{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp11_mechanism_context_external20_20260804","research_id":"eoa_inverse_innovation_exp11_external_scrutiny_20260804","cell_id":"layer_decay_and_expiration_management__economics_finance","opaque_id":"layer_decay_and_expiration_management__economics_finance__C","search_lanes":{"direct_problem":{"queries":["IFRS 9 post model adjustments overlays governance review expiration sunset stale overlays credit loss","management overlays expected credit loss periodically reviewed removed when no longer needed guidance","credit loss model overlays inventory governance audit trail post-model adjustments","site:bis.org IFRS 9 overlays post model adjustments governance"],"source_ids":["SRC1","SRC2","SRC3","SRC4","SRC8"],"no_result_note":null},"closest_prior_art":{"queries":["PRA SS1/23 post-model adjustments time limited independent review tracked ongoing appropriateness","\"post-model adjustment\" \"time-limited\" credit loss","\"management adjustment\" \"sunset\" IFRS 9 overlay","\"post-model adjustments\" register review date audit trail"],"source_ids":["SRC1","SRC4","SRC5","SRC6"],"no_result_note":null},"historical_terminology":{"queries":["CECL qualitative factors adjustments documentation review sunset policy Q factors allowance historical terminology","bank annual report management overlay released no longer required IFRS 9 overlay 2023","\"management overlay\" \"no longer required\" expected credit loss"],"source_ids":["SRC1","SRC3","SRC7"],"no_result_note":null},"products_practices_standards":{"queries":["PRA SS1/23 post-model adjustments time limited independent review tracked ongoing appropriateness","credit loss model overlays inventory governance audit trail post-model adjustments","bank annual report management overlay released no longer required IFRS 9 overlay 2023"],"source_ids":["SRC1","SRC6","SRC7"],"no_result_note":null},"non_english_regional":{"queries":["site:banque-france.fr IFRS 9 \"overlays\" ajustements post-modèle gouvernance","IFRS 9 ajustes post modelo overlays provisiones gobernanza revisión español","IFRS 9 modeloverlays management adjustments Aufsicht Deutschland","ajustements post-modèle IFRS 9 surcouche provisionnement réexamen"],"source_ids":["SRC8"],"no_result_note":null},"composition_subproblems":{"queries":["\"overlay\" \"expiry date\" \"IFRS 9\"","\"post-model adjustments\" register review date audit trail","\"management overlay\" \"no longer required\" expected credit loss","credit loss model overlays inventory governance audit trail post-model adjustments"],"source_ids":["SRC1","SRC4","SRC6","SRC7"],"no_result_note":null}},"sources":[{"source_id":"SRC1","title":"Model risk management principles for banks (Supervisory Statement SS1/23)","url":"https://www.bankofengland.co.uk/-/media/boe/files/prudential-regulation/supervisory-statement/2023/ss123.pdf","publisher":"Bank of England, Prudential Regulation Authority","date_or_year":"May 2023","source_type":"OFFICIAL_GUIDANCE","language":"English","claims_supported":["Banks should maintain a documented, firm-wide governance and control process for post-model adjustments.","Material PMAs require documentation, senior-management support, approval by appropriate authority, transparent reporting with and without the PMA, and independent review.","Review must assess continued relevance to the portfolio, reasoning, inputs, outputs, and root causes.","Documentation must specify criteria for reducing or removing PMAs and triggers through which prolonged or recurring use activates validation, remediation, recalibration, or redevelopment."]},{"source_id":"SRC2","title":"Overlays and in-model adjustments: identifying best practices for capturing novel risks","url":"https://www.bankingsupervision.europa.eu/press/blog/2023/html/ssm.blog230526~29af0452d6.sl.html","publisher":"European Central Bank, Banking Supervision","date_or_year":"26 May 2023","source_type":"OFFICIAL_GUIDANCE","language":"English","claims_supported":["Overlays became prevalent during the pandemic and are widely used for novel risks that models do not capture.","Evidence-based overlays can be appropriate, but require strong governance, transparency, and sound methodology.","Overlay, management adjustment, post-model adjustment, and top-level adjustment are overlapping industry terms."]