Answers are grounded in this paper and cite their sources.
I build a country-year notification-based phase-in index of WTO Trade Facilitation Agreement commitments for 2015–2023. The clock is common; only notified A/B/C shares vary, so the index records scheduled notification intensity, not border implementation. Across eleven baseline tests (725 / 105 at the two-year lag; 829 / 106 at one year), none survives Holm’s family-wise correction (survivors = 0). Two are nominally significant (exploratory): merchandise at 0.0082 and trade-to-GDP at 0.0416 (adjusted 0.0903 and 0.416). Neither is robust: the trade-to-GDP estimate changes sign when the mechanically zero pre-2017 lags (208 of 725) are excluded, a restriction that also removes the early years; under trends both are imprecise (trade 1.2853; merchandise 4.2759). Co-movement collapses, not variance: the correlation of residualised index with openness falls from 0.091 to 0.019 while about three-quarters of identifying variance survives (SD 0.105 → 0.091). Post-entry lags reverse the sign (−17.0963) and equal weights remove significance. On the post-entry L2 window, the constructed index is exactly collinear with member intercepts and linear trends, so its coefficient is not separately identified.
No takes yet. Share an insight, caveat, or question.
Baoyong Feng (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: