Compare Measures
Updated estimates, 1995–2022 (80 countries).
Replication Data
This dataset accompanies Trade Interdependence in the Modern Global Economy by Michael-David Mangini and Stephen Chaudoin. Forthcoming at APSR.
The dataset is a directed-dyad-year panel covering 76 countries from 1995 to 2020 (148,200 observations). It contains the following variables:
| Variable | Description |
|---|---|
COU | Country of interest (ISO 3-letter code) |
PAR | Partner country (ISO 3-letter code) |
TIME | Year (1995–2020) |
upper_bound_bayes | Upper bound dependence measure |
compensation_bayes | Compensation dependence measure |
RussettOnealCOU | Russett & Oneal dependence measure |
BarbieriCOU | Barbieri dependence measure |
Dataverse Link Coming Soon
Explore the Data
Beta version
Interactively explore dependence measures across countries and trading partners.
These interactive figures use updated estimates through 2022 (80 countries). The dataverse files for the paper cover 1995–2020.
Monadic Dependence Map
This map shows how dependent each country is on its trading partners, aggregated across all partners. Choose a dependence measure and an aggregation method (average, maximum, or total across partners), then press Play to see how dependence evolves from 1995 to 2022.
About the four dependence measures
Our measures (from the paper) capture how much real harm a partner could do, not just how much trade there is:
- Upper bound — the maximum damage a partner can inflict by withdrawing trade in all products at once, holding all other trade flows constant. It is the worst-case share of trade-based welfare a single partner could destroy.
- Compensation — the percentage increase in trade from a country's other partners that would be needed to compensate it for losing a given partner, holding the product mix constant. Higher = harder to replace that partner.
Aggregate-statistic measures (for comparison) are simple trade ratios. For country i's dependence on partner j, with dyadic trade tradeᵢⱼ = (imports of i from j) + (exports of i to j):
- Russett–Oneal = tradeᵢⱼGDPi — dyadic trade as a share of i's GDP. (Summed over all partners = total trade ÷ GDP — the “Total” option.)
- Barbieri = tradeᵢⱼ(total importsi + total exportsi) — dyadic trade as a share of i's total trade. (These shares sum to ~1 across partners, so only the maximum is meaningful on this map.)
Dyadic Dependence Map
This map shows directed trade dependence between country pairs. Each arrow points from a country to the partner it depends on — e.g. an arrow from the USA to China reads “how dependent is the USA on China?”. Arrows are colored and sized by the measure value (darker/thicker = higher dependence). Use the slider to show the top N strongest dependencies per country, or switch Show to “Dependence on chosen countries” to see how everyone depends on one or two chosen countries. Use the year controls to watch the network evolve.