TL;DR
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
A Swiss National Bank working paper by Severin Joho and Enzo Rossi analyzes 1,063 speeches by senior SNB officials from 1928 to 2024. It reports rising speech frequency, shorter and more accessible language, and a stronger focus on inflation in the flexible-exchange-rate era; the researchers also compare Python-based text analysis with ChatGPT.
Severin Joho and Enzo Rossi have published a Swiss National Bank working paper analyzing 1,063 speeches by senior officials from 1928 to 2024, documenting changes in how often the central bank speaks, how its officials write, and which economic topics they emphasize. The study reports that speeches became more frequent, shorter and more accessible over time, while inflation gained prominence in discussion during the flexible-exchange-rate era.
The authors say they assembled and processed a dataset of 1,063 speeches delivered by high-ranking SNB officials across the period. Their analysis examined speech frequency, length, readability and thematic content. The paper is dated October 23, 2026, and is listed in the SNB’s working-paper series as working paper 2026/13.
Among the findings reported in the study, the number of speeches per year followed an upward trend from the mid-1950s, with a more marked acceleration after Switzerland moved to flexible exchange rates. The authors also identify a shift toward shorter speeches and say their language became more accessible from the 1960s onward.
The paper finds a change in subject matter as well: inflation moved to the forefront of policy discussion in the flexible-exchange-rate era, while cyclical economic development and unemployment received less emphasis. The authors also compare conventional Python-based text-mining methods with ChatGPT, presenting the approaches as complementary and discussing methodological risks in using large language models for central-bank communication research.
How SNB Communication Changed
The findings offer a long-run view of how a central bank communicates as its operating environment and policy concerns change. Speech frequency, length and readability affect how much information officials put into the public domain and how accessible that communication may be to readers. The study describes patterns in communication; it does not, on the information provided, establish that those changes caused shifts in policy or public understanding.
The reported change in themes may also help researchers trace how the SNB’s public discussion evolved alongside its monetary-policy setting. The increased prominence of inflation after the move to flexible exchange rates gives analysts a way to examine changes in the institution’s stated priorities across decades. The paper’s methods discussion matters beyond this dataset because it considers how AI-assisted text analysis can be checked against more conventional computational methods.
Python natural language processing books text analysis
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
A Century of SNB Speeches
The dataset covers speeches delivered between 1928 and 2024, giving the authors material to compare communication across distinct periods rather than relying on a short-term sample. The paper highlights the transition to flexible exchange rates as a point after which the annual pace of speeches accelerated and inflation became more prominent in thematic analysis.
The researchers used two types of tools: deterministic Python-based text-mining frameworks and ChatGPT, a generative AI system. They frame the comparison as a methodological benchmark, not as a claim that one approach can replace the other. Their stated focus includes the complementary uses of the methods, potential algorithmic pitfalls and open questions for future research.
“The frequency of annual speeches has followed an upward trajectory since the mid-1950s, accelerating notably after the transition to flexible exchange rates.”
— Severin Joho and Enzo Rossi, in the working paper abstract
Limits of the Reported Findings
The source summary does not provide the year-by-year speech counts, measured changes in speech length or readability scores, or the details of how topics were classified. It also does not specify the exact size of the increase in speech frequency or quantify how much emphasis on inflation changed. Those results should not be inferred from the overall descriptions alone.
The supplied material does not give a detailed account of how the Python and ChatGPT outputs compared, which model or prompts were used, or how the authors tested accuracy and bias. It is also unclear from the summary how the dataset treats differences in speech availability across decades. The paper identifies methodological pitfalls and open questions, but the full scope of those limitations is not detailed in the source material.
Further Tests of Language Models
The paper points to open questions for future research on the use of large language models in central-bank communication analysis. Further work could examine how reliably AI systems identify themes and readability across historical texts, and how their results compare with reproducible Python-based procedures. The source does not announce a specific follow-up study or timetable.
Readers seeking the underlying measures and method details can consult the full SNB working paper, which is the primary source for the dataset and analysis. Until those details are assessed, the reported trends are best understood as the authors’ findings for the speeches they compiled, rather than as a complete account of every form of SNB communication.
Key Questions
What did the SNB working paper examine?
It analyzed 1,063 speeches by high-ranking Swiss National Bank officials delivered from 1928 to 2024, including their frequency, length, readability and themes.
What changes did the authors report?
The authors report that annual speech frequency rose from the mid-1950s, speeches became shorter, and language became more accessible from the 1960s. They also found that inflation became more prominent in discussion during the flexible-exchange-rate era.
Did the study say inflation was the only topic discussed?
No. The paper says inflation moved to the forefront of thematic discussion while cyclical economic development and unemployment received less emphasis. It does not say those other topics disappeared.
How did the researchers use ChatGPT?
They benchmarked generative AI using ChatGPT against traditional Python-based text mining to study central-bank communication. The source summary describes the methods as complementary and says the paper addresses algorithmic pitfalls, but does not provide detailed comparison results.
Does the supplied summary quantify the changes?
No. It describes the direction of the reported trends but does not provide exact annual counts, readability scores or measurements of the change in topic emphasis.
Source: primary
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
