Global flows of political information: Measuring the information content of news reporting on foreign countries worldwide

The flows of information across countries constitute a key dimension of globalized international politics. Our paper presents a novel methodological framework for estimating the volume of information presented in the media space of a country about other foreign countries. The framework is rooted in information theory (IT) and utilizes the notion of communication entropy as the measure of information content. We highlight the principal difference between the amount of news and the actual volume of information transmitted when reporting about other countries. The measure shows that a flood of news on a country does not necessarily translate into high information content of reporting. We show the applicability of the framework by analyzing dyadic information ties among more than 80 states. Empirically, we draw on a database created within the project GLOWIN (Global Flows of Political Information), where we analyze more than 4.7 million news articles from 2018-2022, in close to fifty languages, with non-English content translated automatically into English. A supervised machine learning model, based on a fine-tuned version of the large BERT language model, classifies the news articles by topic. Named entity recognition is used in detection of news references to other (foreign) states. The article thus combines three innovative features: 1) it offers an analysis of a large volume of news articles across the world, 2) it uses suitable machine learning-based tools to analyze the content of these articles, and 3) it feeds this data into a theoretically derived quantitative measure to estimate robustly the volume of information that flows between states in the globalized world.