},{"source_id":"SRC3","title":"The EBA’s monitoring of IFRS 9 implementation by EU institutions confirms need to timely address practices misaligned with expectations","url":"https://eba.europa.eu/publications-and-media/press-releases/ebas-monitoring-ifrs-9-implementation-eu-institutions","publisher":"European Banking Authority","date_or_year":"17 November 2023","source_type":"OFFICIAL_GUIDANCE","language":"English","claims_supported":["Overlays have become an integral part of ECL frameworks and can be temporary or more permanent.","Observed divergence in calibration and application creates prudential concern and requires stronger methodological and governance framing.","Periodic review and backtesting remain areas needing improvement."]},{"source_id":"SRC4","title":"COVID-19: post model adjustments on expected credit losses for banks","url":"https://www.pwc.ch/en/insights/accounting/post-model-adjustments-on-expected-credit-losses-for-banks-during-covid-19.html","publisher":"PwC Switzerland","date_or_year":"2 June 2020","source_type":"TRADE_PROFESSIONAL","language":"English","claims_supported":["Overlays require robust processes, governance, internal controls, and transparent documentation.","Overlay governance should record the addressed limitation, quantification, assumptions, data, affected exposures, and performance assessment.","Management should determine how an overlay will be consumed over time through model redevelopment, new data, or realized loan-level losses."]},{"source_id":"SRC5","title":"PMA implementation: Don't let overlays become oversights","url":"https://www.deloitte.com/uk/en/services/consulting-risk/blogs/2025/pma-implementation-dont-let-overlays-become-oversights.html","publisher":"Deloitte UK","date_or_year":"13 January 2025","source_type":"TRADE_PROFESSIONAL","language":"English","claims_supported":["Post-model adjustment management is described as a lifecycle covering identification, quantification, and implementation.","Clear risk taxonomy and root-cause analysis are recommended to distinguish unmodelled risks, internal or external parameter effects, and model or data limitations.","PMA implementation is already an area of elevated regulatory attention."]},{"source_id":"SRC6","title":"SAS Allowance for Credit Loss","url":"https://www.sas.com/en_gb/software/allowance-for-credit-loss.html","publisher":"SAS","date_or_year":"Accessed 4 August 2026","source_type":"FIRST_PARTY_PRODUCT","language":"English","claims_supported":["A commercial product centralizes ECL data, models, scenarios, adjustments, and workflows in one repository.","The product supports rule-based post-model adjustments and overlays, simulations, attribution, traceable workflows, change tracking, and auditable artifacts.","Commercially available workflow tooling already implements important parts of overlay identity, lineage, governance, and shadow analysis."]},{"source_id":"SRC7","title":"BDC Annual Report 2023","url":"https://www.bdc.ca/globalassets/digizuite/42073-bdc-annual-report-2023.pdf","publisher":"Business Development Bank of Canada","date_or_year":"2023","source_type":"FIRST_PARTY_PRODUCT","language":"English","claims_supported":["A reporting institution disclosed that its COVID-era management overlay was no longer required once economic uncertainty was adequately captured by quantitative macro-forecast models.","The disclosure demonstrates actual retirement of an ECL overlay when its rationale was absorbed by modelled results."]},{"source_id":"SRC8","title":"Memoria de Supervisión 2024, Recuadro 2.5: Cobertura de riesgos mediante NIIF 9 overlays","url":"https://www.bde.es/f/webbe/Secciones/Publicaciones/PublicacionesAnuales/MemoriaSupervisionBancaria/24/MemoriaSupervision2024_Cap2_Rec2.5.pdf","publisher":"Banco de España","date_or_year":"2024 supervisory report","source_type":"OFFICIAL_GUIDANCE","language":"Spanish","claims_supported":["A regional supervisory review covering 53 significant institutions found that overlays are commonly used for risks not captured by conventional provisioning models.","Approximately half of reviewed institutions lacked either clear assignment of responsibilities or robust review mechanisms.","The review found generic overlays, excessive dependence on expert judgment, and inconsistent treatment of identifiable risks, directly supporting a continuing governance problem."]