Tracking Economic Uncertainty Through News Homepage Links

Economic Uncertainty is typically measured through analyzing the fulltext of news articles or television captioning and can offer fascinating insights into the forces driving global economic change. How might it manifest itself in the links on the homepages of the world's news outlets?

The GDELT Global Frontpage Graph (GFG) monitors around 50,000 news homepages worldwide every hour on the hour since March 2018, cataloging all of their links. Total the GFG totals around 130 billion links and grows at the rate of around a quarter billion links per day.

How many of these frontpage links per day contain the words "economic/economy" and "uncertain/uncertainty" in the text of the link and how has that changed over the past year and a half? The fixed space and maximally curated nature of news front pages offers a particularly insightful look into which stories news outlets around the world view as the most important day by day.

The timeline below shows the total number of links per day whose link text matched the search "(economic OR economy) AND (uncertain OR uncertainty)".

Download Chart in CSV Format.

It is important to note that the highly editorialized nature of homepage links and their extreme space constraints means homepage editors frequently use creative rather than clinically descriptive text and thus present a worst case scenario for matching two separate keywords – a more expansive keyword list that included myriad related terms or which examined the text of the articles linked to from the homepage would likely return more homepage links. As-is, the graph above reflects editorial decision-making to explicitly choose EU-related terminology in homepage links over time.

The daily link volume in the GFG changes little over this time and normalizing the timeline above as a percentage of GFG links does not change the results.



For those interested in how this timeline was created, the following Standard SQL query was used processing 4.3TB in 22 seconds.

SELECT SUBSTR(CAST(DATE AS STRING), 0, 8) DAY, COUNT(1) COUNT FROM `gdelt-bq.gdeltv2.gfg_partitioned` WHERE (LOWER(LinkText) like '%economic%' OR LOWER(LinkText) like '%economy%') AND (LOWER(LinkText) like '%uncertain%' OR LOWER(LinkText) like '%uncertainty%') group by DAY order by DAY desc