WHAT WE DO / SOLUTIONS BY FUNCTION
Risk Management
Originations & Underwriting
ACQUIRE IDEAL CUSTOMERS IN A SMART WAY
80% of the default risk related to an account can be attributed to account characteristics at originations. This is why it is important to acquire the right customers. Targeting good customers for acquisition will lower the portfolio bad rate more effectively than working to decrease the number of bad customers. At acquisition, smart lenders work towards achieving several objectives, including enticing the good prospects to respond / apply, understanding the risk-reward trade-offs of targeted accounts, establishing a speedy, cost-effective approval process, assigning initial lines appropriately and minimizing fraud.
LatentView helps lenders achieve these objectives using an analytics-driven, decision management process for originations, pre-approvals and pre-screening processes.
Refresh / Develop Models: LatentView has the expertise to help create or refresh originations models. Our technical expertise includes both statistical and non-statistical methods, including Discriminant Analysis, Logistic Regression, Non-linear Regression Methods, Neural Networks, and Genetic Algorithms. Our consultants have experience in developing originations models for credit cards, auto lending and personal lines of credit, using application data, and characteristics obtained from credit bureaus. Once the models are developed, they are adjusted for reject inference using a variety of approaches. Finally, a variety of management reports are created to monitor and track model performance.
Segmentation: Key segments are identified, based on past knowledge of managers in marketing and risk management functions. Techniques such as decision trees or clustering can be used to identify key segments. Variables for key segments include demographic or behavioral characteristics obtained from credit bureau reports.
Strategy Formulation: Managers responsible for originations strategy formulation make decisions involving setting cut-offs, to maintain a given approval rate, a given bad/charge-off rate, set the most profitable level or a combination of the three. The strategies are tested for adverse impact on the key segments, especially those considered important to the client. Another key aspect of the strategy involves credit line assignments to make, on being approved. Each strategy has its own business objectives, such as profit and retention.
Implementation: The scoring system is implemented to automate accepts and rejects. The scores are reviewed to ensure scoring errors are caught. Implementation also involves specifying the high-side and low-side overrides to the decisions of the scoring system.
Champion / Challenger Testing: LatentView can assist in testing the new strategies against their expected business objectives, and existing “champion” strategies. For instance, an offer of a higher line of credit or an additional credit card product can be made to an applicant with higher affordability. To do this, test and control groups are formed, and the strategies are applied. The results provide a learning tool for identifying effective strategies.
Model Validation: LatentView can help in monitoring and tracking to ensure that scorecard objectives are being met. Monitoring includes evaluating the divergence between development and test samples through characteristic analysis or population stability reports. Tracking involves identifying relationships between the scores and the portfolio performance characteristics, across different sub-segments of the population.
LatentView’s approach allows an organization to observe the effect of new strategies, and offers the potential for meeting your unique needs to craft custom solutions that deliver measurable results.
Talk to us today to maximize the return on your analytic investments!
Risk Management
Originations & Underwriting ACQUIRE IDEAL CUSTOMERS IN A SMART WAY
80% of the default risk related to an account can be attributed to account characteristics at originations. This is why it is important to acquire the right customers. Targeting good customers for acquisition will lower the portfolio bad rate more effectively than working to decrease the number of bad customers. At acquisition, smart lenders work towards achieving several objectives, including enticing the good prospects to respond / apply, understanding the risk-reward trade-offs of targeted accounts, establishing a speedy, cost-effective approval process, assigning initial lines appropriately and minimizing fraud.
LatentView helps lenders achieve these objectives using an analytics-driven, decision management process for originations, pre-approvals and pre-screening processes.
Refresh / Develop Models: LatentView has the expertise to help create or refresh originations models. Our technical expertise includes both statistical and non-statistical methods, including Discriminant Analysis, Logistic Regression, Non-linear Regression Methods, Neural Networks, and Genetic Algorithms. Our consultants have experience in developing originations models for credit cards, auto lending and personal lines of credit, using application data, and characteristics obtained from credit bureaus. Once the models are developed, they are adjusted for reject inference using a variety of approaches. Finally, a variety of management reports are created to monitor and track model performance.
Segmentation: Key segments are identified, based on past knowledge of managers in marketing and risk management functions. Techniques such as decision trees or clustering can be used to identify key segments. Variables for key segments include demographic or behavioral characteristics obtained from credit bureau reports.
Strategy Formulation: Managers responsible for originations strategy formulation make decisions involving setting cut-offs, to maintain a given approval rate, a given bad/charge-off rate, set the most profitable level or a combination of the three. The strategies are tested for adverse impact on the key segments, especially those considered important to the client. Another key aspect of the strategy involves credit line assignments to make, on being approved. Each strategy has its own business objectives, such as profit and retention.
Implementation: The scoring system is implemented to automate accepts and rejects. The scores are reviewed to ensure scoring errors are caught. Implementation also involves specifying the high-side and low-side overrides to the decisions of the scoring system.
Champion / Challenger Testing: LatentView can assist in testing the new strategies against their expected business objectives, and existing “champion” strategies. For instance, an offer of a higher line of credit or an additional credit card product can be made to an applicant with higher affordability. To do this, test and control groups are formed, and the strategies are applied. The results provide a learning tool for identifying effective strategies.
Model Validation: LatentView can help in monitoring and tracking to ensure that scorecard objectives are being met. Monitoring includes evaluating the divergence between development and test samples through characteristic analysis or population stability reports. Tracking involves identifying relationships between the scores and the portfolio performance characteristics, across different sub-segments of the population.
LatentView’s approach allows an organization to observe the effect of new strategies, and offers the potential for meeting your unique needs to craft custom solutions that deliver measurable results.
Talk to us today to maximize the return on your analytic investments!
