{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp12_substrate_denial72_20260805","cell_id":"incentive_compatible_rule_design__accounting_auditing","arm":"ORDINARY_MAX","candidate_id":"incentive_compatible_rule_design__accounting_auditing__ORDINARY_MAX","proposal_index":1,"version":0,"title":"Protected-Exception Choice Rule for Balance-Sheet Reconciliations","problem":"At month-end, an account preparer may already know that a balance-sheet reconciliation contains an unresolved item. If disclosed exceptions worsen a clean-account metric, attract open-ended review, or threaten an on-time rating, while concealment is detected only occasionally, temporarily clearing, reclassifying, re-aging, or omitting the item can be the preparer's locally advantageous choice. The certification rule then rewards an apparently clean reconciliation rather than truthful exposure of unresolved accounting uncertainty.","actors":["Balance-sheet account preparers","First-level reconciliation reviewers","Business-unit controllers and finance managers","Central controllership and the Chief Accounting Officer","Internal audit or an independent controls-testing team","Audit committee and any personnel-evaluation authority","External financial-statement auditor acting independently"],"observable_state":"The candidate problem is observable through validated clearing entries that reverse after close, aged items that disappear before certification and later reappear in another account, clean attestations contradicted by re-performance, disclosures clustered after audit selection, and post-certification corrections concentrated around rating deadlines. These signals require transaction-level examination and evidence that the preparer knew of the item before certification; none alone proves strategic concealment.","consequence":"Known uncertainty can remain outside the exception workflow, leaving balances and control attestations weakly supported, delaying correction, and forcing auditors to reconstruct events later. Promptly candid preparers can also incur more visible workload or evaluation cost than peers who conceal comparable items.","affected_objective":"Timely and truthful identification, accounting, escalation, and resolution of material or risk-relevant reconciling items, with reliable close evidence and fair treatment of good-faith preparers; minimizing the reported exception count is explicitly not the objective.","intervention":"Replace the single clean-or-late certification with a locked, pre-review choice for scoped accounts. Under the protected-exception route, the preparer timestamps each unresolved item and supplies its amount, age, source evidence, uncertainty, containment action, and remediation owner. A timely good-faith disclosure does not worsen the clean-account score merely because an issue exists and receives a bounded validation review. Under the clean-certification route, the preparer affirmatively declares no known item above published accounting and risk criteria. Only after the route is locked does an independent system draw risk-weighted, randomized re-performance samples. A validated pre-known omission or unsupported clearing entry triggers expanded testing and a proportional process-rating consequence after independent review and appeal; the mere existence of an error does not. Reviewer evaluation emphasizes disclosure completeness and resolution quality rather than raw finding counts. The protected route never waives booking, escalation, fraud, legal, control, or external-audit duties.","structural_mapping":[{"archetype_element":"Participant Role Map","domain_realization":"Preparers choose and hold account-specific knowledge; reviewers and controllers influence workload and ratings; controllership defines reconciliation rules; internal audit verifies claims independently; governance authorities control personnel consequences and appeals."},{"archetype_element":"Desired Outcome Specification","domain_realization":"The target is timely exposure and resolution of relevant reconciling items, separated from the easily gamed proxy of a low open-item count."},{"archetype_element":"Action and Choice Set","domain_realization":"A preparer can disclose with evidence, certify clean, defer investigation, split or re-age an item, post a clearing or reclassification entry, move the item to another account, or exit/escalate through an exception channel."},{"archetype_element":"Incentive Payoff Map","domain_realization":"The current rule can make disclosure costly through lower ratings and expanded review while making concealment attractive if detection is unlikely. The proposed rule caps good-faith disclosure burden and attaches probability-weighted review and consequences to a contradicted clean claim."},{"archetype_element":"Information Structure Map","domain_realization":"Preparers privately know an item's context before auditors do. Route criteria are public, but audit selection remains unknown until the declaration is locked; timestamps, journals, support, messages, and reversals can establish whether knowledge preceded certification."},{"archetype_element":"Truthfulness Condition","domain_realization":"For a preparer with a qualifying unresolved item, truthful routing is the best response only when the predictable burden of protected disclosure is lower than the probability-weighted cost of a validated knowing omission under clean certification. Genuinely clean accounts must still prefer the clean route."},{"archetype_element":"Verification Rule","domain_realization":"Independent testers validate protected disclosures within a fixed scope and randomly re-perform locked clean certifications using original-source evidence, journal history, subsequent reversals, and published evidentiary standards."},{"archetype_element":"Penalty or Reward Rule","domain_realization":"Timely evidenced disclosure receives metric safe harbor and bounded review. Only an independently substantiated inconsistency between prior knowledge and a clean declaration can produce expanded testing or a proportional rating consequence, subject to appeal."},{"archetype_element":"Participation and Fairness Constraints","domain_realization":"Both routes use published thresholds; selecting the protected route creates no adverse inference; review burden is capped; risk may affect sampling probability but not access to safe harbor; and preparers can appeal knowledge, materiality, or evidence determinations."