# L2 — PRD / CTH dataset feasibility outline (NO TESTS) — v0.2 variable inventory

**Status:** concrete variable names/coverage from local SCA + ANES + NY Fed SCE public docs · **H1–H4 UNTESTED** · MODEL_CHANGE=NO  
**Seed job:** `7a01871f-2615-43a0-8d5f-a70ca8a7ae42` (PRD backlog) · lane `lane-L2-prd-cth-feasibility` · wave ACK `b8c1c81e`  
**Related:** `docs/prd/WAVE2C_PRD_COMMITMENT_THRESHOLD_BACKLOG.md`, `docs/m1/WAVE2B_M1_LONG_FEASIBILITY.md`

## Non-actions
- Do **not** estimate thresholds, interactions, or hysteresis.
- Do **not** use political transition / election labels as DVs.
- Do **not** treat this outline as evidence of PRD existence.

---

## Required triad (feasibility framing only)

| Role | Meaning for PRD/CTH |
|------|---------------------|
| Objective household | Admin/survey measures of conditions (employment, income, inflation) |
| Subjective perception | Self-reported current assessment / hardship / sentiment |
| Political alignment | PID / lean / ideology / affect — **predeclared**, not shopped |

Nonlinear CTH / hysteresis remain **unidentifiable** from this outline alone.

---

## 1. Michigan SCA (priority lead) — concrete locals

### 1.1 Landed micro extract (`data/raw/michigan_sca/microdata/sca_micro_2015_2023_pilot.csv`)
| Variable | Role | Coverage in extract |
|----------|------|---------------------|
| `CASEID` / `ID` | Case keys (not permanent panel ID) | 2015–2023 |
| `YYYYMM` | Interview month | 108 months continuous |
| `IDPREV` / `DATEPR` | Prior-interview linkage | present |
| `IDPREV2` / `DATEPR2` | Second-prior linkage | present (IDPREV2 often blank historically) |
| `PEXP` | Subjective **expectation** (year ahead) | present |
| `PAGO` | Subjective **current vs year ago** | present |
| `ICS` | Index of Consumer Sentiment (person-level component input) | present |
| `WT` / `WT_HH` | Design weights | present |
| `AGE` / `SEX` / `EDUC` / `REGION` | Demographics | present |
| `SAMPLE` / `METHOD` | Sample/method flags | present |
| **PID / ideology / party lean** | Political alignment | **ABSENT in this 18-col pilot extract** |

**Row count:** 63,748 respondents · **108** unique YYYYMM · Codex metadata `e01abcea`.

### 1.2 Public aggregates (local)
| File | Variables (concept) | Span |
|------|---------------------|------|
| `tbmics.csv` | `ICS_ALL` monthly | 1952–2026 (676 months; early years incomplete) |
| `table6_pago_monthly.csv` | PAGO relative | ~1978–2026 (584 rows) |
| `table8_pexp_monthly.csv` | PEXP relative | ~1978–2026 (584 rows) |
| `tbmiccice.csv` / Table 5 | `ICC` (current), `ICE` (expected) | monthly modern era |
| Official wording | PAGO / PEXP 12-month horizons | continuous core items |

### 1.3 PRD open questions (still)
| Item | Status |
|------|--------|
| Objective items inside SCA micro (income/employment/inflation-expect) | **Need fuller public micro codebook** — not in 18-col pilot |
| Political PID in SCA | **UNKNOWN / likely sparse or absent** in standard public files — audit required before claiming triad |
| 2024 phone→web | Mode strata mandatory if series extends past Jun 2024 |
| Panel attrition | Documented in M1-LONG v2 (origin≤2019 linked pairs etc.) — reuse diagnostics; **no PRD test** |

---

## 2. NY Fed SCE — concrete public-doc variables

**Sources inspected:** SCE core questionnaire PDF (NY Fed CMD); M1-LONG feasibility notes; landing `newyorkfed.org/microeconomics/sce`.

