# Methodology: Medicaid Estate Recovery Aggressiveness Scorecard

**Package date:** 2026-09-17 (America/New_York)  
**Scope:** 50 U.S. states + District of Columbia  
**Purpose:** Transparent, source-cited research table comparing how aggressively state Medicaid programs pursue estate recovery, for Family Estate Guide editorial use.

## What this is / is not

- **Is:** A reproducible synthesis of published secondary sources (MACPAC, KFF, Justice in Aging) into structured CSV + a reader-facing draft.
- **Is not:** Legal advice, a substitute for state Medicaid manuals/state plans, or a prediction of recovery in a particular estate.
- Cells marked `unknown` mean the primary sources used here did **not** support a confident state rule. **We do not invent rules.**

## Primary sources and vintages

| Source ID | Document | Vintage | Role in scorecard |
|-----------|----------|---------|-------------------|
| `kff-2024` | KFF *What is Medicaid Estate Recovery?* + appendix | Survey **March 2024**, published **Sep 13, 2024** | Recovery scope for age 55+ (all / some optional / LTSS-minimum); under-55 permanent institutionalization; survey hardship flags |
| `jia-2025` | Justice in Aging PDF *Mitigating the Harmful Effects…* | **March 2025** | Table 2 hardship waiver categories (8 pathways); Table 3 cost-effectiveness strategy categories |
| `macpac-2021` | MACPAC March 2021 Chapter 3 + Appendix 3C | State-plan review ~**2020**, report **March 2021** | TEFRA liens; cost-effectiveness narrative thresholds; older beyond-minimum flags as fallback |
| `cms-estate-recovery` | Medicaid.gov overview | Accessed 2026-09-17 | Federal baseline (not state-coded) |

**Staleness warning:** MACPAC Appendix 3C is useful but **may be stale**. Prefer KFF (2024) and Justice in Aging (2025) where they conflict on topics those newer sources cover. TEFRA lien status still leans on MACPAC until a newer 50-state lien census is available.

Local archives (with access dates) live in `sources/`. Extraction notes live in `extracted/`.

## Dimensions and weights

Composite **Aggressiveness Score** ranges **0–100** (higher = more aggressive recovery posture *as coded from sources*).

| Dimension | Nominal weight | Direction | Coding |
|-----------|----------------|-----------|--------|
| `recovers_beyond_minimum` | **25%** | Broader recovery → higher | KFF: `all_medicaid_benefits`=100; `some_optional_benefits`=55; `ltss_minimum_only`=0. If KFF blank, fall back to MACPAC non-mandatory services for 55+ (confidence medium). |
| `tefra_liens` | **15%** | Yes → higher | MACPAC 3C-1 TEFRA column: yes=100, no=0, missing=`unknown` (excluded). |
| `hardship_inverse` | **30%** | More pathways → **lower** | JIA Table 2: count of pathways / 8. `Yes`=1.0, `Yes*` (deferral-only)=0.5, `No`=0. Score = `100 * (1 - count/8)`. |
| `cost_effectiveness` | **20%** | Higher $ floor → **lower** | If numeric USD floor known: `100 * (1 - usd/50000)` capped 0–100. Else JIA category heuristics (e.g., `no_threshold`≈95). |
| `modest_home_hardship` | **10%** | Yes → **lower** | JIA modest-home pathway: yes=0, deferral=40, no=100. |

### Renormalization

If a dimension is `unknown` / blank for a state, it is **excluded** and the remaining nominal weights are renormalized to sum to 1.0 for that state. This avoids inventing a midpoint.

### Intentionally excluded from composite

**`estate_definition` (probate-only vs expanded non-probate)** is present as a CSV column but **not scored**. Primary sources discuss the federal option to expand the estate definition and give examples (e.g., Minnesota in JIA 2025) but do **not** provide a complete, consistent 50-state census in the materials archived here. Coding every state would require invention or unvetted tertiary scraping.

## Confidence flags

Each scored dimension and the composite carries `high` / `medium` / `low`:

- **high:** Newer survey/table cell directly supports the coding (KFF 2024 or JIA 2025).
- **medium:** Older MACPAC cell or numeric threshold extracted from narrative text.
- **low:** Category-only heuristic, missing MACPAC asterisk, or sparse evidence.

## Known source tensions

- **Modest-home hardship:** KFF 2024 and JIA 2025 Table 2 do not always agree. This package **prefers JIA Table 2** for pathway taxonomy and records KFF values + `modest_home_source_conflict` when they differ—for human legal review.
- **Cost-effectiveness dollars:** MACPAC 3C-3 narratives are heterogeneous (claim floors vs estate-value floors vs litigation thresholds). Numeric USD fields are best-effort encodings; category fields from JIA 2025 should be read alongside.
- **North Carolina estate floor:** JIA 2025 cites a **$50,000** estate exclusion example; MACPAC 2021 text referenced lower figures. Package uses **$50,000** with medium confidence and flags vintage conflict in STATUS.
- **Missouri / North Dakota:** JIA notes absence of tabulated hardship pathways and flags potential federal noncompliance—coded as zero pathways, which strongly raises aggressiveness; still needs counsel review.

## Reproducibility

1. Download sources into `sources/` (see `data/sources.csv`).
2. Extract text (`pdftotext` / `pdfplumber`; rotate MACPAC landscape appendix pages 90°).
3. Run `scripts/build_scorecard.py` (writes `data/*.csv` and can refresh `draft/page-body.html`).

## Editorial / legal review required before publishing claims about a specific state

See `STATUS.md`.