Reproducibility snapshot v4.23.0 for "The anti-maladaptation filter hypothesis: cohort-scale empirical test that NAM-classification-filtered portfolio allocation delivers higher expected avoided-loss NPV than five naive baseline allocations for climate adaptation of 725,462 electricity substations across 39 OECD countries"

  Frozen v4.23.0 reproducibility snapshot accompanying the P4 manuscript submitted
  to Energy Economics (Elsevier). Contains: main manuscript v3 (12,947 words) with
  §4 empirical results from the Phase P4-5 39-country × 725,462-substation cohort
  solve (2026-07-26T16:28:03Z); Supplementary Materials §S1 per-country cost matrix
  + §S3 LP-vs-integer 0/1 + CVaR extended + shared-model τ² caveat + Cochran's Q
  refresh note + uniform-CV limitation + §S8 pooled DerSimonian-Laird per-country
  weight tables; five main-text SVG figures (Wong 2011 accessibility palette);
  full cohort JSON outputs from the 5,265 LP solves; one-command reproducibility
  scripts (cohort_solve.py + cohort_extended.py) with orchestrator runbook; cover
  letter + author declaration for the Energy Economics submission; audit response
  memo documenting cumulative 25-finding Beta remediation closure + Part 3 R19.1
  CLOSURE ADDENDUM.

  Empirical headline: 5-of-5 baselines strictly dominated at reference cell
  (SSP2-4.5 × 2050 × 15% budget) with cohort-mean reference-cell dominance
  margins +145.08% (population-weighted), +152.65% (GDP-weighted), +97.50%
  (substation-count-weighted), +55.14% (capacity-MW-weighted), and +2.04%
  (hazard-weighted). Multi-correction on 135-cell test family: Bonferroni
  117/135 (86.7%); Holm 126/135 (93.3%); BH-FDR 126/135 (93.3%). DerSimonian-
  Laird random-effects pooled effects strictly positive at 95% CI on all 5
  baselines. CVaR-α at α ∈ {0.95, 0.99} preserves dominance direction. LP-vs-
  integer 0/1 gap ≤ 0.5% of theoretical ceiling in 38 of 39 countries.

  Companion papers in the SSI Index publication slate:
  - P1 (Bérard 2026, Climate Risk Management, under review) — NAM operationalisation
    as first-stage screening classifier at cohort scale
  - P2 (Bérard 2026a, Environmental Research: Energy, DOI 10.1088/2753-3751/ae87a5,
    published August 2026) — atmospheric-corrosion damage-function library
  - P3 (Bérard 2026b, Journal of Infrastructure Preservation and Resilience,
    DOI 10.1186/s43065-026-00193-z, published July 2026) — Markov degradation prior