A Condensed Look Inside the Credit Scoring Industry

#creditscoring #analytics #machinelearning #bigdata

Key Terminologies

Credit risk Scoring is a tool used to evaluate the level of credit risk associated with applicants or customers providing statistical odds, or probability that an applicant with any given score will be good or bad.
Credit scoring offer a quick and proven way to use the data , not just for reducing losses but also lead to greater profitability.
Credit Score is a digit (commonly 3 digits ranging from 200-900 depending on institution) that summarizes a customer's credit risk based on their credit report and it predicts how you manage your credit(debt) and it helps the lenders to assess the rusks upon lending.
Scorecard is a management tool that generates credit score of an applicant based on their creditworthiness. It consists of a group of characteristics, statistically determined to be predictive in separating good and bad accounts. It can be developed In-house or outsourced.

scorecards can also be defined based on the type of data used to develop them.

Custom scorecards are those developed using data for customers of one organization exclusively e.g., if a bank uses the performance data of its own customers to build a scorecard to predict bankruptcy

Generic or pooled data scorecards are those built using data from multiple lenders. e.g., few banks which did not have enough data to build their own custom scorecards, they decide to pool their data for auto loans , build a scorecard with that data and share the scorecard or customize the scorecards based on unique characteristics of their portfolios.
Also scorecards build using industry bureau data and marketed by credit bureaus such as [(https://creditinfo.co.ke/)], are types of generic scorecards.

Factors driving the increased use of scorecards

Increased regulation.
Availability of greater educational material and training for developers
Ease of access to sizable and reliable data
Corporate knowledge management fostering retention and sharing of subject-matter expertise.
Efficiency and process improvement.
Creating value and boosting profitability.
Improved customer experience.

Examples of Scorecard Characteristics

Demographics (e.g. age, postal code)
Existing relationship (e.g. payment performance, product type)
Credit bureau (e.g. inquiries, delinquencies) and so on.

Management reports produced during scorecard development

Gains chart It tells us the expected performance of the scorecard. An example of gains chart

The above gains chart exhibits;

The score bands which are arranged so that there are approximately 10% of accounts in each bucket.
The marginal event rate (marginal bad rate) rank orders from a minimum of 0.2 percent to a maximum of about 15.7%. There is some variability between the bad rate based on counts and predicted bad rate from the model due to low counts.
For the score range 163-172, for example, the expected marginal bad rate is 5.31%. This means 5.31% of the accounts that score in that range are expected to be bad.
For all accounts above 163, the cumulative event rate is 2.45%. This would be the total expected bad rate of all applicants above 163.
If we use 163 as a cutoff for an application scorecard, the acceptance will be about 70%.

Based on the outlined factors, a company can then decide , for example, to decline all applicants who score below 163, or charge them higher pricing in view of greater risk they present.

Strategies for high-risk applicants

These strategies will maximize revenue and minimize bad debt

Decline credit if the risk level is too high
assign a lower starting credit limit on a credit card or line of credit
Ask the applicant to provide a higher down payment or deposits for high ticket size loans such as mortgages or asset finance loans
Charge a higher interest rate on a loan
Charge a higher premium on insurance policies
Ask high risk applicants for further documentation on employment, assets or income

Despite of Risk scoring being used to evaluate levels of risks, they have also been applied in other operational areas , such as:

Streamlining the decision making process
Reducing turnaround time for processing applications through automated decision making
Evaluating quality of portfolios intended for acquisition through bureau-based generic scores.
Setting economic and regulatory capital allocation.
Forecasting
Comparing the quality of business from different channels

The amount of confidence in any scorecard or model must be based on both the quality and quantity of the underlying data and decision making strategies adjusted accordingly. Models are very useful when used judiciously, along with policy rules and judgement, recognizing both their strengths and weaknesses. The most accurate model in the world will not help if the financial institution chooses not to confirm any information from credit applicants or to verify identities. One need to be very realistic when it comes to using scorecards and not have an unjustified level of trust in them.