}],"problem_evidence":{"status":"SUPPORTED","finding":"The problem exists, although the evidence more directly establishes governance weakness than a measured stock of stale layers. European supervisors report widespread and integral overlay use, divergent calibration, and continuing review deficiencies; Banco de España reports that roughly half of reviewed significant institutions lacked clear responsibilities or robust review mechanisms. PRA guidance explicitly addresses continued relevance, prolonged use, recurring PMAs, and criteria for reduction or removal, while BDC documents an overlay becoming unnecessary after its rationale was captured in models.","source_ids":["SRC1","SRC2","SRC3","SRC7","SRC8"],"uncertainty":"No retained source quantifies how often successive overlays double-count risk, become ownerless, or cannot be reconstructed. Evidence is supervisory and illustrative rather than a representative longitudinal estimate of stale-overlay prevalence."},"adopter_evidence":{"status":"SUPPORTED","finding":"Identifiable adopters are banks and other lenders using IFRS 9 or CECL models. The authorizers are senior management, risk committees, audit committees, and equivalent provisioning governance bodies; PRA guidance expressly assigns approval to an appropriate authority, and commercial ECL workflow products target these institutions.","source_ids":["SRC1","SRC6","SRC8"],"uncertainty":"Exact committee names, decision rights, and record-retention obligations vary by institution and jurisdiction."},"implementation_evidence":{"status":"PARTLY_SUPPORTED","finding":"Core implementation elements are already supported: documented rationale, authority approval, independent continued-relevance review, criteria and triggers for reduction or removal, remediation of recurring overlays, transparent with/without comparisons, centralized traceable workflows, simulations, audit artifacts, and actual retirement when models absorb the risk. The retained sources do not establish routine use of the proposal's complete state machine, including explicit supersession links, orphan detection, preservation exceptions, archived rehydration, and a rollback window.","source_ids":["SRC1","SRC4","SRC5","SRC6","SRC7"],"uncertainty":"Product marketing establishes capability rather than adoption quality, and the single bank disclosure does not demonstrate a complete reconstructable lifecycle process."},"prior_art":{"disposition":"ESTABLISHED_PRACTICE","closest_analogues":[{"name":"PRA SS1/23 Principle 5.1 post-model-adjustment governance","source_ids":["SRC1"],"same_problem":true,"same_causal_lever":true,"overlap":"Requires a documented PMA process, rationale, authority approval, transparent impact reporting, independent continued-relevance review, reduction/removal criteria, prolonged-use triggers, and remediation of recurring overlays.","remaining_difference":"It does not prescribe the proposal's named active/review-pending/archived states, overlay-level supersession and orphan flags, archival rehydration procedure, or rollback window."},{"name":"PwC overlay establishment and consumption framework","source_ids":["SRC4"],"same_problem":true,"same_causal_lever":true,"overlap":"Treats overlays as governed, documented adjustments whose limitation, assumptions, data, exposure scope, performance, and consumption over time must be specified.","remaining_difference":"It is professional guidance framed as questions, not an integrated register with explicit lifecycle states, exception handling, and restoration controls."},{"name":"SAS Allowance for Credit Loss governed workflow","source_ids":["SRC6"],"same_problem":true,"same_causal_lever":true,"overlap":"Commercially implements a centralized repository, overlay and manual-adjustment workflows, simulations, attribution, change tracking, repeatability, and audit-ready artifacts.","remaining_difference":"The public product description does not show expiry triggers, supersession relationships, orphan detection, archived restoration, preservation overrides, or controlled reinstatement."