New here? The Knowledge Centre has a first-time orientation, a searchable FAQ for every audience (banks, Bank of Ghana, EPA, IFC, World Bank) and a glossary of every abbreviation. This page is the detailed methodology reference.
PortfolioCheck screens a bank's mortgage / property book for physical-climate & environmental risk in Ghana, and turns it into the numbers and the report a Ghanaian lender needs for the Bank of Ghana Climate-Related Financial Risk Directive. This page is both a user guide (how to drive the tool) and a transparent methodology (how every number is built, and what it deliberately does not do).
lat/lon, a
gps GhanaPostGPS digital address (e.g. GA-183-4317, resolved to the precise
5 m grid square), or a free-text address/region (geocoded to
locality level). Optional asset_id,
exposure, ltv, property_type. Points outside Ghana are dropped./api/screen) to wire into a
loan-origination system.Privacy: an uploaded book is held in memory only for your session — nothing is persisted to disk or shared. For maximum security the whole tool can be deployed on-premise so the loan data never leaves the bank.
A 0–100 composite = 50% weighted hazard load + 50% of the single worst hazard, so one extreme hazard that can impair the security is not averaged away. Default (untagged) weights: flood 34% · coastal 18% · heat 16% · riparian 12% · land 10% · drought 10%. Bands: Low <25 · Moderate 25–50 · High 50–75 · Critical ≥75.
From v2.6.0 these weights adapt per asset to its materiality profile (coastal · floodplain · farmland · urban) so the asset's dominant physical threat leads the composite — a sea-threatened villa is led by coastal / sea-level, a field by drought. A coastal or floodplain designation also floors that lead peril at High — always raising, never lowering, a measured or satellite-observed score, and overridable by re-tagging. Untagged assets use the default weights above, and every weight is recalibratable to the bank's own loss experience.
Banks act on cedis, not scores. The translation is built on one transparent fundamental — an Average Annual Loss (AAL) per hazard (a depth-damage × event-frequency proxy, = max-AAL × score², combined across hazards) — and keeps two loss channels separate so neither is double-counted:
These feed an indicative annual marginal ECL = PD × LGD × EAD: the climate LGD uses the repriced collateral and a smooth downturn liquidation recovery, floored at 5% (Basel-style); the climate PD rises with negative equity (strategic default) and uninsured damage. The single-event figure is a correlated-tail proxy — one severe event hitting every High+ asset in a region at once.
Every constant above is an illustrative, reasonable default, exposed for the bank to recalibrate against its own loss experience. This is decision-support — not a lifetime/12-month staged IFRS 9 ECL, not a valuation, and not the bank's PD/LGD model.
The baseline is the locked "BoG baseline" parameter set (so figures are comparable across banks). A bank that has run a calibration exercise loads a calibrated parameter set with no code change — a JSON of the scalar knobs (a single AAL multiplier to match observed average annual property-damage loss, the base PD, the repricing intensity, the forced-sale recovery, the uninsured share). Back-solve helpers fit the AAL multiplier to the bank's observed loss rate; the calibrated set's name then appears on every report and BoG return, so a supervisor always sees which parameters produced the numbers. The dashboard recomputes from the same set automatically.
Every asset can be pulled through the three headline IPCC AR6 pathways — SSP1-2.6 (low), SSP2-4.5 (intermediate) and SSP5-8.5 (high) — across 2030 / 2050 / 2100. Each hazard is intensified by an AR6-grounded factor versus today: extreme-rainfall intensity (Rx1day) for flood/riparian, the count of hot/heat-stress days for heat, consecutive-dry-days for drought, and absolute sea-level rise (cm) for coastal — sourced from AR6 WG1 (Atlas / Ch.11 extremes / sea-level), the World Bank Climate Knowledge Portal and the Greater-Accra extremes study (Aja et al. 2024). The factor is applied to today's 0–1 hazard score with a bounded, saturating odds map s′ = s(1+kf)/(1+s·kf), so a projection never exceeds 1.0 and an asset already at high risk barely moves while mid-range assets move most. Pick scenario and horizon in the top bar; open any asset to see its risk trajectory under all three pathways. Screening estimates, not downscaled forecasts — deep uncertainty grows with horizon, which is why scenario and horizon are user-selectable rather than baked into a single future. Data quality on each asset is flagged on the PCAF 1–5 scale (screening is never better than 3).
