Vintage analysis
Vintage analysis follows loans originated in the same month or quarter and plots a measure such as the share 30 or 90 days past due by months on book, so cohorts are compared at the same age. A vintage is all loans originated in one period, and months on book counts from origination. Each vintage produces a curve, and results form a triangle because recent vintages have fewer observed months.
Example
For illustration, a January vintage has 1,000 loans, and 50 have reached 30 days past due by month on book 3. Its share at that age is 50 ÷ 1,000 = 5%; compare it with another vintage at month 3, not its month 1.
How a vintage curve is built
Set the observation points first, such as the end of each month after origination. At each point, calculate the selected delinquency measure for loans in that cohort that have reached that age. The plotted points form one curve for that vintage.
The full results form a triangle. A recent cohort has values only for its first few months on book, because later months have not happened yet. Older cohorts fill more of the chart.
Comparing vintage curves and reading changes
Compare vintages at the same months on book. If one curve is higher at that age, a larger share of that cohort has reached the selected delinquency measure by then.
A change in the curves around a period when policy or channel changed can flag a pattern worth examining. The curve does not by itself show that the change caused it. A growing book can also show a falling delinquency rate because recent loans have had less time to go bad.
Questions we get asked
What does “ever 30 days past due” mean?
It means a loan has reached 30 days past due at least once by the age shown. The measure is cumulative to that point, rather than limited to loans currently 30 days past due.
Can vintage analysis track 90 days past due?
Yes. A curve can show the share of a vintage that has reached 90 days past due at each month on book, just as it can show a 30-day measure.
What can a vintage curve tell you about the cause of delinquency?
A curve shows how the selected delinquency measure changes with age for a cohort, not why it changed. Treat a difference as a pattern to investigate, rather than proof of a cause.
Decisimo's free vintage calculator draws curves of 30 or 90 days past due by months on book from a loan list CSV, and the file stays in your browser.
