A credit committee reads the loan tape before it reads anything else. It tests cohort and vintage performance, loss curve shape, roll rates between delinquency buckets, and whether underwriting policy documentation matches what the tape shows actually happened.
Field completeness and what missing fields signal
Before a credit committee tests any performance metric, it tests whether the loan tape is complete. Missing origination dates, absent FICO or equivalent risk scores, blank delinquency status fields — each gap is a data quality problem, but it is also a signal about the originator's own systems and reporting discipline.
A tape with scattered gaps across several fields reads differently from a tape with one field missing consistently across every record, because the latter usually indicates the field was never captured rather than lost in a data extract. Committees ask which explanation applies before proceeding, since the answer changes whether the gap is fixable.
Field completeness is checked first because every downstream analysis — cohort performance, loss curves, roll rates — depends on having enough clean data to run it. A tape that fails this check is returned to the originator before any of the more analytical work begins.
Cohort and vintage analysis — do recent originations behave like older ones
Vintage analysis groups loans by the period in which they were originated and tracks each group's performance over time, which lets a committee compare how a loan originated eighteen months ago is performing at month twelve against how a loan originated this year is performing at the same point in its life.
The question this answers directly is whether underwriting standards have held steady, tightened or loosened over time. A recent vintage performing worse at an equivalent point in its life than an older vintage did is one of the clearest early warning signs a committee looks for, well before aggregate portfolio metrics would show any deterioration.
Cohort analysis, grouping loans by shared characteristics rather than purely by origination date, is used alongside vintage analysis to isolate whether a specific product, channel or underwriting change is driving a divergence rather than time alone.
Loss curve shape and where it should flatten
A loss curve plots cumulative losses against loan age, and for a stable, well-underwritten portfolio it should flatten at a predictable point in the loan's life, after which further losses accrue slowly if at all. A committee's first question is not the level of the curve but its shape — does it flatten where the asset class and product would suggest it should.
A curve that continues rising well past the point where comparable products typically flatten suggests either that losses are being recognised late, that underwriting has changed partway through the observation period, or that the portfolio genuinely has a longer loss-emergence profile than assumed, and each explanation carries a different implication for pricing.
Where the tape does not yet have enough seasoning to show the flattening point, static pool analysis of the earliest, most seasoned cohorts is used as a proxy for how the rest of the book is likely to behave.
Roll rates between buckets as the leading indicator
Roll rates measure the proportion of loans that move from one delinquency bucket to the next worse bucket over a given period — for example, what share of accounts thirty days past due become sixty days past due the following month. Because they measure a rate of transition rather than a stock of past losses, roll rates move before aggregate loss figures do.
A committee treats a deteriorating roll rate, particularly in the early buckets, as a leading indicator of future losses even where the current non-performing loan ratio still looks acceptable. By the time the headline loss ratio moves, the roll rate has usually already signalled the change several months earlier.
This is why roll rate trends, tracked monthly across at least the previous twelve months, are requested as a standard part of the tape rather than treated as a supplementary exhibit.
Reconciling underwriting policy against observed behaviour
An originator's stated underwriting policy sets out criteria — minimum income thresholds, maximum debt-to-income ratios, required documentation — and a committee tests whether the loan tape's actual characteristics are consistent with that stated policy having been applied, rather than accepting the policy document as evidence of practice.
Where a meaningful share of originated loans sit outside the stated policy parameters, that gap is investigated directly with the originator. It may reflect a documented and approved exception process, in which case the exception rate itself becomes a metric to track, or it may indicate the policy is not being consistently enforced.
This reconciliation is one of the more time-consuming parts of the review, because it requires cross-referencing individual loan-level data against policy documents rather than relying on portfolio-level summaries, but it is also where the largest discrepancies between stated and actual practice tend to surface.
Static pool analysis and why lenders prefer it to portfolio averages
Static pool analysis fixes a specific cohort of loans at origination and tracks its performance in isolation over time, without allowing new originations to dilute the picture. This differs from a portfolio average, which blends new and seasoned loans together and can mask deterioration in older vintages behind the appearance of a healthy blended metric.
A growing portfolio's blended delinquency rate can look stable or even improving purely because new, unseasoned originations dominate the denominator, while the older cohorts within that same portfolio are quietly deteriorating. Static pool analysis is the tool that isolates the older cohorts and shows their performance without that dilution effect.
Lenders prefer static pool analysis specifically because portfolio growth is the most common way underlying deterioration gets obscured, and a committee that relies only on blended metrics is, in a fast-growing originator's book, systematically looking at the wrong number.
Last reviewed September 2026
Start with the structure, not the pitch.
Tell us the transaction and the timetable. If it is not something we should run, we will say so.