What is roll rate analysis?
Roll rate analysis measures how loans move between delinquency stages from one period to the next, usually month to month. A roll rate answers a simple question: of the accounts that were 30 days past due last month, what share are now 60 days past due? Repeat that for each bucket and you have a picture of how delinquency is flowing through the portfolio toward charge-off.
Retail lenders use roll rates on credit cards, auto loans and personal loans because they react faster than charge-offs. A rise in the share of current accounts rolling to 30 days shows up months before those accounts are charged off, which gives risk and collections teams time to act.
How do you calculate a roll rate?
A roll rate is a ratio with a clear numerator and denominator. The denominator is the number of accounts, or the balance, in a delinquency bucket at the end of one month. The numerator is the part of that group that sits in the next worse bucket at the end of the following month.
For example, the 30-to-60 roll rate equals the accounts that were 30 to 59 days past due last month and are 60 to 89 days past due this month, divided by all accounts that were 30 to 59 days past due last month. Always state whether you are counting accounts or dollars, because the two can tell different stories.
- Current to 30: share of current accounts that became 30 days past due.
- 30 to 60, 60 to 90, 90 to 120: share that rolled one bucket worse.
- Cure rate: share that paid back to current.
- Stay rate: share that remained in the same bucket, often through a partial payment.
Worked example: reading a roll-rate table
Example only, with made-up round numbers. A personal loan portfolio starts the month with 50,000 current accounts, 1,000 accounts 30 days past due, 400 accounts 60 days past due and 200 accounts 90 days past due. One month later the team counts where each group ended up.
- Current: 49,000 stayed current and 1,000 rolled to 30 days. Current-to-30 roll rate: 2.0%.
- 30 days: 450 cured to current, 150 stayed at 30 days and 400 rolled to 60 days. 30-to-60 roll rate: 40%.
- 60 days: 100 cured, 80 stayed and 220 rolled to 90 days. 60-to-90 roll rate: 55%.
- 90 days: 20 cured, 40 stayed and 140 rolled to 120 days. 90-to-120 roll rate: 70%.
- Flow to loss: if 90% of 120-day accounts are charged off, multiplying the forward rolls (2.0% x 40% x 55% x 70% x 90%) gives about 0.28% of current accounts reaching charge-off over the next five months, if these rates held.
How do roll rates differ from a migration matrix?
A roll rate usually tracks only the forward movement from one bucket to the next. A migration matrix, sometimes called a transition matrix, shows every possible movement: each starting bucket is a row, each ending bucket is a column, and each cell holds the share that moved there. That includes cures, accounts that skip a bucket, payoffs and charge-offs.
The matrix gives a fuller picture but is harder to read at a glance. Many teams watch the simple forward roll rates every month and look at the full matrix when something changes. Both are built from the same month-end snapshots, so the work to produce one supports the other.
What do changes in roll rates tell you?
Each roll rate points to a different part of the problem, so it helps to read them separately before combining them. Risk analysts, portfolio managers and collections leaders will often look at the same table and focus on different rows. Agreeing on what each movement usually means, and who acts on it, turns the table from a report into a tool for deciding what to do next.
- A rising current-to-30 rate means more borrowers are missing a first payment. Look at recent originations, payment shock or a servicing change.
- A rising 30-to-60 rate with a stable current-to-30 rate suggests early-stage collections are curing fewer accounts.
- Rising late-stage rolls, 90 to 120 and beyond, mean losses are becoming more certain and recovery strategy matters.
- Falling cure rates can signal weaker borrowers or changes in contact strategy.
- Seasonal patterns are normal. Many consumer portfolios see delinquency improve when tax refunds arrive and weaken after the holidays, so compare the same month across years before reacting.
What are the common mistakes in roll rate analysis?
Roll rates look simple, which makes errors easy to overlook. Small denominators are the first trap: a 60-to-90 rate based on 40 accounts can swing widely from month to month on noise alone, so show the counts and consider grouping months or segments when the base is thin. Mixing products with different behavior, such as cards and auto loans, in one table hides what is really happening in each.
Changes in operations can also distort the numbers. Re-aging, payment deferrals, hardship programs and changes to how days past due are counted all move accounts between buckets without any change in the borrower's actual condition. Finally, a shift in portfolio mix can move roll rates even when every segment is stable. Splitting roll rates by vintage, product or risk tier helps separate real deterioration from mix effects.
How do roll rates connect to charge-offs and collections?
Under the federal banking agencies' uniform retail credit classification policy, open-end loans such as credit cards are generally charged off at 180 days past due, and closed-end retail loans generally at 120 days. That makes the chain of roll rates a reasonable short-term forecast of charge-offs: today's 30-day accounts are a large part of the charge-offs several months from now.
Collections teams use roll rates to judge treatment. If a new call strategy for 30-day accounts raises the cure rate and lowers the 30-to-60 roll, the effect should show up in later buckets too. Measuring cure and repeat delinquency for each treatment, rather than just contact rates, tells you whether the work is actually changing outcomes.
How do you set up a monthly roll-rate report?
A reliable roll-rate report starts with consistent month-end snapshots of every account: delinquency status, balance, product, origination date and flags for hardship programs, deferrals or bankruptcy. Each month you match the current snapshot to the prior one by account and count the movements between buckets. Accounts that paid off, were sold or were charged off during the month need their own categories, so the movements add up to the starting count and nothing silently drops out.
Keep the report short enough to read every month. A useful layout shows forward roll rates and cure rates for the latest month, the same month last year and a trailing average, split by product. Show the counts behind every rate so readers can see when a percentage rests on a small base. Add a line for the implied flow to charge-off, and note any month where a policy or operational change, such as a new hardship program, affected the buckets.
How Aarvion helps
Aarvion Risk OS Retail Credit covers cards, auto and personal loans. Its performance explorer shows trends, roll rates and migration alongside vintages at equal months on book, with numerators and denominators shown so analysts can see when a rate rests on a small base. A portfolio overview tracks exposure, 30+ delinquency, net charge-offs and utilization.
Collections treatment outcomes, including cure and repeat delinquency, are tracked by treatment, and review cohorts are sized to analyst capacity. Investigations and approved monitoring plans keep follow-up organized, and every AI step is checked against the bank's own rules and recorded.
