BACI

Overview

BACI (Base pour l’Analyse du Commerce International) is a comprehensive database of bilateral trade flows developed by CEPII (Centre d’Études Prospectives et d’Informations Internationales), a leading French research center on international trade.

This dataset provides:

Bilateral trade data: Export and import flows between country pairs
Product-level detail: Over 5,000 distinct products classified by the Harmonized System (HS)
Global coverage: More than 200 countries and territories
Historical depth: Data spanning multiple decades of international commerce
Quality assurance: Reconciliation of mirror flows (exports reported by one country matched with imports reported by trading partner)

The BACI dataset is widely used in academic research, policy analysis, and international trade studies due to its comprehensive coverage and quality control procedures.

Data Source and Methodology

BACI is constructed from UN Comtrade data with significant processing: - Reconciliation of discrepancies between reported imports and exports - Imputation of missing values using econometric techniques - Classification using the Harmonized System (HS) nomenclature - All values converted to USD for comparability

For detailed methodological information, visit the CEPII BACI documentation.

Available Dataset

HS92 (Harmonized System 1992)

The HS92 classification provides trade data organized under the 1992 version of the Harmonized System nomenclature.

Coverage: Bilateral trade flows for 5,000+ products (HS 6-digit level)
Countries: 200+ countries and territories
Time period: 1995 onwards (varies by country; most comprehensive from 2000 onwards)
Variables: Trade value (USD), quantity (where available), exporter, importer, product code
Use cases:
- Analyzing trade specialization and comparative advantage
- Studying trade relationships between countries
- Product-level trade analysis
- Research on globalization and supply chains

Important Information About Downloads

Download Size and Time

Important: The BACI dataset is very large. All data is packaged in a single compressed file on the CEPII server. Even if you only need data for specific years, the entire dataset must be downloaded first, then processed locally.

Typical download size: 500+ MB (compressed)
Uncompressed size: Several GB
Download time: Can take 30-60+ minutes depending on internet connection
Processing time: Additional time required after download for extraction and filtering

Recommendation: Plan your download during off-peak hours or use a stable, high-speed connection.

Function Parameters

1. dataset

Currently only one dataset is available:

dataset = "HS92"  # Harmonized System 1992 classification

2. raw_data

Controls whether to download the original data or the processed/cleaned version.

TRUE: Returns raw CEPII data with original formatting and column names
FALSE: Returns treated data with standardized English variable names and cleaned formatting

raw_data = FALSE  # logical

3. time_period

Specifies which year(s) to download. You can request single or multiple years.

Available: Generally 1995 onwards, though coverage varies by country
Most complete: 2000 onwards for the majority of countries
Note: Specifying years helps with filtering but still requires downloading the full dataset

time_period = 2016              # single year
time_period = c(2010, 2015)     # multiple specific years
time_period = 2010:2020         # range (will select available years)

4. language

Output language for variable names and documentation.

"pt": Portuguese
"eng": English

language = "eng"  # character string

Examples

# download treated trade data for 2016 (HS92 classification)
# Warning: large download, may take a long time
trade_2016 <- load_baci(
  dataset = "HS92",
  raw_data = FALSE,
  time_period = 2016,
  language = "eng"
)

Data Notes

Raw vs. Treated Data

Raw data (raw_data = TRUE): CEPII’s original format with variable codes and potentially non-standard naming
Treated data (raw_data = FALSE): Standardized with English variable names, consistent column formatting, and ready for immediate analysis

Data Structure

Each row typically represents a trade flow with: - Exporter: Country code (ISO 3-letter code) - Importer: Country code (ISO 3-letter code) - Year: Calendar year of the trade flow - Product code: HS 6-digit classification - Product name: Description of the product - Value: Trade value in USD - Quantity: Physical quantity (where available)

Country and Product Coverage

Most countries covered from 2000 onwards
Coverage varies by country; some countries have earlier data available
HS 6-digit codes ensure consistency with international standards
Product classifications may change over time as the HS is updated

Quality Notes

BACI reconciles mirror flows to improve data quality
Missing values may occur for certain country-product-year combinations
Some countries report data more frequently/accurately than others
Small trade flows may have been imputed or estimated