Claim at v1.0.0

stalled-421a-vesting-footings

8caae96aaf3e7ee8ffbb591ba5519e021188c0d30c43a18801d4ac1721725032

Data

{
  "has_outcome": true,
  "kind": "confirmation",
  "lineage": {
    "audit": "docs/lineage_audit.md; data/derived/lineage_audit.json",
    "class": "c",
    "class_meaning": "a shared cache: the query was fetched for an earlier claim and holds this claim's outcome",
    "classes": [
      "c",
      "a"
    ],
    "early_sources": [
      {
        "class": "a",
        "dataset": "ipu4-2q9a",
        "evidence": "Legacy NB permits define the jobs and foundation start dates (the cohort), cached from the construction round. Not the outcome.",
        "fetched": "2026-11-12T15:26:51-05:00",
        "name": "DOB Permit Issuance"
      },
      {
        "class": "a",
        "dataset": "rbx6-tga4",
        "evidence": "DOB NOW foundation permits define the start date and window (exposure), fetched at 11:54:48 for this claim. Not the outcome.",
        "fetched": "2026-11-13T11:54:48-05:00",
        "name": "DOB NOW: Build – Approved Permits"
      },
      {
        "class": "a",
        "dataset": "w9ak-ipjd",
        "evidence": "DOB NOW job filings define New Building jobs (the cohort). Not the outcome.",
        "fetched": "2026-11-12T15:21:36-05:00",
        "name": "DOB NOW: Build – Job Application Filings"
      },
      {
        "class": "c",
        "dataset": "bs8b-p36w",
        "earlier_claim": "construction-four-year-completion",
        "evidence": "Legacy certificates of occupancy (the outcome: no CO within 1,096 days), fetched 15:20:43 on 12 Nov for construction-four-year-completion. Referee: \"the registrant was not blind to completion data in general\"; the doc's phrase \"before any completion data were read\" \"is true only of this cohort\".",
        "fetched": "2026-11-12T15:20:43-05:00",
        "name": "DOB Certificate Of Occupancy",
        "revealed_deciding_number": "partly: four-year completion shares by first-permit year (for example 0.797 for 2017, 0.756 for 2018) were published; 36-month rates by foundation window were not"
      },
      {
        "class": "c",
        "dataset": "pkdm-hqz6",
        "earlier_claim": "construction-four-year-completion",
        "evidence": "DOB NOW certificates of occupancy (outcome), fetched 15:21:05 on 12 Nov for construction-four-year-completion; as for bs8b-p36w.",
        "fetched": "2026-11-12T15:21:05-05:00",
        "name": "DOB NOW: Certificate of Occupancy",
        "revealed_deciding_number": "partly: as for bs8b-p36w"
      }
    ]
  },
  "outcome": "stalled-421a-vesting-footings",
  "prereg_branch": {
    "name": "prereg/stalled-421a-vesting-footings",
    "sha": "ea9ff50ad9c60b3e9071ab8ff5f0b60e9bb98c75"
  },
  "prior": 0.45,
  "registered_at": "2026-11-13T11:55:23-0500",
  "registrant": "claude-opus (stalled-construction agent), 13 Nov 2026",
  "registration": {
    "alternatives": [
      "Interest-rate and cost shock of 2022-2023 affecting all jobs started in 2022 (D2 separates it partly; the 2023 cohort faces the same rates)",
      "System change: 65% of the 2022 window is DOB NOW and all controls are legacy BIS; DOB NOW CO recording or BBL coding may differ (legacy-only secondary comparison)",
      "Composition: the rush brought larger projects, which take longer than 36 months whether or not they stall",
      "The 2024 extension to 2031 removed the pressure to finish by 2026, so slower completion need not mean a stall",
      "CO recording lag in late 2026 for jobs near the 36-month horizon (affects the 2023 window most, biasing D2 down)"
    ],
    "graph_evidence": [
      {
        "concepts": [
          "vacant lots",
          "percent vacant land for 36 months or longer",
          "vacant housing demolitions"
        ],
        "direction": "n/a",
        "n_findings": null,
        "n_studies": null,
        "pubmed_ids": [
          "23814668",
          "26954939",
          "36716893"
        ],
        "summary": "The System graph holds findings on vacant land and health or crime (Philadelphia and other cities; PMIDs 23814668, 26954939, 36716893) but no finding on construction finance, tax-incentive deadlines or stalled construction. Variable searches for 'construction', 'tax', 'incentive' and 'development' on 13 Nov 2026 returned only clinical and fiscal-health variables. The graph therefore supplies the consequence side (long-vacant land) but not the mechanism tested here."
