# MacroVision v94: UAS employment/transaction linkage metadata audit
29 October 2026. Public documentation only; no UAS microdata, participant IDs, account access, outcomes or estimates.

## Decision
Advance UASFin from a broad candidate to a **conditional linkage candidate**: public employment instruments contain potentially useful event questions. Do not approve estimation. Actual survey/transaction intersection, complete coverage and access remain unknown. This is progress in measurement feasibility, not evidence for the thesis mechanism.

## Verified public anchors
| Instrument | Metadata / fields | Interpretation boundary |
|---|---|---|
| UAS625 | July 2024 monthly survey, fielded July 1–31; catalog reports 9,859 observations. Routing includes empl3 and empl_when_month/day/year; year can be assigned 2024 when empty. | Survey completion count is not job losses or linked households. Defaulted year is not necessarily respondent-reported. |
| UAS782 | March 2026 monthly survey, fielded March 1–31; catalog reports 9,661 observations. empl1 describes change; empl3 includes business closure and fired/laid-off options; employment-change date fields are present. | Multiple reasons and approximate dates require adjudication; job change and hour reduction are not necessarily job loss. |
| UAS548 | Plaid Consent Survey 6; April 17, 2023–January 14, 2024. Custom invited sample 1,139; completed 891. Special-project invitation available on request. | These are invitation completions, not consents, linked accounts, continuous transaction coverage or employment-survey intersection. |
| UAS548 routing | sc_001 screens online banking; consent records willingness. Subsequent steps create credentials and connect institutions. | Consent is not completed linkage; linked accounts may later be removed. |

No observed intersection count is available from this audit. Consent fieldwork ending before later employment surveys proves only calendar ordering of invitations; it does not establish transaction continuation into 2024 or 2026. No transaction start/end dates, account-level missingness table, balance snapshot dictionary or approved researcher linkage key were obtained.

## Proposed source-specific event adjudication — not an estimator
The unchanged v91 target requires involuntary separation, baseline paid employment, an index employee and household coverage.
- Candidate branch: empl1 indicates working-to-not-working, with empl3 recording closure or fired/laid-off. First verify export encoding for multiple selections, routing and missing responses.
- Do not treat empl1 job changes, full-to-part-time, retirement or other transitions as the same primary exposure. Preserve their counts separately in an authorized audit.
- A selected closure reason does not by itself prove permanent employer closure or eliminate confounding.
- Retain mixed voluntary/involuntary or conflicting reasons as unresolved until a rule is frozen before outcomes. Never choose the favorable reason.
- Approximate day/month/year, defaulted year and disagreement with survey reference windows must be flagged, not silently converted to exact dates.
- fin3 combines job loss with hours/income reductions and is not an interchangeable primary exposure.
- A no-change answer covers the stated reference window, not necessarily the entire required control window. Check actual previous response date and reference-text generation.
- These instruments refer to the respondent's change; highest-earning baseline index employee and partner job losses still require a household roster/history crosswalk.

These are proposed adjudication rules based on metadata. No claim is made that the necessary baseline paid-employment or household fields have already been mapped.

## Required provider-side audit inputs
1. Approved transaction-product access, linkage key definition and permitted environment.
2. Monthly active/complete account coverage and observed balance timestamps, distinguishing historical backfill from contemporaneous recording.
3. Authorized linkage with each monthly employment instrument and actual interview/reference dates; deduplicate repeated reports of the same event.
4. Aggregate-only flow: invited -> consented -> connected -> complete baseline -> eligible employment response -> valid event/control -> complete follow-up. Define denominators; do not conflate survey response with financial linkage.
5. Joint event/buffer cell counts, coverage of partner accounts, cash and benefit-funded purchases, pure-food classification validation and refund/transfer rules.
6. Appropriate weights/variance and a source-specific power plan fixed before outcome contrasts.

Record unavailable items explicitly. Do not impute the entire household from a single account or count repeated monthly surveys as independent households. No approved microdata application, contact or DUA submission is performed.

## Access boundary and recommended next step
UAS policy discussed in v93 excludes microdata from AI tools. Any authorized record-level audit must be run by an approved researcher in a permitted environment; only permitted code/metadata and aggregate output can return to this workflow.

Next actionable deliverable: an institution-ready access concept note, including the fixed research question, minimal requested files and the aggregate feasibility flow above. It can be drafted without sending or registering. More public questionnaire browsing cannot establish the missing transaction intersection. Source-specific estimation remains paused until access and coverage are confirmed.

## Reproducible evidence scope
Documentation inspected on October 29, 2026 via public official catalog/routing search excerpts. Some direct UAS625/UAS782 pages timed out; those excerpts do not certify a pinned file release or full routing reconciliation. UAS548 routing was directly readable. No survey data were scraped/downloaded and no raw records are included.

Public reference URLs:
- UAS625 catalog: https://uasdata.usc.edu/index.php?pid=Surveys+by+experiment&type=3
- UAS782 catalog: https://uasdata.usc.edu/index.php?pid=Surveys+by+topic&type=3
- UAS548 metadata: https://uasdata.usc.edu/index.php?ssid=UAS+548&type=4
- UAS625 routing: https://uasdata.usc.edu/output/paperversion/index.php?r=eNpLtDK2qi62MrFSKkhMT1WyLrYyBLFLk3Iyk_WK8ktLMvPSQaJAweLizBQQ09hKyczIFCZYChE0tFIyVLKuBVwwx2oXUg
- UAS782 routing: https://uasdata.usc.edu/output/paperversion/index.php?r=eNpLtDK2qi62MrFSKkhMT1WyLrYyBLFLk3Iyk_WK8ktLMvPSQaJAweLizBQQ09hKydzCCCZYChE0tFIyVLKuBVwwx84XVg%2C%2C
- UAS548 routing: https://uasdata.usc.edu/output/paperversion/index.php?r=eNpLtDK2qi62MrFSKkhMT1WyLrYyBLFLk3Iyk_WK8ktLMvPSQaJAweLizBQQ09hKydTEAiZYChE0tFIyVLKuBVwwx8YXVg%2C%2C
- [Prior access comparison](TRANSACTION_DATA_FEASIBILITY_V93.md)

Validation: identifiers/dates checked against official public metadata, with fetch limitations disclosed. No new computation; site CI and post-live public-asset checks required.