### 2.1 Panel geometry
| Variable | Role |
|----------|------|
| `userid` | Stable panel ID (public micro Excel) |
| `tenure` | Months in panel |
| `weight` | Survey weight |
| History | Mid-**2013** → present (monthly) |
| Typical spacing | ~**11** calendar months tenure 1→12 vs 12m question wording |

### 2.2 Subjective (core module)
| Code | Construct |
|------|-----------|
| **Q1** | Financially better/worse than 12 months ago (realization analogue to PAGO) |
| **Q2** | Expect better/worse 12 months ahead (expectation analogue to PEXP) |
| Q1intro* | Intro stems new vs repeat respondents |

### 2.3 Objective / household (core module — examples)
| Code | Construct |
|------|-----------|
| **Q10** | Current employment situation (FT/PT/unemployed/…) |
| Q11 / Q12new / ES* | Jobs count; employee vs self-employed |
| Q13new / Q14new | Job-loss / quit probability next 12m |
| Q15 / Q16 | Unemployment spell length |
| Household income expectations | Forward 12m total HH income direction/percent |
| Earnings items | Current job earnings expectations |
| Inflation block | Q1A / Q1A_part2 inflation/deflation expectations |
| Credit / housing price expectations | C-series / credit access |

### 2.4 Demographics (core)
Age, gender, HH composition, HH income category — present in questionnaire.

### 2.5 Political alignment
| Finding | Implication |
|---------|-------------|
| **No party ID / ideology / vote items located in SCE core questionnaire PDF** | Core SCE **does not** by itself complete the PRD triad |
| Other SCE modules | **Codebook audit still required** before asserting political vars exist in public micro |
| Feasibility stance | SCE = strong objective+subjective panel; political leg likely needs **ANES (or other)** — not silent merge |

---

## 3. ANES — concrete political + economic co-measure vars

**Landed extract:** `data/raw/anes/extracts/anes_cdf_s1_minimal_pid_ft.csv` (73,745 rows).

### 3.1 Political (present in extract)
| Variable | Construct |
|----------|-----------|
| `VCF0004` | Study year (1948–2024; 33 years in full extract; 18 years both leans+party FTs for S1 series) |
| `VCF0301` / `VCF0302` / `VCF0303` / `VCF0305` | Party ID / lean family |
| `VCF0316` / `VCF0320` / `VCF0322` | Ideology / related |
| `VCF0218` / `VCF0224` | Dem/Rep party feeling thermometers |
| `VCF0201` / `VCF0202` / `VCF0290` / `VCF0291` | Additional affect/people FTs in extract |
| `VCF0009z` (+ `VCF0010z`/`VCF0011z`) | Design weights |
| `VCF0017` | Mode |

### 3.2 Economic objective/subjective in ANES
| Status | Note |
|--------|------|
| Not in **minimal** S1 extract | Full CDF has economic items (income, unemployment, retrospective/prospective personal finance) — **requires separate variable list from codebook Varlist PDF** before claiming co-presence by year |
| Cadence | Election-cycle — **weak** for high-frequency CTH/hysteresis |
| Pew affective | **Separate lineage** — never splice into ANES for PRD either |

---

## 4. Cross-source feasibility matrix (NO TESTS)

| Triad element | SCA micro pilot | SCA aggregates | SCE core | ANES CDF |
|---------------|-----------------|----------------|----------|----------|
| Subjective current | PAGO | Table 6 | Q1 | (CDF econ items — audit) |
| Subjective expectation | PEXP | Table 8 | Q2 | (CDF — audit) |
| Objective employment/income | **not in pilot cols** | no | Q10 + income/earnings | CDF — audit |
| Political PID/lean | **absent in pilot** | no | **not in core PDF** | **VCF030x yes** |
| Panel linkage | IDPREV/DATEPR | no | userid/tenure | mostly cross-section |
| Best PRD role | Subjective + panel geometry | Density backdrop | Obj+subj panel | Political density |

**Implication (feasibility only):** No single landed source currently exposes the full objective+subjective+political triad in one micro file on disk. Plausible future designs (still untested): (a) richer SCA micro with political if codebook confirms; (b) SCE obj+subj + external political (hard — linkage); (c) ANES within-survey triad at election cadence (weak dynamics).

---

## 5. Hypotheses (registered only — DO NOT TEST)
- H1 subjective ≠ objective systematically  
- H2 incumbent-party alignment may condition relationship  
- H3 Commitment Threshold (stronger prior commitment → larger adverse objective change needed)  
- H4 deterioration vs recovery thresholds may differ (hysteresis)

## Explicit non-claims
No threshold estimates. No interactions. No holdout evaluation. No evidence of PRD. MODEL_CHANGE=NO.