Latin American exports during the first globalization: How statistical aggregation and standardization affect our understanding of trade

• Published in: Historical methods Vol. 56; no. 2; pp. 97 - 114
• Main Authors: Badia-Miró, Marc, Carreras-Marín, Anna, Rayes, Agustina
• Format: Journal Article
• Language: English
• Published: Washington Routledge 03.04.2023; Taylor & Francis Inc
• Subjects: and Other Services: Latin America Caribbean; Diversification; Energy; Export-led-growth model; Exports; F14: Empirical Studies of Trade; Globalization; International trade; Latin America; Latin American history; margins of trade; N01:Economic History: Development of the Discipline: Historiographical; N76: Transport; Sources and Methods; statistical standardization; Technology; Trade
• ISSN: 0161-5440; 1940-1906
• DOI: 10.1080/01615440.2022.2160399

Data constraints determine the scope of historical research. The gradual digitalization of large sources has increased the number of approaches that can be applied to comprehend the past. Here, we show an example of how trade data can shed new light to better understand growth patterns of Latin America at the end of nineteenth century. Latin American exports during the First Globalization has mainly focused on the high concentration of few products to few markets. In this article, we propose a complementary way to measure diversification by considering the relative number of goods and the number of trade partners. To do so, we had to deal with historical official trade data hardly comparable, which has been homogenized for some countries in a 1910 benchmark (SITC-rev2). From that, we can offer a new measure of trade diversification, internationally comparable over time and across countries. Standardizing trade data also implies some consequences in the sense that the number of items for industrial goods is always greater than those for primary goods, arising the question if lower diversification is an inevitable result of specialization on commodities, or instead, it is a statistical artifact driven by the standard criterium we impose on data.