Current Expected Credit Loss (CECL)

Crisil provides banks around the world with support and automated modelling platforms for current expected credit loss (CECL).

For banks, implementing effective CECL models is just the first step in establishing a robust risk management platform for credit losses from loans, held-to-maturity (HTM) bonds, available for sale (AFS) bonds, lease, guarantees, trade receivables, and debt securities. Following implementation, CECL models require regular validation, monitoring, and internal controls. This in turn requires the development of sound governance processes to ensure that the new models are functioning appropriately and in line with their intended use.

Crisil supports banks through every step in that process. We help clients meet CECL compliance requirements by providing gap analysis, data management, model development and documentation, model validation, model monitoring and maintenance, business analysis, and testing services.

To do so, we leverage comprehensive knowledge of accounting standards such as FASB and International Accounting Standards Board (IASB), expertise in various regulations (like CCAR/DFAST, Basel norms), industry experience, industry-leading modelling platforms, proprietary templates, and execution-enabling frameworks.

Our automated platform for the validation and documentation of regulatory and non-regulatory models can reduce downtime by up to 40%. The platform is easy to integrate, modular and fully transparent.

Solution Components

Data Management

Development data quality, relevance, validity, sufficiency
Managing end-to-end data and standardisation
- Data sourcing validation
- Proxy/synthetic
- Data cleansing
- Data transformation
Definition of dependent and independent variables
Data reconciliation
Managing loan level vs. aggregate, borrower level, third party and macro data vs. gross level data
Adjust the vintages selected in historical data
Reasonable and supportable forecasts

Model Development

Data collection and portfolio analysis
Data cleansing, data preparation and data quality checks
Computing life of the loan for various portfolios
Life time loss modelling using PD/LGD/EAD based approach, roll rate, loss rate, aging and DCF approach
Leveraging existing models
Model performance testing
Sensitivity analysis and ongoing monitoring plan

Model Validation

Performing conceptual soundness like scoping, analyzing model theory, model assumptions and highlighted limitations
Validating approach, life of loan, segmentation/sampling and variable selection
Model estimation, challenger and benchmarking model development
Model output and result analysis
Model performance testing
Backtesting and sensitivity analysis
Validated GEN1 CECL models in 2017-2018

Model Monitoring and Maintenance

Designing and reviewing policies
Model performance testing and override analysis
Stability analysis
- Risk Parameter (PD/LGD/EAD) monitoring
- Variable sensitivity analysis
Benchmarking
Trend analysis
Backtesting
Handling breaches or escalation resolutions
Reporting & Documentation

ECL Industry Benchmarking

Access to industry benchmarks (e.g. ECL basis) vis-a-vis industry peers
Summary statistics, trend analysis, variance and impact analysis at risk component level
Reporting and documentation

Comprehensive Automation of Documentation

Compliance with internal requirements
Documenting theoretical framework, model assumption and limitations, model methodology, variable selection, assumption testing, results and output discussion etc.
Inbuilt documentation templates (e.g., Model development documentation; Model validation report; Vendor contingency plan; Ongoing model monitoring)
Existing documentation review and remediation

Financial and Regulatory Reporting

Regulatory Interpretation
Regulatory Submissions
Regulatory Audits
Reconciliations
Supporting with playbooks and white papers for smooth regulatory submission

Crisil EDGE

Proprietary data collection, model development, validation and monitoring frameworks

Deep domain expertise in Credit Risk covering different portfolios (e.g. mortgages, credit cards, personal loans etc.), models (e.g. PD, LGD, EAD, etc.), risk data aggregation and reporting

Automated and integrated CECL components including model development, validation and documentation can reduce downtime by almost 30% - 40%

Extensive experience and strong expertise in validation, review and documentation of CECL models, led by senior credit risk experts