What is the Vintage Methodology for CECL?

The Vintage Methodology under CECL (Current Expected Credit Loss) 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. There are a handful of characteristics you should look at when segmenting your loans with the Vintage methodology. The most important risk driver is that all loans share a common origination period. Contrasted to the cohort method, loans are only included in tracking historical losses in the period in which they originated. Upon renewal of a loan, a new vintage is created. Critical data elements needed to run Vintage include loan number, balance at origination, loan balance, maturity date, renewal date, and loss information. The loans in the pools for this methodology should have similar risk characteristics and can be sub-segmented by an optional risk driver, like risk rating, although many times pools will become too granular to use this optional driver. Loan pools should  have very similar weighted average lives because loss rates in year two of a three-year loan looks vastly different than year two of a seven-year loan. Vintage works well with indirect auto loans and other consumer loans, credit portfolios, etc. To determine loss rates with this methodology, start with historical loss rates for each vintage and examine any trends in recent vintage loss rates. Fill in loss rates for future periods based upon historical trends as well as factoring in any changes for current conditions and reasonable and supportable forecast periods where you anticipate these periods are different from historical. Depending on differences in the makeup of the vintages, different adjustment factors may be required for each vintage. Depending on forecasted conditions, adjustments could be either positive or negative. When future years are no longer reasonably forecastable, revert to adjusted historical averages. Re-evaluate your Q factors with shifts in the economic landscape. When figuring out [...]