Name: UI_INDEX Methodology & Data Validation
Creator: Sierra Napier
Published: 2026-06-11
License: https://github.com/thedatavigilante/UI_INDEX/blob/main/LICENSE

Every statutory value driving the indices was checked against a live primary source on 2026-06-11. This is the assertion table the analysis notebook runs on load.

Data Point	Value Used	Live Source	Status
MD max WBA 2026	$430	MD DLLR Benefits Schedule	✅ EXACT
VA max WBA 2026	$430	VEC SB 1056 (eff. Jan 4 2026)	✅ CORRECTED
DC max WBA 2026	$444	DC.gov Unemployment	✅ EXACT
MD SUI wage base 2026	$8,500	EY Tax News	✅ EXACT
VA SUI wage base 2026	$8,000	EY Tax News	✅ EXACT
DC SUI wage base 2026	$9,000	EY Tax News	✅ EXACT
David Trone FEC receipts	$63.83M	FEC.gov S4MD00319	✅ EXACT
DC avg annual wage 2026	$112,400	BLS QCEW (data vintage in employer gap calc)	⚠️ VINTAGE NOTE
Weekly housing cost	see note	HUD FMR / market-rate proxy	⚠️ METHOD-SENSITIVE

✅ EXACT = matches primary source to the dollar. ⚠️ = correct but methodology-dependent; see notes below.

Index Formulas

BAI = Max Weekly Benefit ÷ Median Weekly Housing Cost | Below 1.0 = systemic failure
WBI = Taxable Wage Base ÷ Average Annual Wage
MIPI = (Earnings − Disregard Threshold) ÷ Max Weekly Benefit
Housing Gap = Weekly Housing − Max Weekly Benefit ($/week deficit)
Employer Gap = (Expected Base − Statutory Base) × DOL Effective Rate × Covered Employment

Real Value Index — what a frozen benefit is really worth

A frozen number isn't static — inflation cuts it every year, without a vote. The Real Value Index adjusts each frozen maximum benefit by BLS CPI-U to its 2026 purchasing power.

RVI = Max WBA ÷ (CPI₂₀₂₆ ÷ CPI_base_year)

Real Value Index — frozen benefits vs. real 2026 value

State	Frozen at	Since	Real value 2026	Erosion
Maryland	$430	2014	~$306	−29%
Virginia	$378	2008	~$244	−35% (worst)
DC	$444	2018	~$335	−25%

Method: BLS CPI-U annual averages; 2014→2026 anchor 40.67% (in2013dollars/BLS). Virginia froze in 2008 — the longest freeze and worst erosion. Confirm exact figures against the BLS calculator before citing to the dollar.

Why Comparative Static Baselines?

This analysis anchors to three specific years — 2010, 2018, 2026 — rather than a continuous time-series. Benefit caps, wage bases, and disregard thresholds change through discrete legislative acts, not gradual drift. Anchoring to the same three moments across all three jurisdictions makes the policy decay visible and comparable. A moving average would obscure the step-function nature of the neglect — the whole point is that these numbers don't move for years at a time, then jump.

Housing Cost Methodology — The Most Consequential Choice

This is the single most important methodological decision in the audit, and it deserves explicit disclosure. The Weekly_Housing column determines whether a jurisdiction's BAI lands above or below the 1.0 survival threshold. With Virginia's corrected $430 cap, the VA 2026 outcome depends entirely on which housing source is used:

HUD FMR 2-BR state median (~$293/wk): VA BAI = 1.47 (well above)
Market-rate average (used in this audit, $420/wk): VA BAI = 1.02 (barely above)
Northern VA market rate (~$550+/wk): VA BAI = 0.78 (below)

This audit uses a market-rate weekly proxy rather than HUD FMR, on the reasoning that claimants face market rents, not subsidized FMR ceilings. This is a deliberately conservative-to-middle choice. A HUD-only methodology would make every jurisdiction look better; a Northern VA / DC-metro methodology would make them look worse. The honest statement: the direction of decay is robust across all three methods; the exact threshold-crossing year for Virginia is methodology-sensitive.

Honest Limitations

FEC contribution categorization uses keyword matching on contributor names — approximate, and captures only 0–30% of each member's total. See the Political Layer methodology disclosure for the full accounting and the correction roadmap.
Self-funding detection currently uses last-name matching and over-excludes joint-committee transfers for Kaine, Warner, and Hoyer. Only David Trone has verified candidate self-loans. Correction logged in OPTIMIZATIONS.md #23.
Employer gap uses the 2010-ratio method (Method B). Alternative methodologies (full wage base, CPI-only, national peer average, wage-growth index) produce different dollar figures — see source code for scenarios A–E.
BLS wage and employment data are 2024 annual averages; 2026 figures are projected forward.
Committee assignments reflect the 119th Congress (2025–2027) and change each Congress.
OpenSecrets API is blocked by Cloudflare challenge — official industry contribution codes were unavailable without manual registration.
Off-cycle candidates: Warner (2026) and Van Hollen (~2028) were not 2024 candidates; their FEC figures reflect off-cycle committee maintenance.

Data Sources & Citations

Source	Data Used	Status
BLS QCEW	Average annual wages, covered employment	✅ LIVE
USDOL / State DOL Statutes	SUI wage bases, max weekly benefits, disregard thresholds	✅ VERIFIED
DOL UI Financial Data	Historical effective SUI tax rates by state/year	✅ VERIFIED
HUD Fair Market Rents	FMR by metro → weekly housing proxies	✅ VERIFIED
FEC Open API (v1)	Campaign receipts, Schedule A itemized contributions	✅ LIVE
Census ACS 2022	Median household income by congressional district	✅ LIVE
Congress.gov	Committee assignments, member metadata (119th Congress)	⚠️ RATE-LIMITED
BLS LAUS — Local Area Unemployment Statistics	State + county unemployment rates (2010/2018/2026) — county choropleth map	✅ LIVE
USASpending.gov	Federal UI grants by state (CFDA 17.225) — spending accountability analysis	✅ LIVE
FRED (St. Louis Fed)	CPI-U, DC metro CPI (CUURA311SA0), PCE — independent inflation cross-validation	✅ LIVE
OpenSecrets	Industry codes, member net worth	❌ BLOCKED

All sources are public record. Every figure is reproducible from the source code at github.com/thedatavigilante/UI_INDEX. See DATA_CATALOG.md for full file lineage and the OPTIMIZATIONS.md running correction log.

🔬 Methodology & Data Validation

Data Validation — Live-Source Verified