      }
    ],
    "id": "stalled-421a-vesting-footings",
    "kind": "confirmation",
    "literature": [
      {
        "citation": "NY Real Property Tax Law 421-a(16) (Affordable New York), definition of commencement date: excavation and construction of initial footings and foundations lawfully begun",
        "doi": null,
        "finding": "A project vested in the program once footings were poured on or before 15 June 2022; completion was due by 15 June 2026, extended to 15 June 2031 by the 2024 state budget for projects commenced 2016 to June 2022."
      },
      {
        "citation": "The Real Deal, 20 Jan 2023, 'Expired 421a deadline jeopardizes 33k housing units' (REBNY member survey)",
        "doi": null,
        "finding": "About 33,000 units in 72 vested projects were at risk of missing the 2026 completion deadline; lenders wanted a 12-month buffer."
      },
      {
        "citation": "Bisnow, May 2024, 'We're ready to go: stalled projects get new life with 421-a extension'",
        "doi": null,
        "finding": "Developers who vested but could not finance construction in time stalled and restarted after the 2031 extension."
      },
      {
        "citation": "REBNY, 2022, new-building filings report",
        "doi": null,
        "finding": "689 NB filings in Q1 2022, the most since 2014."
      },
      {
        "citation": "NYC DOB Stalled Sites Program (Local Law 2009); The Real Deal 2010 and 2014 reports",
        "doi": null,
        "finding": "After 2008 the city listed about 600 stalled sites and offered permit extensions; 364 restarted by May 2010. The program ended in 2013."
      }
    ],
    "plan": "Script analyses/stalled/vesting_cohorts.py (committed with this registration), mode 'run'. Unit: NB job, legacy BIS (ipu4-2q9a FO INITIAL permits on NB jobs, cached data/big/construction/permits_nb_legacy.parquet) and DOB NOW (rbx6-tga4 work_type Foundation, Initial Permit, on New Building jobs in w9ak-ipjd). Start = first initial foundation permit. Outcome = no CO (bs8b-p36w NB rows or pkdm-hqz6) on the same BBL within 1,096 days. Windows 15 Dec (y-1) to 14 Jun y for y = 2017..2023; 2020 and 2021 are reported but not used (COVID). Bootstrap 2,000 draws, seed 20261113. Window sizes before the outcome was read: 2017 541, 2018 516, 2019 497 (controls 1,554, all legacy), 2022 807 (521 DOB NOW, 286 legacy), 2023 366. Power check (rule 8): with a stall rate near 0.4 the standard error of D1 is about 0.021 and of D2 about 0.031, so an effect of 0.10 gives z_null about 4.8 for D1 (D = 23) and 3.2 for D2 (D = 10); D1 discriminates for effects above about 0.06 and D2 for effects above about 0.06. A secondary, non-deciding comparison restricts all windows to legacy BIS jobs.",
    "prediction": {
      "direction": "+",
      "magnitude": "Share of jobs with no CO on the BBL within 1,096 days of the first initial foundation permit. Registrant's expectation: D1 = rate(15 Dec 2021-14 Jun 2022) - rate(same window 2017-2019 pooled) between +0.05 and +0.15; D2 = rate(2022 window) - rate(15 Dec 2022-14 Jun 2023) between +0.03 and +0.12.",
      "refute_if": "upper 95% bootstrap bound of D1 < 0.03 (no excess stalling in the rush cohort), OR upper 95% bound of D2 < 0 (the 2023 cohort, which could not vest, stalled more; that points to the 2022-2023 rise in interest rates and costs rather than footings-only vesting)",
      "support_if": "lower 95% bootstrap bound of D1 > 0.03 AND lower 95% bootstrap bound of D2 > 0"
    },
    "prior": 0.45,
    "registered_at": "2026-11-13T11:55:23-0500",
    "registrant": "claude-opus (stalled-construction agent), 13 Nov 2026",
    "sha256": "0ff6d1daae72533b4326615a6c4b725cd04e1dc2ef73c315fde3a3a2df7c1628",
    "statement": "New-building jobs that began foundation work in the six months before the 421-a vesting deadline of 15 June 2022 failed to reach a certificate of occupancy within 36 months more often than jobs that began foundations in the same calendar window of 2017-2019, and more often than jobs that began them in the same window of 2023, because the program counted a project as commenced once its footings were poured."
