Credit Scoring "Reduces Risks"

Discussion in 'Credit Talk' started by roni, Sep 23, 2001.

  1. roni

    roni Well-Known Member

    Customized credit scoring helps manage risk.
    PARTNERSHIP; COUNTER Intelligence Associates (Company); BETHPAGE Federal Credit
    Union; CREDIT unions; EXPERIAN Inc.
    Credit Union Magazine, Jul2000, Vol. 66 Issue 7, p45, 3p
    Meli, Jeff
    Focuses on the partnership formed by Bethpage Federal Credit Union with Experian and
    Counter Intelligence Associates to build a credit scoreboard that will increase approval rates and
    reduce delinquencies and losses. Development and design of the credit scoreboard solution;
    Implementation of the credit scoreboard.
    Business Source Elite
    Click here to mark for print.

    [Go To Citation]




    Bethpage Federal says it boosts loans and lowers delinquencies

    Changing market conditions continually challenge credit unions to adopt better technologies and strategies to
    serve members while minimizing overall risk. One effective tool is credit scoring.

    Many credit unions have gone beyond traditional generic scores and developed custom scoring systems based
    on their own applicant pool. When developed and implemented to suit the lending environment, and then
    monitored effectively, these systems provide significant benefits.


    Bethpage (N.Y.) Federal Credit Union is the state's fifth-largest credit union, with more than 93,000 members
    and $980 million in assets. Through the mid-1990s, Bethpage Federal experienced a decline in its primary
    membership as a result of downsizing and layoffs. Meanwhile, changing competitive conditions encouraged the
    credit union's more creditworthy members to go elsewhere for better rates, leaving a disproportionate
    percentage of less-creditworthy members.

    The credit union responded by incorporating new sponsor groups, but realized some critical improvements were
    necessary to improve its loan portfolio, boost loan application volume, and stem the rising tide of delinquencies.
    To remain competitive, the credit union needed to adjust its loan pricing, streamline the loan-origination process,
    and better differentiate members of varying creditworthiness.

    The credit union called on Experian in Orange, Calif. (714-385-5827;, and Counter
    Intelligence Associates in San Juan Capistrano, Calif. (800-424-4951; to create a solution to
    tackle two seemingly conflicting objectives: increasing approval rates while reducing delinquencies and losses.
    The credit union determined a custom scorecard built on its own portfolio data would be the optimal solution.


    Bethpage Federal worked with Experian to design a sample of applications that best reflected its applicant
    pool. The first step was to construct definitions of bad and good performance based on the credit union's
    experience. Portfolio reports were then generated for each product incorporated into model development.
    Based on the performance definitions and portfolio counts, the credit union selected a representative sample of
    the applicant pool for model development, excluding applicants or product types that wouldn't be scored by the
    model when implemented. All applicants in the sample fell within a time period recent enough to reflect the credit
    union's current applicant pool, yet mature enough to meet aging requirements for good accounts and to
    incorporate a sufficient sample for modeling.

    As is typical for applicant scoring systems, Bethpage Federal included application, credit bureau, and
    performance information in the sample. All application and credit bureau information was collected from the date
    of the loan decision, while performance information was collected from throughout the life of the loan. Experian
    assessed the sample for data integrity, weighted the sample counts to reflect the actual portfolio, and used
    inferencing techniques to assign performance to the declined accounts. The sample was then divided into an
    analysis sample and an independent test sample. A preliminary model was developed based on the analysis
    sample, and the model was then validated on the test sample to ensure effectiveness on an independent

    The intent of the preliminary model development was to build the strongest statistical model. Even the best
    statistical solutions, however, can fail if they're not logically valid, operationally effective, or easy to implement
    and use.

    To ensure the solution would function effectively, the credit union and Experian discussed the model's
    components and fine-tuned it. All desired model adjustments were tested for their impact on model
    performance, allowing the credit union to weigh the importance of a modification against its effect on the model.
    This process occurred in an interactive fashion until all desired changes were made. Only when both parties
    were satisfied by the result was the model finalized.


    Bethpage Federal obtained the model specifications from Experian and initiated the implementation process
    with its application processor. Once the model was programmed, Experian audited the model and its individual
    characteristics to ensure compliance with the specifications.

    Meanwhile, the credit union set minimum scores in line with model performance forecasts Experian provided.
    Applicants scoring in the upper ranges of the scorecard received automatic approvals, dramatically increasing
    loan-processing efficiencies. The credit union also conducted training and user acceptance sessions to prepare
    the credit union for the scoring model's introduction. Once auditing and training were complete, the model was
    ready to go live. The credit union implemented the scorecard in August 1998, and later worked with Counter
    Intelligence Associates to introduce a risk-based pricing program.

    From the date of implementation, the credit union has been dedicated to achieving maximum benefit from the
    scorecard, both from a loan volume and a risk perspective.

    On a quarterly basis, the credit union sends Experian a data file containing all accounts booked since scorecard
    implementation. Each of the variables in the scorecard, along with a worst-performance indicator, is included.
    Experian then monitors trends in population stability, scorecard adherence, and scorecard effectiveness. If any
    red flags appear, Experian alerts the credit union and suggests alternative credit underwriting approaches.


    As of the first quarter of 2000, the scorecard is performing well. As anticipated, the credit union has seen some
    minor shifts in the overall population distributions, but the model has met or exceeded expectations.

    Since scorecard implementation, loan approval rates have increased from 60% to 65%, and the credit union's
    loan-to-share ratio has grown from 47% to 51%. Approved loans that scored below the cutoff show
    delinquency rates about four times those of approved loans that scored above the cutoff. Delinquency rates of
    loans that have been on the books at least one year are in the 1% to 2.5% range on the model development
    population. This percentage is expected to increase as these accounts age, but early indications are promising.

    Overall, the investment in a custom scorecard has already paid dividends for Bethpage Federal. "Our new
    applicant scorecard is an effective tool that has enabled us to improve turnaround time for applicant decisions,
    offer convenient 24/7 loan approval, increase our approval rates, and reduce losses," says Kirk Kordeleski, the
    credit union's president/ chief executive officer. So long as no major economic shifts occur in the near future, the
    scorecard will probably perform at this level for many quarters to come.

    When the scorecard first exhibits a measurable and sustained drop in performance, the credit union will consider
    a model adjustment. Because Experian already receives data on a quarterly basis and the credit union has the
    model components programmed, any model updates should be accomplished quickly.


    By Jeff Meli

    Jeff Meli is a project manager in the custom analytical solutions group of Experian Strategic Solutions' Atlanta
    office. He can be reached at 404-841-1436 or at

    Copyright of Credit Union Magazine is the property of Credit Union National Association, Inc. and its content may not be
    copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However,
    users may print, download, or email articles for individual use.
    Source: Credit Union Magazine, Jul2000, Vol. 66 Issue 7, p45, 3p.

Share This Page