Skip to main navigation Skip to main content Skip to page footer

Working Paper

PLAID Product-Level AI-Derived Indicators Database for International Trade

Authors

  • Brockhaus
  • C.
  • Hinz
  • J.
  • Iodice
  • I.

Publication Date

JEL Classification

F14 F17 C80 C38

Key Words

Product-level Indicators

Large Language Models

International Trade

Gravity Models

Open Data

Related Topics

Globalization

We introduce PLAID (Product-Level AI-Derived Indicators Database), an open, versioned data-base of product-level indicators for international trade, generated by an ensemble of frontier large language models (LLMs). The current beta covers six indicators at the HS 6-digit level across all major HS revisions since 1992: (i) the Rauch classification of goods into organized-exchange, reference-priced, and differentiated products; (ii) UN Broad Economic Categories (capital, inter-mediate, consumption); (iii) economic perishability on a five-class scale; (iv) a hazardous-materials flag; (v) microchip/semiconductor content; and (vi) 3TG conflict-mineral content. We validate each indicator against established external benchmarks and demonstrate its empirical relevance through a targeted gravity application. PLAID is designed to grow: future releases will add indicators for technology intensity, R&D intensity, regulatory sensitivity, and other product attributes. All data and code are publicly available.

Kiel Institute Experts

  • Carsten Brockhaus
    Kiel Institute Researcher
  • Prof. Dr. Julian Hinz
    Research Director

More Publications

Topics

  • View over cargo ship deck with containers

    International Trade

Research Center

  • Research Center

    Trade