Methodology

Panoperty collateral-monitoring methodology

How the Collateral Monitor turns a portfolio tape into local evidence and a ranked review queue, what each output means, and how evidence quality and provenance are recorded.

Portfolio snapshots

A lender supplies a portfolio tape. Each row is validated and either accepted or rejected with a stated reason. Row counts and exposure totals are reconciled before the snapshot is committed, and the reconciliation is stored with it. A committed snapshot is fixed: an identical file returns the existing snapshot, and a new tape creates a new one. The same collateral asset is linked across successive snapshots, so its evidence history continues.

Location

Monitoring is anchored on coordinates supplied in the tape for residential collateral in Ireland. Assets without usable coordinates are rejected with that reason and are not monitored. The location source is recorded with the evidence.

Local comparator evidence

Each month, for every located asset, Panoperty builds local comparator pools and measures buyer engagement and participation behaviour observed around comparable residential listings from proprietary data sources. Pools are built separately for five attributes, so evidence can differ by attribute:

  • Location. Comparable listings near the asset.
  • Property type. Comparable listings of a similar property type.
  • Bedrooms. Comparable listings of a similar size.
  • BER. Comparable listings with a similar energy rating.
  • Price cohort. Comparable listings in a similar price range.

Pools widen geographically only when local evidence is too thin, and the distance actually used is recorded. Several behavioural measures are taken for each pool and compared with that pool's own history, so evidence reflects change in the local market rather than a regional average.

The lender-supplied collateral value remains the recorded collateral value used in the monitoring workflow, including for LTV. Panoperty does not replace it or display any internal reference used to select comparable listings.

What each evidence state means

Each attribute carries one of four states:

Weakening
Local buyer demand is materially weaker than its own recent history.
Stable
Local buyer demand is within its normal range.
Strengthening
Local buyer demand is materially stronger than its own recent history.
Insufficient evidence
There is not enough comparable activity, or not enough history, to reach a view. Shown explicitly and never filled in.

States describe buyer behaviour relative to its own history. They are monitoring categories, not estimates of value or probability. An asset is shown as weakening when any of its attributes is weakening.

Persistence

Persistence records how long an attribute has held its current state across consecutive monthly reviews, for example WEAKENING · 3 consecutive reviews · 62 days. A missing or insufficient observation breaks the sequence, and a change in the asset's energy rating starts a new sequence for that attribute. Persistence describes continuity of evidence. It does not predict property value. Research on whether weakening evidence persists is summarised on the buyer-demand evidence page.

LTV and valuation age

LTV is the loan exposure divided by the lender's supplied collateral value. Valuation age is the time since the lender's valuation date. Both use the supplied value and date exactly as supplied.

How assets are ranked

Each asset's position in the queue reflects three components: LTV, valuation age and the strength of weakening local demand evidence across its attributes. Assets with no weakening attribute are still ranked on LTV and valuation age. Each component, the contributing attributes and the calculation version are stored with the rank, so every position can be explained to a reviewer.

Provenance and versioning

Every evidence record carries:

  • the asset, attribute and monitoring period;
  • the size of the comparator pool and the distance at which it was formed;
  • the location source of the asset;
  • the evidence state and whether enough history was available;
  • the calculation version that produced it.

Re-running a period replaces that period's records rather than duplicating them. Earlier periods are retained.

Calibration and governance

How the ranking components are weighted against each other, and how many assets are reviewed each month, are governance and calibration decisions for each deployment, not established credit policy. The lender's own review policy decides what happens to an asset in the queue.

Limitations

  • Evidence reflects listings observed in Panoperty's proprietary data sources. Coverage varies by location and property type, which is why pool size, distance and insufficient-evidence states are reported.
  • New assets and new comparator pools need some history before a state other than insufficient evidence can be assigned.
  • Monitoring currently covers residential collateral in Ireland, with exposures in euro.
  • Evidence states describe buyer behaviour. They are not valuations and have not been validated as predictors of value, loss or default.

Detailed methodology

Parameters, measure definitions, comparator rules and calculation detail are provided to institutions within the Collateral Monitor and, on request for model-risk or validation review, as part of a portfolio assessment.

Request the detailed methodology

Parameters, measure definitions and calculation detail are shared with institutions as part of a portfolio assessment.