},{"archetype_element":"Strategic Response and Gaming Monitor","domain_realization":"Monitoring covers route selection, validation yield, clearing-and-reversal patterns, migration to excluded accounts, item splitting, resolution age, review hours, appeals, retaliation reports, and burden disparities, with rule changes made through documented governance."}],"mechanism_mapping":[{"mechanism_slug":"anti_gaming_scoring_rule","role":"Replace the raw count of reported exceptions with measures of timely evidenced disclosure, truthful certification, and resolution quality, so suppressing an item no longer directly improves the intended performance score.","counterfactual_removal":"If raw exception counts still lower the preparer's or reviewer's rating, disclosure retains an immediate penalty and concealment can remain attractive despite the other features."},{"mechanism_slug":"self_selection_menu","role":"Offer an evidence-bearing protected-exception route and an affirmative clean-certification route whose different obligations cause participants to reveal the state they privately observe through their choice.","counterfactual_removal":"Without distinct precommitted routes, there is no low-cost truthful path and no precise clean claim against which later contradictory evidence can be evaluated."},{"mechanism_slug":"blind_or_randomized_review_rule","role":"Select clean certifications for risk-weighted re-performance only after route lock, preventing participants from conditioning concealment on known audit selection while avoiding universal review.","counterfactual_removal":"If reviews are predictable, a preparer can conceal only in accounts unlikely to be tested; if every account is reviewed, the bounded-burden participation advantage disappears."},{"mechanism_slug":"audit_and_penalty_system","role":"Combine independent validation with expanded testing and a proportional, appealable consequence for substantiated pre-known omissions rather than for good-faith errors or issue existence.","counterfactual_removal":"Random review would discover some items but would not change the relative payoff of a false-clean declaration if contradiction carried no consequence."}],"causal_chain":["Preparers acquire private, account-specific knowledge of unresolved items before reviewers or auditors can fully observe them.","A clean-account or on-time metric and open-ended review burden make visible disclosure immediately costly.","Because later detection is uncertain, temporary clearing, reclassification, re-aging, or false-clean certification can dominate prompt disclosure for a self-interested preparer.","The protected route removes the raw-count penalty, requires contemporaneous evidence, and makes the review burden predictable.","The clean route converts silence into an affirmative, locked claim and prevents route changes after audit selection becomes known.","Randomized independent re-performance gives concealment a nonzero expected detection cost without reviewing every account.","Expanded testing and proportional consequences apply only after evidence establishes a pre-known omission, while appeal and safe-harbor boundaries limit false punishment.","If the truthfulness condition holds, a preparer with a qualifying item expects a better outcome from protected disclosure, while a genuinely clean preparer still prefers clean certification.","Earlier truthful routing makes the item available for accounting, escalation, and remediation before later audit reconstruction is required."],"baseline":"The comparison baseline is the existing single-route monthly reconciliation process: preparers certify by a deadline, dashboards emphasize clean or timely completion and open-item counts, disclosure can expand review, and periodic testing occurs through known thresholds or retrospective audit. Errors and omissions are generally addressed after discovery, with no protected evidence-bearing route that distinguishes candid disclosure from a contradicted clean claim.","nearest_rivals":["Metric removal alone: abolish clean-account and raw finding-count penalties and rely on reviewer judgment. This directly addresses reporting cost and may be sufficient if metric pressure is the main cause, but it provides neither a structured revelation route nor a truthfulness check.","Continuous automated reconciliation and journal-anomaly detection: identify clearing entries, reversals, and unsupported movements without relying on voluntary disclosure. This may dominate where data capture is comprehensive, but it cannot necessarily observe the contextual knowledge held by preparers and does not itself reward volunteering that knowledge.","Universal evidence review or complete re-performance: verify every reconciliation rather than designing around private information. It reduces reliance on truthful routing but may impose greater close delay and review burden; if it is affordable and sufficiently accurate, the proposed mechanism is unnecessary.","Stronger sanctions and certification training: clarify duties and increase consequences for inaccurate attestations. This is simpler, but it leaves honest disclosure costly and may produce defensive concealment when standards or evidence are ambiguous.","Random audit alone: make detection less predictable without changing the exception metric. This can deter omission, but candid preparers remain disadvantaged if reporting an issue still worsens their rating or creates unbounded work."],"remaining_contrastive_claim":"The bounded mechanistic claim is that, when preparers possess private pre-close knowledge and verification is necessarily incomplete, combining an evidence-bearing protected disclosure route with locked, unpredictable verification of clean claims changes both sides of the payoff comparison: it lowers the cost of truthfully revealing an item and raises the expected cost of knowingly concealing it. No novelty, prevalence, demand, or effect-size claim is made.","authority_safety":{"decision_authority":"The Chief Accounting Officer or controller owns reconciliation-policy and dashboard changes; the Chief Audit Executive controls independent