  },
  "registration_commit": {
    "commit": "ea9ff50ad9c60b3e9071ab8ff5f0b60e9bb98c75",
    "committed_at": "2026-11-13T11:55:24-05:00"
  },
  "registration_file_sha256": "e1f9ca0f68822b303fa1c8c0ef9fb88f58b1361a2d5d096f342d91338eaabafe",
  "registration_path": "claims/registered/stalled-421a-vesting-footings.json",
  "registration_sha256": "0ff6d1daae72533b4326615a6c4b725cd04e1dc2ef73c315fde3a3a2df7c1628",
  "registration_text": "{\n \"id\": \"stalled-421a-vesting-footings\",\n \"kind\": \"confirmation\",\n \"statement\": \"New-building jobs that began foundation work in the six months before the 421-a vesting deadline of 15 June 2022 failed to reach a certificate of occupancy within 36 months more often than jobs that began foundations in the same calendar window of 2017-2019, and more often than jobs that began them in the same window of 2023, because the program counted a project as commenced once its footings were poured.\",\n \"graph_evidence\": [\n  {\n   \"concepts\": [\n    \"vacant lots\",\n    \"percent vacant land for 36 months or longer\",\n    \"vacant housing demolitions\"\n   ],\n   \"n_findings\": null,\n   \"n_studies\": null,\n   \"direction\": \"n/a\",\n   \"summary\": \"The System graph holds findings on vacant land and health or crime (Philadelphia and other cities; PMIDs 23814668, 26954939, 36716893) but no finding on construction finance, tax-incentive deadlines or stalled construction. Variable searches for 'construction', 'tax', 'incentive' and 'development' on 13 Nov 2026 returned only clinical and fiscal-health variables. The graph therefore supplies the consequence side (long-vacant land) but not the mechanism tested here.\",\n   \"pubmed_ids\": [\n    \"23814668\",\n    \"26954939\",\n    \"36716893\"\n   ]\n  }\n ],\n \"literature\": [\n  {\n   \"citation\": \"NY Real Property Tax Law 421-a(16) (Affordable New York), definition of commencement date: excavation and construction of initial footings and foundations lawfully begun\",\n   \"doi\": null,\n   \"finding\": \"A project vested in the program once footings were poured on or before 15 June 2022; completion was due by 15 June 2026, extended to 15 June 2031 by the 2024 state budget for projects commenced 2016 to June 2022.\"\n  },\n  {\n   \"citation\": \"The Real Deal, 20 Jan 2023, 'Expired 421a deadline jeopardizes 33k housing units' (REBNY member survey)\",\n   \"doi\": null,\n   \"finding\": \"About 33,000 units in 72 vested projects were at risk of missing the 2026 completion deadline; lenders wanted a 12-month buffer.\"\n  },\n  {\n   \"citation\": \"Bisnow, May 2024, 'We're ready to go: stalled projects get new life with 421-a extension'\",\n   \"doi\": null,\n   \"finding\": \"Developers who vested but could not finance construction in time stalled and restarted after the 2031 extension.\"\n  },\n  {\n   \"citation\": \"REBNY, 2022, new-building filings report\",\n   \"doi\": null,\n   \"finding\": \"689 NB filings in Q1 2022, the most since 2014.\"\n  },\n  {\n   \"citation\": \"NYC DOB Stalled Sites Program (Local Law 2009); The Real Deal 2010 and 2014 reports\",\n   \"doi\": null,\n   \"finding\": \"After 2008 the city listed about 600 stalled sites and offered permit extensions; 364 restarted by May 2010. The program ended in 2013.