Information is the resolution of uncertainty – Claude Shannon
Case Study: Extracting Attention Trends from an Investment Newsletter
Natural Language Processing methods can bring considerable insights for economists and investment strategists. Here is one example. I receive dozens of newsletters and content updates from various macro / financial sources and content providers in my mailbox. One of our favourite readings I must say is BloombergSenior Editor and Columnist John Authers we read his newsletter “Points of Return” (https://lnkd.in/ez4zMAyT) whenever we have some time.
So, we wanted to know how John’s focus on different topics evolved over the last three years, since June 2020. We used a statistical algorithm called Latent Dirichlet Allocation pioneered by David Blei. For more clarity, we averaged the resulting topic distribution on a weekly basis. You can see below the result. It is striking to see how monetary policy and the Fed’s fight against inflation came to largely dominate the newsletter. In contrast, China and the Pandemic receded almost in tandem. Politics come back from time to time as can be seen from the pink color patches. Overall, John is still very much focused on inflation … as is the Fed.
Following the publication of a post on Linkedin outlining these results, I have been very pleased and honored by the feedback received from John Authers. This analysis even made it to an Issue of Points of Return published on August 14, 2023. Here is the part in which John talks about our analysis:
Mind-Reading
It’s a little strange to open LinkedIn and find a post entitled “What’s in John Authers’ Head?” It’s even stranger to see a beautiful dynamic graph produced with Natural Language Processing to answer the question. But that’s exactly whatAlexandre Kateb of the Multipolarity Report produced in this post. He went to the trouble of using sophisticated programs to trawl through the last three years of Points of Return, and analyze how the priorities and topics of the newsletter have changed over that time. Here’s the result:
The orange bloc that grows ever bigger at the bottom is for “inflation and monetary policy.” This is no surprise, and it’s an emphasis I’m happy to defend. It matters a lot, and the ultimate outcome of 2020’s extraordinary monetary interventions is still unknown. The big green splurge at the top that steadily recedes is for the Covid pandemic — also as I would have expected, and completely justifiable. You can also see brief blotches of interest in the Ukraine invasion and the UK gilts crisis.
What surprises me is the fall in the amount of time the newsletter has devoted to China and the emerging markets over those three years. I find this hard to defend. China can never be safely ignored, and if people aren’t paying it much attention that tends to be a bad sign. The current situation there, with deflation appearing to be a real risk, combined with the crucial shift in its commercial relationship with the US, might well in future be viewed as the most important developments of our time. So, expect more Points of Return on China and the emerging complex around it; I’ve been remiss in neglecting them.
NLP has made itself useful to pointing to some topics I’ve not focused on as much as I should have. However, I’m glad to say that there’s still a lot going on in my head to which it has no access. I hope it stays that way.