sampling and validation. The audit committee, and HR or compensation authority where applicable, must approve any use in personnel evaluation or discipline. The external auditor retains independent judgment and cannot be directed to rely on the mechanism.","authorized_first_step":"Joint written approval from controllership and internal audit authorizes only a two-close pilot on low- or medium-risk, nonmaterial accounts. Existing close controls remain controlling; pilot results cannot affect compensation or discipline; normal accounting, escalation, retention, and external-audit obligations continue unchanged.","excluded_actions":["No alteration, suppression, backdating, or delayed recording of ledger entries or audit evidence for the pilot.","No safe harbor for fraud, intentional material misstatement, knowing legal violations, retaliation, or failure to perform required escalation.","No automatic adverse action based on route choice, an algorithmic flag, or a single contradictory sample.","No undisclosed employee surveillance or reuse of pilot data outside the approved accounting-control purpose.","No substitution for required management certifications, internal controls, regulatory obligations, or external-audit procedures.","No disclosure of sampling seeds or selected accounts before route lock, and no pressure on preparers to choose either route."],"halt_rollback":"Pause new route assignments and revert prospectively to the current attestation process if the pilot delays close, exceeds its validation-capacity cap, mishandles a material item, generates retaliation or privacy concerns, produces unreliable knowledge determinations, or conflicts with an auditor or regulator requirement. Preserve the original audit trail, independently review affected decisions, continue remediation of discovered items, and draw no adverse inference from having selected the protected route."},"negative_tests":{"strongest_counterevidence":"Blinded file review and interviews may show that apparent concealment instead reflects ambiguous materiality rules, system cutoffs, inadequate staffing, insufficient skill, or legitimate timing—and that preparers neither knew of the items before certification nor gained workload or evaluation advantage from omission. That evidence would favor process clarification or capacity investment rather than incentive redesign.","problem_falsifier":"The problem is falsified in the scoped setting if no validated item is both known before certification and omitted, cleared, re-aged, or moved in a way that improves a rating or avoids review; false-clean discrepancies are unrelated to deadline or evaluation stakes and disappear when definitions and supporting systems are corrected.","intervention_falsifier":"The intervention is falsified for the scoped use if the combined rule does not outperform the baseline or random-audit-only rival on validated pre-known omissions, if route choice fails to correspond to privately known account state, or if any directional improvement is offset by worse close timeliness, unresolved-item aging, review burden, low-value overreporting, inequitable appeals, retaliation, or migration of concealment to excluded accounts.","risks":["Defensive over-disclosure of immaterial items could overwhelm validation and obscure consequential exceptions.","Preparers could label deliberate conduct as good-faith uncertainty to seek safe harbor.","Items could be split, re-aged, or moved below published thresholds, shifting rather than removing gaming.","Reviewers and preparers could collude or pressure one another about route selection.","Leaked sampling logic could let participants concentrate concealment in low-probability accounts.","Small or poorly resourced teams could bear disproportionate documentation and audit burdens.","False findings about prior knowledge could chill professional judgment or enable retaliation.","Expanded testing could become punitive surveillance rather than a proportional verification response.","The protected label could be misunderstood as permission to defer required accounting or escalation.","The new disclosure-quality metrics could themselves become narrow proxies that crowd out judgment."]},"next_evidence_step":"Pre-register a two-close, four-team crossover covering at most 80 low- or medium-risk, nonmaterial account-months. Randomly assign the order in which comparable teams use the combined rule or the current baseline, keep all normal controls active, validate protected disclosures up to a fixed cap, and re-perform a concealed 20% random sample of clean claims. Independent reviewers should code whether knowledge preceded certification, clearing entries later reversed, omitted items met published criteria, and appeals were sustained. Record validated pre-cutoff disclosures, pre-known omissions, resolution age, review hours, close delays, route migration, appeals, burden disparities, and retaliation reports. Define stopping thresholds before launch and treat the result only as evidence about problem existence, incentive direction, feasibility, and safety—not as an effect-size or prior-art determination.","prior_art_status":"UNSEARCHED","diversity_from_prior_proposals":"Not assessed against prior proposals because runtime isolation forbids access to them. Within this record, the candidate is specifically organized around a locked choice between evidence-bearing protected disclosure and clean certification followed by post-choice randomized re-performance, rather than a generic recommendation for more auditing or harsher penalties.","revision_record":{"parent_version":null,"progress_targets_addressed":["Initial complete candidate; no inherited progress targets."],"conceptual_changes":["Initial formulation of reconciliation concealment as a private-information and best-response problem."],"operational_changes":["Initial specification of the two-route attestation, metric safe harbor, post-lock randomized validation, bounded consequences, appeal, and rollback."],"evidence_changes":["Initial bounded crossover pilot and explicit problem, intervention, safety, and alternative-explanation tests."],"claim_changes":["Restricted the claim to the proposed payoff-changing mechanism and expressly excluded novelty, prevalence, demand, and effect-size claims."]}}