Useful journalism to understand, manage and grow a business.
SearchDelinquency, default, losses, recoveries, and concentration describe distinct aspects of credit risk. This guide explains what each indicator measures, what questions to ask when interpreting it, and why it is useful to analyze its evolution by vintage.
Credit metrics help track the health of a portfolio, but no single metric tells the whole story. A rate can change because payments have worsened, the composition of loans has shifted, or the accounts being observed have been active for longer. To interpret the signals, risk, finance, and management teams need to understand what each indicator measures, what its denominator is, and what period it covers.
Days past due (DPD) indicate how many days have passed since a missed payment's due date. Categories of 30, 60, or 90 days past due make it possible to observe the severity of delinquency and distinguish recent delays from more prolonged delinquencies. The thresholds and treatment of each account should be defined consistently in internal reports.
A delinquency rate expresses the proportion of accounts or balance that exceeds a delinquency threshold. It can be calculated, for example, based on the number of loans or on the outstanding balance; these approaches answer different questions. When comparing results, it is useful to check that the same threshold, denominator, and period are being used.
Useful questions: Is the rate calculated by account or by balance? Does it include all products? Is the increase concentrated in recent delinquencies, or does it also extend to more severe delinquency stages? Has the classification criterion changed?
The default rate measures the proportion of borrowers or exposures that meet a definition of default during a given period. That definition—for example, reaching a certain number of days past due—must be made explicit: the measure is not comparable if each product or report uses a different criterion.
The charge-off rate reflects the balance that the creditor records as a loss in its accounts, in accordance with its policies. It does not necessarily mean that the payment obligation has disappeared, nor does it indicate, by itself, how much money will be recovered later. It is therefore useful to distinguish observed default from accounting treatment and subsequent recoveries.
The loss rate summarizes losses relative to a defined base, such as originated balance or exposure. Its interpretation depends on which losses it includes, which recoveries it deducts, and what time horizon it uses. The rules should be consistent when comparing products.
The recovery rate shows what portion of a defaulted debt is recovered through payments or other means considered by the entity. For the indicator to be interpretable, it is necessary to clarify whether it is measured against the defaulted amount, the charged-off amount, or another base, and how long recoveries are tracked.
Loss and recovery are connected, but they are not the same indicator: one describes the net or gross outcome, depending on its definition; the other describes amounts recovered under a specific criterion. It is useful to review both alongside the time elapsed since default, because a figure observed over a short window might not reflect subsequent recoveries.
Concentration risk arises when a significant portion of the portfolio depends on a small number of borrowers or is grouped in segments exposed to common factors. Concentration can be examined by customer, product, sector, region, or other categories relevant to the business.
A concentrated distribution does not, by itself, prove that the portfolio will incur losses. It does help identify whether a common problem could affect multiple exposures at once. To interpret the figure, it is necessary to look both at the proportion of balance in each segment and at the quality and evolution of its payments.
Useful questions: Which segments account for the most balance? Has concentration increased compared with previous periods? Do the exposures share conditions or risk factors? Does delinquency evolve differently across segments?
Vintage analysis, also called cohort analysis, groups loans according to the period in which they were originated and tracks their performance as they age. Instead of comparing the entire portfolio on a given date, it makes it possible to observe how different cohorts evolve after a comparable number of months since opening.
The age of each cohort is usually expressed as months on book (MOB). In a vintage table or curve, each row or line represents an origination period; the columns or horizontal axis show the age of the loans. The indicator being tracked may include, among others, delinquency by stage, the cumulative default rate, or cumulative loss.
This view helps identify whether recent cohorts are behaving differently and observe when defaults or losses accumulate. However, newer cohorts have had less time to demonstrate their performance: comparing them with mature cohorts without adjusting for age can lead to misleading conclusions. Changes in credit policies, customer composition, or economic conditions can also have an impact.
When choosing an observation window, it is useful to review how the metric evolves with age and whether its change is moderating. A curve that continues to rise should not be considered stable just because a cohort has reached a certain age; the appropriate window depends on the product, observed behavior, and purpose of the analysis.
Credit models use three concepts that help break down potential loss:
These estimates are not observed delinquency rates: they are components of a modeled assessment and depend on data, assumptions, and definitions. When presenting them, it is important to document the horizon, population, and calculation criteria.
To turn metrics into a useful assessment, teams can compare the overall trend with the details by product, segment, and cohort. A stable aggregate rate, for example, does not allow us to conclude that all segments are evolving in the same way; nor is a new cohort with little time on book directly comparable to another whose performance has been observed for longer.
Before communicating a change, it is useful to verify the definition, denominator, period, and data maturity. This discipline makes it possible to distinguish real changes in portfolio behavior from differences in composition, measurement, or observation time, without attributing more to a single indicator than it can demonstrate.
A partnership can accelerate innovation, but it also increases operational and compliance dependencies. This guide proposes evaluating the strategy, financial soundness, controls, technology, and responsibilities before signing and throughout the partnership.
Read story
A practical guide to turning quarterly goals into prioritized tasks, owners, deadlines, and reviews—without scheduling every hour or leaving the team no room for unexpected events.
Read story
A useful calendar is not just a collection of dates: it brings together the obligations that actually affect the business, assigns responsibilities and documentation, and includes reviews to adapt to changes.
Read story