Vintage Analysis for Gold Loans
| Financial Services
Key Highlights
- Purpose: Vintage analysis tracks a cohort of loans disbursed in the same month over its life, isolating underwriting quality from portfolio mix and seasonality.
- Signal: A worsening vintage curve for recent disbursement cohorts is one of the earliest, cleanest signals of underwriting or valuation drift - earlier than the headline gold auction ratio.
- Segmentation: Cutting vintage curves by branch, LTV band, and ticket size pinpoints exactly where quality is slipping, not just that it is.
- Pitfall: Recent cohorts are "immature" and cannot be fairly compared to older, fully-seasoned cohorts - a common analytical mistake.
What Vintage Analysis Is and Why Gold Loans Need It
Vintage analysis groups loans into cohorts by the month they were disbursed, then tracks a chosen outcome - auction rate, rollover rate, delinquency, or closure - for each cohort as it ages, month by month. Instead of asking "what is our auction ratio right now," which blends loans of every age and origination quality into one number, vintage analysis asks "how is the January 2026 cohort performing at month 6, compared to how the January 2025 cohort performed at month 6." That comparison is what makes it possible to isolate genuine underwriting or valuation drift from portfolio mix, seasonality, or the gold price cycle - the same distortion our branch KPI and branch economics posts warn against when reading AUM growth at face value.
Building a Vintage Curve
The mechanics are straightforward but require clean, branch-level disbursement data: group every loan by disbursement month into a cohort, then at each subsequent month-end calculate the chosen metric - most commonly cumulative auction rate or rollover rate - as a percentage of that cohort's original disbursed value. Plotting each cohort as its own line, with months-on-book on the x-axis, produces a family of curves that can be compared directly. A cohort whose curve sits meaningfully above prior cohorts at the same age is underperforming; a cohort that sits below is outperforming - and because every curve is measured against its own age, not calendar time, the comparison is fair even as the book grows.
What a Healthy Gold Loan Vintage Curve Looks Like
| Months on Book | Typical Cumulative Auction Rate (Healthy Cohort) | What's Happening |
|---|---|---|
| 1-3 months | Near 0% | Standard tenure loans are still current; auctions this early usually flag origination-level red flags |
| 4-6 months | 0.1-0.3% | First wave of genuinely stressed accounts begins to surface |
| 7-9 months | 0.3-0.6% | Curve should be flattening, not accelerating, if underwriting is sound |
| 10-12 months | 0.5-1.0% | Most healthy cohorts approach a plateau near their terminal auction rate by this point |
| 12+ months (terminal) | <1.5% for a disciplined book | The number that ultimately matters for provisioning and capital planning |
These bands are indicative starting points, not universal benchmarks - every NBFC should build its own baseline from its historical cohorts and treat deviation from that baseline, not an external number, as the real signal.
Reading Vintage Curves to Catch Underwriting Drift Early
The single most valuable use of vintage analysis is early warning. If the cohort disbursed three months ago is already tracking above where the cohort from a year ago was at the same age, that is a strong, statistically grounded signal that something changed in underwriting, valuation, or customer selection in the intervening period - well before it would show up in the company-wide gold auction ratio, which by definition lags because it blends immature and mature cohorts together. This is also the analytical backbone behind sound incentive design: paying incentives on a cohort before its vintage curve has had time to mature is exactly how volume-driven incentive plans end up rewarding loans that later default or get auctioned.
Segmenting Vintage Curves by Branch, LTV Band, and Ticket Size
A single company-wide vintage curve tells you whether something is wrong; it does not tell you where. Cutting the same analysis by branch isolates whether drift is company-wide (a policy or gold-price problem) or localized (a specific branch manager or appraiser problem) - directly complementing our branch productivity metrics framework. Cutting by LTV band at origination confirms whether loans originated near the regulatory ceiling are disproportionately responsible for deterioration, and cutting by ticket size often reveals whether growth into larger, unfamiliar ticket sizes is being underwritten with the same discipline as the core small-ticket book.
Common Pitfalls
The most common analytical mistake is comparing an immature recent cohort's early-months curve directly against a fully-seasoned older cohort's terminal number - the recent cohort simply has not had time to develop the same cumulative auction rate, and this apples-to-oranges comparison either falsely reassures or falsely alarms. The second is failing to adjust for the gold price cycle: a period of falling gold prices will mechanically push more loans toward or above their LTV threshold across all cohorts simultaneously, which needs to be separated from genuine underwriting-driven deterioration before conclusions are drawn.
Key Takeaways
Vintage analysis is the earliest, cleanest lens for catching gold loan underwriting drift - well before it shows up in headline portfolio metrics. Build cohort curves by disbursement month, compare each cohort fairly against prior cohorts at the same age, and segment by branch, LTV band, and ticket size to pinpoint exactly where quality is slipping.
- Cohort discipline: Always compare cohorts at the same age, never calendar-time snapshots
- Early warning: Vintage drift surfaces underwriting problems months before company-wide metrics do
- Segment to localize: Branch, LTV band, and ticket size cuts show exactly where the problem sits
- Watch for gold price effects: Separate commodity-cycle noise from genuine underwriting drift
Technovative Consulting builds vintage analysis, cohort tracking, and PAR reporting pipelines for gold loan NBFCs. Learn more about our data pipelines and portfolio analysis services.