Pre-Registration — Solar Phase → Sovereign-Default Onsets, INDEPENDENT SAMPLE

Registered: 2026-09-27, before any analysis was run on either dataset below. Column layouts were inspected to write this document; no phase assignment, count or statistic has been computed. Why this document exists: two registered tests on the Reinhart–Rogoff chronology (SOLAR_PHASE_CRISIS_TEST_RESULT_2026-09-17.md, SOLAR_PHASE_SOVEREIGN_DEFAULT_TEST_RESULT_2026-09-17.md) found a consistent 12–16 % excess of sovereign-default onsets in the declining phase of the solar cycle that never reached the declared threshold on the conservative count (p = 0.051–0.066). A tilt seen twice in one dataset is a lead, not a finding. This registration tests it on chronologies compiled by other people, by other rules, at monthly rather than annual resolution.

Status: RUN 2026-09-27 — REFUTED on this sample (primary ratio 0.99, p = 0.59). Result: SOLAR_PHASE_DEFAULTS_INDEPENDENT_SAMPLE_RESULT_2026-09-27.md. Nothing below was revised. Revisions require a new dated document and both remain public. This is a replication attempt and is reported with the two earlier tests.

---

1. Hypothesis, one direction only, unchanged

H1: Sovereign-default onsets fall in the declining phase of the solar cycle (SILSO maximum to the following minimum) at a rate above the phase's share of time. H0: onsets are distributed across phase in proportion to phase length.

The effect size the earlier tests suggest is a ratio of about 1.12–1.16 (observed ÷ expected). A replication that returns a ratio near 1.0 refutes; a ratio near 1.14 with p ≥ 0.05 on this sample is "consistent, still not established"; p < 0.05 with the ratio in the registered direction supports.

2. Data — fixed; both compiled without reference to solar activity and independently of Reinhart–Rogoff

Primary: Asonuma & Trebesch, monthly default and restructuring database, May 2026 update (sites.google.com/site/christophtrebesch/data; JEEA 2016). 217 spells, 1970–2026, foreign private creditors only, monthly start dates. Their README states the dataset "differs … from existing ones, e.g. by Standard & Poor's (2006) and Reinhart and Rogoff (2009)" and is coded from press, IMF archives and arrears data. Sheet DATASET Defaults & Restruct.; onset field = "Start of default or restructuring".

Secondary: Bank of Canada – Bank of England Sovereign Default Database, 2025 update (bankofcanada.ca, 1960–2024, annual, all creditor classes including official and domestic). Onset = the first year a country's total debt in default is positive after a year in which it was zero or absent. Dated to mid-year, as in the first test, because the data are annual.

Independent-sample replication — RESULT: the sovereign-default tilt does not replicate

Registration: PREREG_SOLAR_PHASE_DEFAULTS_INDEPENDENT_SAMPLE_2026-09-27.md, committed before the script existed, unchanged. Run: 2026-09-27. Script: scripts/research/prereg_defaults_independent_sample_test.py (re-runs from the two public downloads and SILSO in one command; raw output beside it). Verdict under §5: REFUTED on this sample. Primary ratio 0.99 (p = 0.59). No row in the registered direction reaches the threshold except the BoC–BoE country-level event count, which is the inflated count the earlier registrations already set aside, and the registration names it as secondary.

Events loaded — A–T spells 217, onset events after the 12-month rule 189 (116 post-default, 73 preemptive); BoC–BoE countries 166, onset events 418.

Row	n	in declining	expected	ratio	p one-sided (H1)	p reverse	weighted share obs / exp	p permutation
A–T onset-MONTHS (PRIMARY)	118	76	76.7	0.99	0.592	0.483	0.674 / 0.650	0.327
A–T onset events (country-level)	178	120	115.6	1.04	0.274	0.776	—	—
A–T post-default onsets only	109	74	70.8	1.04	0.298	0.769	—	—
A–T preemptive onsets only	69	46	44.8	1.03	0.438	0.659	—	—
BoC–BoE onset-YEARS