This article tests whether public mirror trade data distinguish listed controlled goods from same-chapter off-list goods, and whether a multi-row Comtrade response can be summed as a cell total. It uses 244 archived UN Comtrade preview responses for 2021?2023 and compares four Common High Priority Items List codes with four controls in the only complete sample: Germany?Kazakhstan in 2023. In 133 of 160 non-empty cells (83.1%), summing the rows gives approximately eight times the cell maximum. The total-bearing row is identified by customsCode=C00, motCode=0 and partner2Code=0 in all 160 cells, although no response field marks it. The measured factor comprises a partner-2 multiplier of about 2 and a combined mode-by-customs multiplier of about 4. Across eight mirror relations, the median absolute log value gap is 0.620 (+85.9%). Listed and off-list medians are 0.879 and 0.330, but the exact Mann?Whitney test (p = 0.343) and permutation test (p = 0.314) detect no difference. With four relations per group, the comparison has little power and does not establish equivalence or absence of rerouting. Six exporter-side weights are publisher-imputed, so weight and unit-value contrasts are not decisive. These records do not identify physical movement; larger samples and independent movement evidence are needed.
Public price and trade series show a sharp rice-price increase and declining Indian export quantity around the July 2023 non-basmati white-rice restriction. Between June and September 2023, the Thai 5% rice benchmark rose 20.62%, while wheat fell 8.92% and maize fell 16.14%; Brent and urea rose 25.52% and 32.17%. A regression fitted to 102 pre-restriction months leaves an August 2023 residual of +0.17819 log points (z = +3.90). Its low explanatory power (R² = 0.0631) and jointly insignificant covariates mean that the residual describes an unusual rice-price movement without isolating the restriction's causal effect. India's annual HS1006 exports fell 19.7% in quantity and rose 20.8% in unit value between 2022 and 2023, with export value down 2.9%. Annual data combine seven pre-restriction and five post-restriction months; the available HS6 category also combines restricted and unrestricted rice. These observations cannot identify the five-month policy effect or a change in global physical supply. An aggregation audit further finds two- to fourfold overstatement in five of six responses when total and component rows are summed together. Selecting total-bearing rows establishes the annual changes while preserving the distinction between observed adjustment and causal attribution.
Friend-shoring models quantify costs under assumed trade-cost shocks, but do not measure the unit values paid for observed supplier switches. This article asks whether US importers paid different landed unit values as sourcing shifted from China toward other suppliers. It uses 1,305 UN Comtrade partner-detail records for 28 selected HS6 goods, seven years (2017?2019 and 2021?2024), and seven reporting partners. Among the chapters shown, China's share fell 16.2 points in machinery and 16.0 points in electrical machinery, while two chapters rose. In the restricted 26-code panel observed at both endpoints, China's share fell 16.6 points and Viet Nam's rose 11.7, about seven-tenths of China's decline. These comparisons span HS2017 and HS2022 classifications, so reclassification effects remain possible. The data do not identify a supplier price premium: 1,114 of 1,305 partner rows (85.4%) carry publisher-estimated weights, and imputed rows often yield nearly identical within-cell unit values (mean coefficient of variation 0.0380, compared with 0.8693 for fully reported-weight cells). This pattern suggests that imputation may reproduce a common per-commodity unit value; the provider's method is undocumented. Only seven goods contain reported weights, and four permit comparisons across at least three suppliers. The results describe this selected panel and landed unit values, not same-quality prices or the cost of friend-shoring.