},{"name":"BDC retirement of a management overlay","source_ids":["SRC7"],"same_problem":true,"same_causal_lever":false,"overlap":"Demonstrates that an overlay can be removed when the model again captures the underlying uncertainty.","remaining_difference":"The disclosure does not reveal whether removal followed an explicit lifecycle register, dependency check, archival reconstruction test, or rollback control."}],"contrastive_claim_remaining":"Beyond existing PMA governance, an overlay-level state register that adds explicit supersession and orphan detection, preservation exceptions, reconstructable archival retirement, and controlled reinstatement may reduce stale-overlay selections, reviewer disagreement, and reconstruction time during a reporting cycle.","contrastive_claim_falsifier":"In a shadow comparison on one portfolio, the enriched lifecycle register produces no practically meaningful reduction in stale-overlay selections, inter-reviewer disagreement, or reconstruction time relative to an SS1/23-aligned baseline, or it causes reviewers to omit justified uncertainty coverage.","confidence":"HIGH","search_limitations":"The bounded search used eight retained public sources and six adversarial lanes. It did not inspect proprietary bank procedures, paid databases, patents, software demonstrations, or non-public audit workpapers; therefore it cannot determine universal adoption or world novelty."},"researchability_gates":{"externally_supported_problem":{"status":"PASS","rationale":"Supervisory sources document widespread overlay use, governance deficiencies, unclear responsibility, weak review, and concern about persistent or recurring PMAs.","source_ids":["SRC1","SRC3","SRC8"]},"identifiable_adopter_or_authorizer":{"status":"PASS","rationale":"Banks and lenders are identifiable adopters, and senior management, risk committees, audit committees, or equivalent provisioning bodies are explicit authorizers.","source_ids":["SRC1","SRC6"]},"distinct_testable_incremental_claim":{"status":"PASS","rationale":"Although the core intervention is established, the retained prior art does not disclose the full combination of explicit lifecycle states, supersession and orphan checks, reconstructable archival retirement, preservation exceptions, and rollback. Its incremental effect can be tested against an existing governed PMA process.","source_ids":["SRC1","SRC4","SRC6"]},"bounded_next_evidence_step":{"status":"PASS","rationale":"One portfolio and one reporting cycle can be run in shadow mode, comparing blinded decisions and reconstruction time without changing booked reserves; SAS also establishes that simulation and traceable workflow support are technically available.","source_ids":["SRC1","SRC6"]},"no_unresolved_safety_or_authority_stop":{"status":"PASS","rationale":"The proposed shadow test leaves booked reserves unchanged and retains evidence. Existing guidance supplies an approval and independent-review structure. Automatic retirement, evidence destruction, and circumvention of legal holds remain excluded.","source_ids":["SRC1","SRC4"]},"adequate_search_evidence":{"status":"PASS","rationale":"All six required lanes were searched, including aliases and older CECL Q-factor terminology, standards, products, actual institutional practice, Spanish regional terminology, and combinations involving expiry, registers, review, audit trails, and retirement. Exactly eight sources from eight publisher contexts were retained and opened; five are official guidance and two are first-party sources.","source_ids":["SRC1","SRC2","SRC3","SRC4","SRC5","SRC6","SRC7","SRC8"]}},"strict_success":false,"screen_survival":false,"remaining_research_value":"MODERATE","recommended_next_step":"Run a controlled one-cycle shadow pilot on one portfolio against an already governed PMA baseline. Pre-register overlay-level stale criteria and primary outcomes: stale-overlay selection rate, inter-reviewer agreement, reconstruction time, missed necessary coverage, and unexplained shadow reserve effects. Preserve all evidence and require committee review before any production change.","world_novelty_boundary":"This bounded public-source search finds the core problem–intervention package to be established practice and leaves only a narrower, testable systems-integration claim. It does not establish world novelty, patentability, freedom to operate, market size, routine global adoption, realized impact, or superiority to existing institutional implementations."}