What is it about?
Traditional measures of country similarity (distance between capital cities, common language, common border etc.) are unambiguous but arguably incomplete. There is potential new information in comparing the overlap between textual descriptions of country pairs. This can be done accurately and automatically using the "semantic fingerprinting" technology developed by Cortical.io. This paper quantifies the similarity of countries based on their names and short economic descriptions and shows it contributes to explaining investor behaviour.
Featured Image
Photo by Cathy Mü on Unsplash
Why is it important?
An easy to use methodology that uses artificial intelligence to extract the "semantic fingerprint" of any text or term makes it possible to quantify the similarity of country descriptions and country names. The country similarity data used in this paper is made available to researchers who wish to test it in other empirical exercises.
Perspectives
Any terms/ texts of any length can be efficiently quantified using an intuitive technology inspired by state-of-the-art neuroscience research, without requiring any investment in software or coding skills by the empirical researcher. The technology has already proved relevant in empirical research in Finance and may add value to research in other fields as well.
Dr. Crina Pungulescu
Read the Original
This page is a summary of: Bilateral home bias: A new measure of proximity., Journal of Neuroscience Psychology and Economics, September 2022, American Psychological Association (APA),
DOI: 10.1037/npe0000162.
You can read the full text:
Resources
Cosine Similarity
Cosine similarity based on semantic fingerprints country names, economic descriptions and keywords extracted from economic descriptions.
Extended Cosine Similarity
Extended cosine similarity based on semantic fingerprints of country names, economic descriptions and keywords extracted from economic descriptions
Contributors
The following have contributed to this page
