side. Some banks have reported mechanisms in place to reduce the chances of violations of fair lending regulations. Such mechanisms include reviews of overrides as well as reviews of the rejected applications.

Performance

Tracking reports - Financial institutions should closely track a scorecard for its stability and to distinguish between the less risky and more risky consumers.

Handling the deterioration of performance

According to industry analysts, in a case where an organisation finds that the performance of their model is deteriorating, three approaches to fix this are available: Scorecard realignment, Scorecard weights re-estimation and redevelop scorecard.

Assessment methodologies - disparate treatment/impact

Banks that find that the performance of their credit scoring model is deteriorating are recommended to re-estimate their model and maintain updated database, according to regulators that Lepus spoke to.

Internal audit procedures

Frequency and validation of models - Frequency of validation or revalidation of credit scoring models usually depends on a case by case basis and it is standard practice to carry out validation at least every 18 - 24 months. Validation is generally carried out more often where the market is changing a lot. Monitoring reports - Large volume banks should aim to produce monitoring reports on a quarterly or monthly basis. The lending institution’s loan volume should determine how often reporting is carried out. Smaller banks that handle smaller volumes may take around six to twelve months or more to originate a large enough loan volume to perform such an analysis.

Monitoring of third party relationships

Lenders cannot have a great deal of control over the practices of their third-party brokers, especially where compliance with fair-lending laws, pricing policies, and the use of credit scoring models are concerned. One way in which lenders can prevent liability for third-party brokers is to test loan application files and documents concerning the pricing of the loan to the customer.

Regulatory compliance

One general way in which organisations can safeguard against fair lending violations, as well as other regulations, is to maintain sound internal controls. The key internal control that can be implemented is to make employees aware that they will be accountable for compliance, as opposed to having one person, or group, responsible for compliance. This helps to prevent violations.

Impact of Basel II -

According to industry experts, banks are spending large amounts of money on Basel II compliance, and consequently other projects have been sidelined. Industry analysts state that with the introduction of Basel II, the financial services market will expect to see the corporate credit scoring models and the retail credit scoring models to converge. How banks forge Win-Win - According to representatives from one regulatory institution, banks focus on two key points: they do not want to violate the law and want to extend the credit to as many customers as possible. An interviewee added that fair lending compliance and profitability are not always consistent.

Future trends

With organisations growing and consolidating in the marketplace, the banks need to address the internal controls as well as the external regulators. Another emerging trend is related to banks trying to attract and attend to niche or thin credit markets.