The implementation of IFRS 9 introduced a fundamental shift in the measurement of credit risk across financial institutions. Unlike the incurred-loss framework under IAS 39, IFRS 9 requires the estimation of Expected Credit Losses (ECL) using a forward-looking, probability-weighted approach that incorporates both current conditions and reasonable future forecasts. This change has heightened the importance of understanding how credit exposures deteriorate over time and how borrowers migrate across delinquency states.
Within this context, Roll Rate Analysis and Vintage (Cohort) Analysis have become essential analytical tools. Although both techniques examine borrower performance, they provide distinct yet complementary insights—one focused on short-term transitions between delinquency stages, and the other on long-term patterns of risk evolution within origination cohorts. Together, these frameworks underpin effective IFRS 9 modelling, Stage classification, and ongoing credit portfolio monitoring.
Why Roll Rates and Vintage Curves Matter Under IFRS 9
IFRS 9 requires banks to classify the exposures into Stage 1, Stage 2, and Stage 3, depending on whether credit risk has significantly increased since origination (SICR) and whether a default event has occurred.
To project expected lifetime losses, institutions need to understand:
- How accounts migrate across delinquency states
- How fast risk builds up after origination
- How macroeconomic conditions impact transitions
- Whether the portfolio quality is improving or deteriorating
This is where our two methods split the workload:
- Roll Rate Analysis is typically the engine for short-term prediction (Stage 1 / 12-month PD). It looks at how accounts “roll” from one delinquency bucket to another month-over-month.
- Vintage Analysis is the engine for long-term prediction (Stage 2 & 3 / Lifetime PD). It looks at how entire cohorts of loans behave over their entire lifecycle.
Roll Rate Analysis: A Transition-Based View of Risk
Roll rate analysis examines the probability that an account moves (or rolls) from one delinquency bucket to the next within a specific time period—typically monthly.
2.1 What Roll Rate Analysis Measures
Roll rate analysis quantifies the likelihood that credit exposures migrate from one delinquency state to another over a defined time interval, typically monthly. For each delinquency bucket—such as Current, 1–30 days past due (DPD), 31–60 DPD, 61–90 DPD, and 90+ DPD—roll rate analysis calculates the proportion of accounts that progress, cure, or remain stable within their respective categories.
This measurement captures critical borrower behaviour dynamics, including the probability that an account transitions from early delinquency to more severe stages, the likelihood of curing back to a non-delinquent status, and the stability of accounts that remain current. These transition probabilities form the foundation for understanding short-term credit deterioration patterns and are essential inputs for Point-in-Time (PIT) probability of default (PD) estimation under IFRS 9.
2.2 Why Roll Rate Analysis Is Important for IFRS 9
Roll rates feed directly into PD term structures and Stage allocation decisions. Following are the ways in which roll rates are incorporated in IFRS 9 modelling
- Stage 2 Identification (SICR):
Certain roll rates, such as a sharp increase in transitions from Current → 30 DPD, signal deteriorating credit quality and justify SICR triggers. - PD Term Structure Estimation:
Roll rates can be chained to estimate the probability that a Current account defaults within 12 months or over lifetime. - Forward-Looking Adjustments:
Roll rates are sensitive to economic changes. In downturns, 30→60 DPD transitions typically worsen faster than baseline PDs—ideal for macro-adjusted modelling. - Model Monitoring:
Significant shifts in roll rates indicate portfolio stress or operational issues (e.g., collection inefficiency, policy changes).
2.4 Best Practices in Roll Rate Analysis
Effective roll rate analysis requires careful data handling and methodological rigour to ensure that the resulting transition probabilities accurately reflect underlying credit behaviour. Key best practices include:
- Assess data quality and ensure consistent definitions:
Delinquency buckets, cure definitions, and default classifications must be applied consistently across all reporting periods to avoid distortions in transition estimates. - Exclude or adjust for artificially cured accounts:
Accounts cured through restructuring, write-off reversals, or policy-driven adjustments should be reviewed and, where appropriate, excluded to prevent inflating cure rates or understating deterioration. - Address volatility in small or heterogeneous portfolios:
Roll rates can fluctuate significantly when exposure counts are low. Applying smoothing techniques—such as rolling averages or pooling similar segments—helps produce more stable estimates. - Segment analysis appropriately:
Roll rate behaviour differs across products, risk tiers, geography, and origination channels. Segmenting exposures ensures that transition matrices are reflective of unique portfolio characteristics. - Monitor stability over time:
Abrupt changes in roll rates may signal operational issues, shifts in customer behaviour, or economic stress. Time-series assessment is essential for understanding trend movements. - Incorporate macroeconomic sensitivity:
Linking roll rates to relevant macro drivers (e.g., unemployment rates, interest rates, inflation) enables Point-in-Time calibration and ensures that PD term structures remain aligned with prevailing economic conditions.
Vintage Analysis: A Cohort-Based View of Portfolio Health
Vintage analysis – often referred to as cohort analysis—examines the performance of credit exposures grouped by their origination period. Each “vintage” represents a cohort of accounts booked within the same month, quarter, or year, and the analysis tracks how these exposures evolve over time. By observing delinquency, default, and loss outcomes across different months-on-book, vintage analysis provides a clear understanding of how risk materialises throughout the lifecycle of an account.
