Vintage analysis and data collection
Oct 24, 2016
Vintage analysis is a method of evaluating the credit quality of a loan portfolio by analyzing net charge-offs in a given loan pool where the loans share the same origination period. It allows the financial institution to calculate the cumulative loss rates of a specific loan pool, thereby determining the loan pool’s lifetime expected loss experience.
The method is widely used in the analysis of retail credit card and mortgage portfolios, but it is also one of several methodologies financial institutions are using for the current expected credit loss (CECL) model.
The vintage methodology under CECL measures the expected loss calculation for future periods based on historical performance by the origination period of loans with similar life cycles and risk characteristics. It’s advantageous to pool similar loans that follow comparable loss curves that may be predictive for future periods
This method is not overly complicated, but its expanded data retention and supporting factor requirements could necessitate changes to current workflows and practices. Banks and credit unions will need a way to document all final reserve calculations, including qualitative and environmental factors applied to quantitative reserve amounts. This process could take time and may compel a best practice of collecting data and implementing process changes ahead of CECL implementation.
Many financial institutions are seeking guidance on determining how far back to go when conducting their vintage analysis. Neekis Hammond, Managing Director of Advisory Services at Abrigo, says the more time the better.
“You can never go too far [back] in a vintage, that’s the nice thing,” Hammond said. “If you’ve got a five-year pool and you run a 10-year vintage analysis, those final five years will just yield 0 percent loss experience.”
However, financial institutions can conduct a vintage analysis over a window of time that is too narrow, Hammond said. “The way you’ll know right away if you’re too short is, say you run a 3-year vintage analysis; you’re going back from 2012 to 2015, and you find that you still have a lot of dollars in your 2012 originations. That would indicate that your vintage analysis period is too short and you need to lengthen it.”
When figuring out an appropriate lookback period for determining life-of-loan losses, it should ideally capture a full economic cycle. Extending the lookback beyond a complete economic cycle could skew the historical data positively or negatively.
One advantage of understanding potential methodologies to use under CECL and to understanding segmentation strategies and measurement techniques as soon as possible is that financial institutions may be able to identify data gaps early enough to reduce those gaps, Hammond said. The vintage methodology is just one of many methodologies that could be best used under CECL. One big advantage is that it best illustrates the life-of-loan concept of CECL.