┌──────────────────────────────┐ │ Lending Decisions │ └──────────────┬───────────────┘ │ ┌───────────────────────┴───────────────────────┐ ▼ ▼ ┌─────────────────┐ ┌─────────────────┐ │ Application │ │ Behavioral │ │ Scoring │ │ Scoring │ │ (New Customers) │ │ (Existing Users)│ └─────────────────┘ └─────────────────┘
Determining how to adjust credit limits, marketing efforts, or collection strategies for existing customers based on their ongoing repayment habits. Key Methodologies
Fair lending is addressed, but the book lacks:
: Standard methods like logistic regression remain popular due to their transparency and ease of implementation. credit scoring and its applications by l c thomas hot
Thomas distinguishes clearly between different types of scoring:
impact consumer lending and requirements for stress testing portfolios. The University of Texas at Austin Diverse Applications of Scoring
[ New Applicant ] ──> Application Scoring (Risk Decision) │ ▼ [ Active Customer ] ──> Behavioural Scoring (Limit Adjustments / Marketing) │ ▼ [ Late Payments ] ──> Collection Scoring (Prioritize Recovery Actions) Application Scoring The University of Texas at Austin Diverse Applications
Traditional models treat default as a binary event. Survival analysis (Cox proportional hazards model, accelerated failure time models) treats default as a time-to-event problem.
One of the primary applications discussed is Application Scoring. This is the process used at the moment a customer applies for credit. By analyzing variables such as income, employment history, and past debt performance, models can estimate the risk of a new account. This objective approach minimizes bias and ensures that lending criteria are applied uniformly across a diverse applicant pool.
, co-authored by Lyn C. Thomas, Jonathan N. Crook, and David B. Edelman and published by the Society for Industrial and Applied Mathematics (SIAM) , stands as the definitive global blueprint for mathematical consumer credit risk management. Originally published in 2002 with a heavily expanded second edition in 2017, this foundational text bridges the gap between raw statistical theory and operational banking strategy. Professor Lyn C. Thomas, a world-renowned pioneer in operational research, systematically transformed retail lending from a subjective, qualitative guessing game into an objective, data-driven science. This is the process used at the moment
: Methods for measuring how well a scorecard discriminates between "good" and "bad" borrowers. Dynamic Modeling Markov chains survival analysis to model how a borrower's behavior changes over time. Regulatory Compliance : Guidance on how the Basel Accords
Often used by credit bureaus, these models look at a borrower’s overall credit history to predict future delinquency, often relying on massive data sets. B. The Scoring Process