When inflation came back (2021-2023), attitudes reversed. The doves adopted the data, theories, and principles they had rejected in previous years when they were used by the hawks... And the hawks did the same, from inflation data to monetary policy principles.

See e.g. on inflation expectations

While having opposed preferences, both camps display similar rhetorical strategies: depending on the "inflation season", they mobilize the data, theories and policy principles that help them make their preferences prevail...

... Even when it means being time-inconsistent!

Great plenary session at #COMPTEXT2025 about new cool datasets!

Here @luissattelmayer.bsky.social presents PartySOME, about social media activity on four platforms by hundreds of parties in dozens of countries, carefully curated with @malojan.bsky.social and soon to be public

I've had this screenshot from Banque de France archives ready for a while...

Perfect job for our cbspeeches.shinyapps.io/explorer/!

They wouldn’t let me finish my 8-year term, canned me before I got to double digits 😭

Last, the fourth panel lets you explore the CBSpeeches dataset.

See how women's voices in central banking (slowly) emerged over time, or analyze the distribution of languages and data sources of specific institutions.

The third can help you browse by category.

For example, you can search for the central banks that are the most vocal on a topic, but also go beyond, coloring by speaker to find the most prominent figures within any institution.

The second aggregates the results so you can explore trends in time (by # of mentions or # of speeches)

You can aggregate by region, but also by position, gender, central bank, speaker, etc. View results individually, or stacked to spot cumulative patterns.

The first panel allows you to explore the most relevant speeches for any topic. Type your keywords of interest, select your data, and see results update in real-time.
The plot is interactive: click on a speech to open it in a new tab, or download table and chart in a click!

9/15 Despite this trend, there is still heterogeneity: emerging economies (e.g. HKG, CHN, THA) tend to adopt more "promotional" green finance narratives, while advanced economies focus more on risk management (e.g. ECB, USA, GBR). But with exceptions (e.g. NOR)!

8/15 Second, "Climate-related Risks": centered around prudential concerns about financial stability and monetary policy. This narrative, popularized by Mark Carney's 2015 speech, has become dominant in the last ten years.

7/15 But not all central bankers speak the same: with topic modeling, we differentiate two main climate narratives. First, "Green Finance": focused on investment opportunities, market development, and supporting the transition through financial innovation.

6/15 Back to our paper, we show that climate-related communication has boomed: from ~30 speeches/year in 2007-2016 to ~550/year in 2021-2023! Early leaders were Southeast Asian central banks, before Western European ones took over after 2015.

4/15 This results in a new dataset of 35,487 speeches from 131 central banks over 1986-2023. This includes 15,435 additional speeches from CB websites and 2,007 from CB archives, representing an 89% increase compared to BIS, with new Gender and Position variables.

3/15 The BIS dataset of CB speeches has been incredibly useful, but has limitations: missing central banks, gaps in coverage, and language constraints. To make sure we did not miss anything, we decided to expand it with systematic web-scraping and targeted archives.

Absolute banger of an article on Bank of England's scientization, combining a clear theoretical contribution with a super-ambitious empirical strategy: new Original data + State-of-the-art natural language processing = I can't recommend it highly enough.

hal.science/hal-04267004/d