Designing Auto Approval Matrices
The first step necessary to the design of an automated decision matrix is to determine:
1. What combination of application and credit bureau variables included in the Hybrid extract file are valid measures of risk? Which ones are not?
2. How can the credit score most efficiently apply the variables selected for the products that will be eligible to be automatically approved?
3. What percentage of applicants is expected to be auto approved in each score range?
4. How do regulatory and policy requirements affect the design of the decision templates?
Without the proper tools to do “what-ifs” and test the most efficient method of implementing the decision templates, lenders often rely on a best guess, or choose conservative strategies that artificially limit the number of applicants able to be approved or without other options simply adopt a template from a 3rd party and hope for the best
Auto Decision Approval Percentages
Repayment ratios are the basis of designing matrices to approve the greatest percentage of applicants in each range without unacceptable risk. The design for each template should include the limits and other lending guidelines.
Qwiklend Program
CFSD has available a modeling tool that we consider an important aid to the design and implementation of effective automated decision templates. Decision templates developed can be copied and integrated with loan origination software. The Qwiklend program uses the data base from your own paying loans so the accuracy and anticipated results can be accurately estimated and monitored.
Decision templates can be created for all major product segments for generic or hybrid scores.
Qwiklend Auto Decision Template
This is an example of a possible secured decision template using a hybrid extract file.
Click here to download the Qwiklend User Guide. I recommend printing it out and following along side by side
For best results, please use a desktop computer when using the program
Qwiklend Users Guide
STEP 1. Begin by selecting the type of loans to be analyzed from the drop down box. Loan types can include vehicles or other secured loans, unsecured or credit card loans and mortgages.
STEP 2. Select a primary credit score and risk ranges. Credit Score 1 should be your current generic score. Credit score 2, if used, can be any supplemental score. Most generic scores have a maximum of 850 points. Hybrid score maximums are 1000.
STEP 3. Enter the minimum score that you wish to use to define A, B, C, D and E classes of risk. The minimum risk scores must be in declining order. Class B for instance must have a minimum score lower than class A.
EXAMPLE SCORE
A = 751
B = 651
C = 501
D = 250
E = 0
Note: Scores that define risk categories may be different for each loan type. For example, minimum scores in A for vehicle loans may be 700 and A for lines of credit may be 750.
STEP 4. After entering the risk scores select the button “Calculate” The program will report the baseline or total number of loans in each risk category. The Number Approved and Percent of the Baseline Approved should begin at 100%.
STEP 5. Enter the Maximum Loan Amount permitted for each risk category.
STEP 6. Select the Minimum or Maximum limits for each of the criteria an applicant would be required to meet in risk category A in order to be qualified for auto approval .It is recommended you familiarize yourself with the program by initially entering qualifying criteria for the A class of risk only.
Select only a few underwriting criteria in your first matrix design for the A class of risk. Add or change the criteria values one at a time and “Calculate” to see the effect on Percent Approved.
To maximize the efficiency of the matrix, begin the testing of your decision matrix by using a minimal number of important underwriting criteria.
To adjust the number and percentage of auto approved applications revise and modify the criteria as wished and Select “Calculate”.after each criteria value is changed.
STEP 7. Review and print the results frequently.
STEP 8. Continue the testing until satisfied the Percent of Approved applicants in each score range meets your goal.
Reviewing the Auto Decision Results
It is recommended that templates designed with the Qwiklend program be copied and integrated into the auto approval section in your loan origination system. A report of the number and dollar amount of automatically approved loans should be published and analyzed monthly.
NOTE: The approval percentages calculated by the Qwiklend are based on the history of paying loans for each subgroup in your extract database and is an ESTIMATE of anticipated results.
To obtain an estimate and proposal for converting your current credit score lending program to a Hybrid scoring system contact sales@creditscorelending.com.