Model expected credit loss for CECL with MATLAB
Current expected credit loss (CECL) is a new standard created by the Financial Accounting Standards Board (FASB) for regulating credit loss reporting of financial institutions. It governs balance sheet accounting of loans, mortgages, and other credit instruments and is due to come into force in 2021, primarily affecting U.S. reporting. CECL incorporates a credit impairment charge on affected assets, resulting in higher provisioning expenses.
Expected credit loss (ECL), a main output for the IFRS9/CECL workflow, is a probability-weighted estimate of credit losses during the expected life of a financial instrument. With the new proposed CECL standard, the estimation method requires point-in-time (PIT) projections of probability of default (PD), loss given default (LGD), and exposures at default (EAD). CECL model types available with MATLAB® include:
- Stochastic modeling of default and recovery
- Macroeconomic modeling and forecasting
- Scenario generation
- Instrument pricing and risk sensitivity
- Automated reporting, incorporating PIT model and data selection
For more detail on doing CECL with MATLAB, see Risk Management Toolbox™, Statistics and Machine Learning Toolbox™, Econometrics Toolbox™, and MATLAB Report Generator™.
Examples and How To
Software Reference
portfolioECL
: Compute the lifetime ECL at the individual or portfolio level – Function- Expected Credit Loss Computation - Documentation
- Overview of Lifetime Probability of Default Models – Documentation
- Model Loss Given Default - Documentation
- Credit Simulation Using Copulas - Documentation
creditscorecard
: Createcreditscorecard
object - Function- Binning Explorer: Bin data and export into a
creditscorecard
object - Documentation - Generalized Linear Models - Documentation
See also: econometrics and economics, Monte Carlo simulation, credit scoring model, risk management solutions, IFRS 9, expected credit loss, Basel IV, fraud analytics