The “roll” can go in both directions. It is more often referred to negatively as a forward roll. A forward roll is when an account that was delinquent in one 30-day period is still delinquent in the next period. A backward roll is when an account moves to a lower level of delinquency, such as when a customer pays down the account. For example, a forward roll is when an account that was delinquent at 30 days past due and is still delinquent when it is 60 days past due. An account at 30 days past due that was paid by the customer comes out of delinquency in a backward roll. Roll rate analysis is the most common modeling practice for loss forecasting and is done at the portfolio level.
Alternate name: Migration analysis, transition rate, transition probabilities, flow models
Roll rates are commonly used to analyze credit card portfolios. For example, if a credit card portfolio has $1,000,000 in delinquent accounts at 30 days and $600,000 at 60 days, then the roll rate would be found by dividing the balance of the current month by the balance of the previous month. In this case, the roll rate would be 6%.
How a Roll Rate Works
To find the roll rate, a lender will examine a portfolio of loans (which could be mortgages, credit cards, or any other number of financial products) and calculate how many (or how much of those) delinquent accounts roll from one 30-day period to the next. The roll rate can be calculated with the number of loans delinquent or by the dollar amount of loans delinquent. It can be expressed as a formula like this: Number of loans delinquent at 60 days / Number of loans delinquent at 30 days = Roll rate percentage For example, if there were 100 loans delinquent at 30 days, and 40 loans delinquent at 60 days, the roll rate would be 40/100, or 40%. It can also be calculated using the amount of loans delinquent: Amount of loans delinquent at 60 days / Amount of loans delinquent at 30 days = Roll rate percentage If there were $1,000,000 worth of loans delinquent at 30 days and $60,000 loans still delinquent at 60 days, the roll rate would be $60,000/$1,000,000, or 6%. After historical data is calculated on net roll rates in a segmented portfolio, future roll rates can be estimated, which can then be used to forecast losses for the next year. “The purpose of determining roll rates is to help banks estimate credit losses and future defaults,” economist and CoinMarketCap VP Shaun Heng told The Balance by email. “Roll rate analysis helps to estimate credit losses by keeping track of the number of credit holders rolling further into stages of delinquency,” Heng explained. “Creditors report late payments in 30-day periods, so it is pertinent for banks to know just what percentage of clients are moving from 30 days late to 60 days late, and so on. This can help banks to anticipate potential future losses due to defaults.”
Roll Rate vs. Vintage Loss Model
Another popular forecasting model banks and lenders use is the vintage loss model. In vintage analysis, however, segmentation is based on the origination month. The origination month is called a vintage. Analysis using the vintage loss model allows a lender to observe trends over time. Here’s a quick look at each:
