Kreditmodeller

Kreditmodeller are quantitative tools used by banks and financial institutions to estimate credit risk. They quantify the probability that a borrower will default, the exposure at default, and potential losses, enabling risk-based decision making in lending, pricing, provisioning, and capital allocation. Kreditmodeller can be applied at the borrower level (credit scoring and rating models) and at the portfolio level (portfolio credit risk, expected loss).

Modellerne bygger ofte på key concepts such as probability of default (PD), loss given default (LGD) and exposure at default (EAD). PD estimates the likelihood of default over a chosen horizon; LGD estimates the portion of exposure that would be lost if default occurs; EAD estimates the outstanding balance at the time of default. Outputs are typically risk scores, credit ratings, or probabilistic loss metrics. Common methods include traditional statistical models such as logistic regression and survival analysis, as well as machine learning approaches like gradient boosting and random forests. Data inputs include borrower characteristics, historical default data, and macroeconomic indicators. Validation involves backtesting, out-of-sample testing, and recalibration to maintain accuracy and stability.

Lifecycle and governance are central to kreditmodeller. They undergo development, independent validation, and ongoing performance monitoring within an established model risk management framework. Models are documented, versioned, and regularly reviewed, with performance tracked through metrics such as discrimination, calibration, and realized losses. Effective governance also includes risk ownership, auditability, and clear escalation procedures for model updates or replacement.

Regulatory context and limitations: kreditmodeller support capital adequacy frameworks (e.g., Basel) and impairment accounting (e.g., IFRS 9). Limitations include data quality, non-stationary economic conditions, model risk, and potential overfitting. Continuous monitoring and governance are essential to mitigate these risks.