COMMENTARY · NPL & DISTRESSED CREDIT
NPL recovery rate: why vintage matters.
Applying average historical recovery rate to an NPL portfolio is the recurring shortcut. And the most fragile one when the operation reaches the auditor or the committee.
COMMENTARY · NPL & DISTRESSED CREDIT
Applying average historical recovery rate to an NPL portfolio is the recurring shortcut. And the most fragile one when the operation reaches the auditor or the committee.
NPL (non-performing loan) portfolio valuation is one of the most active engagements of the moment. Banks divest, structured funds buy, distressed advisors evaluate. The question that defines value is one: how much, and when, will be received from this portfolio of credits in default.
The easy answer is to apply average historical recovery rate. A portfolio of R$ 100 million face value × 25% historical recovery = R$ 25 million expected recovery. Discount to present value, mark fair value, ready.
The problem is that the average recovery rate is a statistic that hides what matters.
Vintage is the year (or quarter) of credit origination. Each vintage carries the imprint of its origination context: prevailing interest rate, underwriting standards of the period, sector of the cedent, judicial recovery legislation in force.
A 2018 vintage in retail credit behaves differently from a 2022 vintage in the same portfolio. Same product, same originator, same channel: different recovery curve. Applying average recovery to a mixed portfolio is mixing statistics that should not be mixed.
Average recovery distorts the value in three ways:
Distortion by stratum. Vintages with different proportions in the portfolio receive equal weight in the average, when the manager has clear visibility of which strata have accumulated default.
Distortion by collection horizon. Recovery does not occur uniformly in time. Vintage A may have already realized 60% of total recovery in three years. Vintage B may still recover 80% over five years. Average abstracts this difference and treats them as comparable.
Distortion by external scenario. Recovery depends on legal/judicial environment of the period. Pre-2020 vintages behaved differently after pandemic adjustments. Vintages originated under new bankruptcy law have different behavior.
"The auditor of a structured fund or distressed advisor will ask, on the first opportunity, the recovery curve per vintage. Failing to be able to present it is the operational signal that the valuation has not done the homework."
The defensible vintage report does three things. Segments the portfolio by origination year and product/segment. Builds the recovery curve of each cohort (% recovered over months). Applies these curves to remaining portfolio balance, instead of multiplying by simple average.
The result is a more precise present value: a year-1 curve in vintage A worth 30% more in PV than year-5 curve in vintage B, even if both promise the same total recovery.
It is more work. It is also the difference between a defensible report and a report that explodes on first auditor review.