A bank can record the adaptation measures completed on a property — engineered/
maintained drainage & SuDS, a raised plinth (+0.3 m / +1 m), floodproofing and barriers, property
flood resilience, sea defence / managed setback, cool or green roof, on-site retention. Each carries a
defensible risk-reduction credit applied to the relevant hazard, so the asset's band, AAL and climate
ECL drop in front of you. Credits are combined multiplicatively on residual risk
(1 − Π(1 − cᵢ)) and capped (flood 85%, coastal 80%, heat 50%) — a +1 m plinth subsumes a
+0.3 m one, and floodproofing/PFR are damped where the property is already elevated, to avoid
double-counting. Values are engineering priors from FEMA/USACE/NIBS benefit-cost, Kreibich et al.
(2005), Defra FD2657 / CIRIA C790 and USACE/LBNL coastal & cool-roof studies — illustrative pending
local calibration. Recording a finished, maintained measure and re-scoring the exposure is the auditable
mitigate + monitor evidence the BoG Directive, IFRS S2 / TCFD and the forward-looking
IFRS 9 ECL expect at asset level, and it surfaces the green-loan / resilience origination candidates the
Sustainable Banking Principles reward. Evidence gating (conservative by default): a recorded
measure is treated as self-reported until supporting evidence is accepted by the supervisor.
A self-reported declaration receives no regulatory credit; full credit is taken only once the asset is
accepted by BoG following review of site inspection / as-built drawing / maintenance evidence. The BoG
return states how many assets are verified versus self-reported, and the gap register records that the bank must
hold that evidence before relying on the reduced figures — so unverified self-attestation can never quietly
lower a regulatory return. The review is a closed maker–checker loop: the bank uploads a supporting
document and note against the asset; the supervisor opens it and records Accept, Reject or
Request clarification with a decision note; only an Accept activates the credit, and it is pinned to
exactly the measures whose evidence was reviewed. Changing the declared measures voids any prior acceptance and
discards the stale document; an already-decided item cannot be re-decided by a replayed request. On the supervisor's
top-down cross-check the console shows the bank's filed High+ share against the share it would report
if every declared-but-unevidenced measure were credited — the unverified adaptation pipeline — and
each drill-down carries a report-delivery panel naming the Directive return section and the SHA-256
submission fingerprint, which moves the instant any figure changes. In the authenticated book the same
loop is persistent and store-backed: from /p/<id>/adaptation-cases an institution opens a dated
case per asset, attaches evidence and submits it; a Bank-of-Ghana supervisor account works the queue at
/supervisor/adaptation and issues an attributable verdict (accept / reject / request-more-information) — an
accepted verdict re-scores the asset (baseline → improved) and a one-click BoG package bundles the return, a
case manifest, the evidence and a fingerprinted receipt as a ZIP. Carry measures in the upload via an adaptations column
(e.g. drainage;elevation_1m) and an optional adaptation_verified column
(yes/no); the Export CSV round-trips them.
The report is structured as the bank's compliance artifact, mapping the screen to the Bank of Ghana Climate-Related Financial Risk Directive, the Ghana Sustainable Banking Principles, IFRS S2 / TCFD, NGFS and PCAF.
Regulatory basis (verified June 2026). The BoG Climate-Related Financial Risk Directive was issued November 2024 (under Act 930 s.92(1) and related Acts). Banks must update governance, risk and policy frameworks by 31 December 2025, with full implementation from 1 January 2026 (SDIs/NBFIs from January 2027), and must file semi-annual reports to the BoG on material risks and exposure to vulnerable sectors — disclosure aligned to IFRS S2 / ISSB. PortfolioCheck's “% of exposure at high risk, by region and hazard, with mitigation” output maps directly onto that semi-annual report. Confirm the latest text against the BoG primary source before filing.
Dr. Minka Aduse-Poku — environmental and climate scientist; PortfolioCheck methodology owner.
Related to MineCheck. Same discipline: ground-truthed Ghana data, transparent methodology, an honest gap register — no black box.
We screen a bank's anonymised property book on Ghana-resolution engines and return a board- and BoG-Directive-ready report — on-premise if required, so the data never leaves the bank.
Email minka@resilicity.de