BACI Dataset Documentation BACI provides data on bilateral trade flows for 200 countries at the product level (5000 products). Products correspond to the "Harmonized System" nomenclature (6 digit code). BACI relies on data from the United Nations Statistical Division (Comtrade dataset). Since countries report both their imports and their exports to the United Nations, the raw data we use may have duplicates flows: trade from country i to country j may be reported by i as an export to j and by j as an import from i. The reported values should match, but in practice are virtually never identical, for two reasons: Import values are reported CIF (cost, insurance and freight) while exports are reported FOB (free on board). Mistakes are made, because of uncertainty on the final destination of exports, discrepancies in the classification of a given product, etc... Licensed EtaLab Open Licence v2.0, original data downloaded from http://www.cepii.fr/CEPII/en/bdd_modele/bdd_modele_item.asp?id=37"

Industry-level measures of export quality that take global input-output linkages into account and associated code; as described in Wacker et al. (2025). The main repository folders are: - codes/: a folder with python codes performing the four steps explained in the Usage Notes below. - data/QualEst/: a folder with zipped CSV files (QualEst_YYYY.zip, where YYYY stands for the year), storing the quality estimate for bilateral traded products () based on Trenczek and Wacker [15]. This folder will also store the quality aggregations generated as intermediate products in the codes. Those estimates are ultimately based on the BACI database for bilateral trade flows [13], version 202301-HS07. Quality estimates and trade flows are available for bilateral trade between countries and administrative regions for the entire world, and data for each year are provided in separated CSV files with the following columns: identifier of importing and exporting countries (i and j), product code in 6-digit HS2007 classification (hs6digit), value of export in thousand US dollars (v), and quality estimates (qual_idx). Country identifiers are integers that follow the classification in the BACI database; more details can be obtained from https://cepii.fr/CEPII/en/bdd_modele/bdd_modele_item.asp?id=37. - data/ICIO/: An empty folder, to which one should download and unzip the Regular ICIO input-output tables from the official OECD website (https://oe.cd/icio) and place them in this folder. We used the 2023 December release of Regular ICIO input-output tables from OECD. ICIO are available for 76 countries and regions, plus a "rest-of-world" entry covering all other economies, and 46 industries based on 2-digit ISIC Rev. 4 classification. The input-output tables are provided in CSV format. More details on the data structure and the definitions for rows and columns can be found in https://oe.cd/icio - data/keys/: folder with correspondence tables for country and product-industry identifiers used in the export quality dataset and ICIO input-output tables; both are in CSV format. The file for product-industry identifiers includes a column (TYPE) that describe the property of traded products: 0: intermediates, 1: mixed-use products, 2: final products for consumption or capital investment only. Further details can be found in the ReadMe.txt file placed in the root folder of the replication package.

Dataset contains bilateral trade flows (from BACI), infrastructure scores (from WEF) and bilateral tariffs (from TRAINS). Date - 2006 to 2017, in panel format. This is the data for our paper - "Infrastructure Quality and Trade Liberalization". We assembled this dataset from the above sources to allow us to test for the impact of infrastructure quality (i.e., using the infrastructure scores from WEF) on trade. We also use the tariff data (from TRAINS) as modelling variables, with a specific focus on bilateral trade liberalization at 50%, 60% and 70% reductions in bilateral tariffs. Note that this is version 2 of the data, with more updated multilateral resistance terms.

Abstract It is difficult to establish the impact infrastructure has on trade balances due to confounding factors. Our empirical approach leverages on episodes of trade liberalization, interacting such episodes with infrastructure. Providing a simple extension to the gravity model, we find that infrastructure quality differences between trade partners result in trade imbalances, which liberalization then accentuates. For robustness checks, we interact the liberalization indicator against other confounding variables and do not find such consistent effects. We conclude that lack of quality infrastructure in developing economies contributes to larger trade deficits, and infrastructure development should be supported alongside trade liberalization. Keywords Infrastructure; Gravity model; Trade; Trade deficits JEL classification O18; F14; F40

This dataverse contains a Read_me.txt, STATA data and do-files. Primary sources used are the EASD (see chapters 3&4), GGDC 10-sector database, Maddison Project database, World Trade flows database and BACI CEPII database (Comtrade).