\"\n  }\n ],\n \"prediction\": {\n  \"direction\": \"+\",\n  \"magnitude\": \"Share of jobs with no CO on the BBL within 1,096 days of the first initial foundation permit. Registrant's expectation: D1 = rate(15 Dec 2021-14 Jun 2022) - rate(same window 2017-2019 pooled) between +0.05 and +0.15; D2 = rate(2022 window) - rate(15 Dec 2022-14 Jun 2023) between +0.03 and +0.12.\",\n  \"support_if\": \"lower 95% bootstrap bound of D1 > 0.03 AND lower 95% bootstrap bound of D2 > 0\",\n  \"refute_if\": \"upper 95% bootstrap bound of D1 < 0.03 (no excess stalling in the rush cohort), OR upper 95% bound of D2 < 0 (the 2023 cohort, which could not vest, stalled more; that points to the 2022-2023 rise in interest rates and costs rather than footings-only vesting)\"\n },\n \"plan\": \"Script analyses/stalled/vesting_cohorts.py (committed with this registration), mode 'run'. Unit: NB job, legacy BIS (ipu4-2q9a FO INITIAL permits on NB jobs, cached data/big/construction/permits_nb_legacy.parquet) and DOB NOW (rbx6-tga4 work_type Foundation, Initial Permit, on New Building jobs in w9ak-ipjd). Start = first initial foundation permit. Outcome = no CO (bs8b-p36w NB rows or pkdm-hqz6) on the same BBL within 1,096 days. Windows 15 Dec (y-1) to 14 Jun y for y = 2017..2023; 2020 and 2021 are reported but not used (COVID). Bootstrap 2,000 draws, seed 20261113. Window sizes before the outcome was read: 2017 541, 2018 516, 2019 497 (controls 1,554, all legacy), 2022 807 (521 DOB NOW, 286 legacy), 2023 366. Power check (rule 8): with a stall rate near 0.4 the standard error of D1 is about 0.021 and of D2 about 0.031, so an effect of 0.10 gives z_null about 4.8 for D1 (D = 23) and 3.2 for D2 (D = 10); D1 discriminates for effects above about 0.06 and D2 for effects above about 0.06. A secondary, non-deciding comparison restricts all windows to legacy BIS jobs.\",\n \"alternatives\": [\n  \"Interest-rate and cost shock of 2022-2023 affecting all jobs started in 2022 (D2 separates it partly; the 2023 cohort faces the same rates)\",\n  \"System change: 65% of the 2022 window is DOB NOW and all controls are legacy BIS; DOB NOW CO recording or BBL coding may differ (legacy-only secondary comparison)\",\n  \"Composition: the rush brought larger projects, which take longer than 36 months whether or not they stall\",\n  \"The 2024 extension to 2031 removed the pressure to finish by 2026, so slower completion need not mean a stall\",\n  \"CO recording lag in late 2026 for jobs near the 36-month horizon (affects the 2023 window most, biasing D2 down)\"\n ],\n \"prior\": 0.45,\n \"registrant\": \"claude-opus (stalled-construction agent), 13 Nov 2026\",\n \"registered_at\": \"2026-11-13T11:55:23-0500\",\n \"sha256\": \"0ff6d1daae72533b4326615a6c4b725cd04e1dc2ef73c315fde3a3a2df7c1628\"\n}",
  "statement": "New-building jobs that began foundation work in the six months before the 421-a vesting deadline of 15 June 2022 failed to reach a certificate of occupancy within 36 months more often than jobs that began foundations in the same calendar window of 2017-2019, and more often than jobs that began them in the same window of 2023, because the program counted a project as commenced once its footings were poured."
}

History in this collection

  1. addedv1.0.0Nov 14, 2026this version