What Vintage Analysis Measures
Vintage analysis evaluates how credit risk develops from the moment an account is originated. For each cohort, key performance metrics—such as delinquency rates, default rates, cure rates, and loss amounts—are measured at various time intervals (e.g., month 1, month 3, month 6, month 12). This framework highlights the rate at which accounts deteriorate or stabilise as they age, allowing institutions to identify how early behaviour influences longer-term outcomes. By comparing vintages across different origination periods, the analysis reveals whether portfolio quality is improving, deteriorating, or remaining stable over time.
Why Vintage Analysis Is Crucial for IFRS 9
The staged approach of IFRS9 is all about detecting significant increase in credit risk (SICR). Vintage curves give a clear, intuitive, and robust way to identify this. Below are some of the uses of vintage analysis in the IFRS9 framework.
- Detecting SICR (Stage 2 Migration):
If newer vintages show faster deterioration compared to historical norms, this indicates a significant increase in credit risk. - PD Model Development:
Lifetime PD models often use vintage curves as a base, particularly for retail installment loans. - Monitoring Portfolio Quality:
Vintage deterioration can be caused by:
- relaxed underwriting
- macroeconomic downturns
- aggressive marketing to risky customers
- operational issues in collections
- Stress Testing and Scenario Analysis:
Vintage curves behave predictably across cycles, enhancing macro-overlay and scenario calibration.
Comparing Roll Rate vs. Vintage Analysis
Although both roll rate analysis and vintage analysis are integral to IFRS9 credit risk modelling, they offer distinct perspectives and serve different analytical purposes. Key differences include:
Focus and Analytical Perspective
- Roll Rate Analysis examines short-term transitions between delinquency states, highlighting how accounts move from one bucket to another on a monthly basis.
- Vintage Analysis evaluates long-term performance of exposures based on their origination cohort, capturing lifecycle deterioration patterns.
Underlying View of Risk
- Roll Rates provide a transition-based view, focusing on month-to-month borrower behaviour.
- Vintages offer a cohort-based view, tracking how credit quality evolves as accounts age.
Primary Use in IFRS 9
- Roll Rate Analysis is primarily used to derive PD term structures, inform Stage migration behaviour, and monitor near-term credit deterioration.
- Vintage Analysis is used to assess Significant Increases in Credit Risk (SICR), evaluate underwriting quality, and model lifetime PD based on historical cohort performance.
Granularity and Data Structure
- Roll Rates use cross-sectional snapshots of delinquency states, requiring detailed monthly status transitions.
- Vintage Analysis requires longitudinal tracking of each cohort across months-on-book.
Risk Insight Provided
- Roll Rates highlight early warning signals, such as worsening transitions into delinquency buckets or weakening cure rates.
- Vintage Curves reveal structural changes in portfolio quality, including early-seasoning risk or deterioration in recent originations.
Best-Suited Applications
- Roll Rate Analysis is best suited for portfolios where short-term borrower behaviour is highly predictive, such as credit cards or revolving products.
- Vintage Analysis is ideal for installment loans or products with clear ageing and seasoning patterns, such as personal loans or auto loans.
To summarize in a tabular form:
| Aspect | Roll Rate Analysis | Vintage Analysis |
|---|---|---|
| Grouping Basis | Current behavioural state (DPD) | Origination cohort |
| Focus | Short-term transitions | Long-term lifecycle patterns |
| IFRS 9 Use Cases | Stage 2 triggers, transitions, PD movement | Lifetime PD/LGD curves |
| Data Dependence | DPD history, payment patterns | Origination + lifecycle data |
| Platform Requirements | High compute for matrix generation | High storage for long cohorts |
| Strength | Behaviourally sensitive | Lifecycle-oriented and intuitive |
| Weakness | No direct lifecycle view | Cohort variability may distort trends |
Steps for Using in IFRS9 Modelling
For Roll Rate Analysis
- Build monthly transition matrices
- Smooth or seasonally adjust the transitions
- Link transitions to macro variables
- Chain transitions to derive PD term structures
- Validate stability and back-test predictions
For Vintage Analysis
- Segment cohorts by origination date
- Track delinquency, default, and loss metrics by months-on-book
- Compare vintages to identify deterioration or improvement
- Fit lifetime PD curves (Weibull, exponential, hazard-based)
- Apply scenario overlays for stressed environments
Conclusion
Roll rate and vintage analysis are indispensable analytical techniques within the IFRS 9 modelling lifecycle. They are more than just risk formulas; they are the blueprints for a credit risk data architecture. Roll Rate requires agile, fast math on recent data, while Vintage requires the heavy lifting of historical big data. Roll rates offer granular insights into month-on-month transitions that inform PIT calibration and PD term structure development. Vintage analysis, meanwhile, provides a stable framework for understanding long-term deterioration patterns and validating lifetime PD model behaviour. The integrated use of both methods supports defensible Stage allocation, strong model governance, and clear justification for forward-looking adjustments—ensuring that the institution’s ECL framework remains methodologically sound and compliant with regulatory standards.
