Trade Reallocation Without Aggregate Diversion
Overview
This completed group study examines sanctions as trade-cost shocks and distinguishes absolute replacement of sanctioned trade from relative gains inside destination markets. It combines structural-gravity estimation with a descriptive analysis of changing intermediary positions in world trade.
Research Question
Do sanctions create enough third-country substitution to replace disrupted bilateral trade, or do they mainly redistribute market shares inside a contracting and more fragmented network?
Economic or Technical Motivation
An exporter can gain share when a competitor is sanctioned without increasing aggregate trade enough to compensate for the disrupted flow. Separating these two margins prevents relative repositioning from being mislabeled as full trade diversion.
Data
The final analytical panel contains 415,631 exporter-importer-year observations from 1991 to 2023. Bilateral trade comes from WITS and sanction episodes from the Global Sanctions Data Base. The construction distinguishes direct sanction relationships, sanction type, diversion exposure in levels, and a normalized diversion-share measure.
Methodology
The main specifications use PPML with importer-year, exporter-year, and bilateral-pair fixed effects. Additional models separate trade, financial, and military sanctions; replace diversion measures; use lagged sanctions; and run placebo specifications. Betweenness centrality is used descriptively to trace intermediary positions in the global trade network.
Main Findings
The baseline sanction coefficient implies approximately 6% lower bilateral trade. Diversion exposure in levels is weak or statistically insignificant, while the normalized diversion-share measure is positive and significant: exporters can improve their relative position without replacing lost trade in absolute terms. Trade sanctions have the strongest direct association with lower trade; financial and military sanctions are imprecise after fixed effects.
Robustness, Validation, or Model Assessment
The negative sanction result persists across alternative sanction definitions and lagged specifications. Relative market-share reallocation remains positive across the principal specifications. Placebo diversion variables are reported as insignificant. The network results are explicitly descriptive rather than causal.
Tools and Technologies
Stata, PPMLHDFE, structural gravity, high-dimensional fixed effects, clustered inference, and network-centrality analysis are directly demonstrated.
Limitations
The country-pair panel can conceal product- and firm-level adjustment. The diversion variable measures exposure to substitution opportunities rather than direct firm decisions. The network section documents reconfiguration but does not identify a causal effect of sanctions on centrality.
Deliverables
- Complete 44-page research report
- Cleaned country-pair-year panel
- PPML specifications, heterogeneity estimates, and robustness tables
- Trade-network figures and policy discussion