The Presidential Economic Pressure Index is the first defined external adapter for GII-US. PEPI is an upstream discovery and evidence source, referenced neutrally: it can nominate candidate events, supply provenance links, implementation milestones, and candidate trade or price observations. It is not a GII subindex, and its numbers never enter GII scores.
The construct boundary#
PEPI measures presidential economic-policy follow-through — implementation, delay, dilution, reframing, reversal, or abandonment. GII measures attributable consequences across nine institutional origins and six stakeholder domains. The following are release-blocking controls, not conventions:
| Rule | Reason |
|---|---|
| No PEPI composite, event score, or reversal magnitude enters GII-M or GII-D | Different constructs; follow-through is not impact. |
| PEPI confidence never maps to GII-C | Source-classification certainty is not causal-attribution confidence. Stored as source_classification_confidence. |
| Status, deadline, or resolution is never proof of legal effect or exposure | Announcement clocks are not implementation clocks. |
| Pending and indeterminate never become zero | Missing and unresolved are states, not numbers. |
| No double counting of shared underlying observations | The same market, price, or trade source cannot enter a GII estimate twice. |
Neutral source referencing#
First-party source disclosure#
PEPI is developed by the GII author and published by API Automations, which supplies the live index platform for this research program. The relationship is disclosed as a potential conflict of interest and mitigated structurally by the boundary rules above; external methods review of the boundary controls is a Phase 2 requirement.
Walking skeleton#
The adapter is implemented in @gii/adapters: it preserves raw payloads, diffs incrementally, quarantines anything it cannot trust, transforms survivors through the crosswalk, and reconciles them against the registry crosswalk. The summary below is a live run over a fictional four-event batch — it computes on fixtures only and promotes nothing.
| Release-blocking boundary control |
|---|
| Do not copy a PEPI composite, event score, reversal magnitude, or outcome-type value into GII-M or GII-D. |
| Do not map PEPI confidence to GII-C; store it as source_classification_confidence. |
| Do not treat a PEPI status, deadline, announcement, or resolution as proof of legal effect, exposure, causal attribution, or direction. |
| Do not convert pending or indeterminate to zero. |
| Do not double count PEPI market, price, trade, or sector observations already entering a GII estimate. |
Adapter health#
In production, this page will publish adapter coverage, last successful run, quarantines, and version-compatibility state — without exposing credentials or secrets. The adapter is acceptable for pilot use only when it passes fixture and live-contract tests for pagination, idempotency, status transitions, schema drift, staleness, checksum change, rate-limit recovery, and log redaction — all covered by the @gii/adapters test suite.