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 working paper by Severin Joho and Enzo Rossi examines 1,063 speeches by senior Swiss National Bank officials from 1928 through 2024. It reports more frequent, shorter and more accessible speeches over time, with inflation receiving greater attention in the flexible-exchange-rate era.
Severin Joho and Enzo Rossi have published a Swiss National Bank working paper analyzing 1,063 speeches by high-ranking SNB officials from 1928 to 2024. The study tracks changes in how often officials spoke, how long and readable their speeches were, and which policy subjects they emphasized, while comparing conventional Python-based text analysis with ChatGPT.
The authors report that the annual number of speeches rose from the mid-1950s, with that increase accelerating after Switzerland moved to flexible exchange rates. They also find a long-term shift toward shorter speeches and language that has become more accessible since the 1960s. The paper presents these as patterns in the historical corpus, not as evidence that any one change caused another.
The study’s thematic analysis finds that inflation became more prominent in SNB officials’ discourse during the flexible-exchange-rate era. By comparison, cyclical economic development and unemployment received less emphasis. The authors do not frame those findings as a measure of policy effectiveness; they describe how the subjects appearing in speeches changed over time.
To examine both the texts and the methods used to analyze them, the researchers benchmarked Python-based natural-language processing against ChatGPT. Their report discusses how probabilistic language models and deterministic text-processing frameworks can complement one another, alongside algorithmic pitfalls and questions that need further research.
How SNB Communication Changed
The findings offer a long-run view of how a central bank’s public communication has evolved across different monetary regimes. More frequent statements can give researchers a larger record of official messaging, while shorter and more accessible language marks changes in the form of communication. The reported shift in themes also shows that the subjects emphasized in public remarks were not constant across the period studied.
The work also has a methodological implication for researchers studying institutional language. A century-scale corpus is large enough to make consistent text analysis useful, but automated tools can misread meaning or context. By comparing ChatGPT with conventional NLP, the authors address how different approaches may be used together rather than treating either as a complete substitute for the other.
Python natural language processing books for 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 by high-ranking Swiss National Bank officials over 97 years, from 1928 through 2024. It is the foundation for the paper’s analysis of speech frequency, length, readability and subject matter. The authors describe it as a novel, comprehensive dataset; the supplied report summary does not provide a breakdown of speeches by decade or speaker.
The paper links an acceleration in speech frequency to the period after the transition to flexible exchange rates. Its comparisons therefore span distinct monetary-policy settings and historical periods. The study examines rhetoric in speeches; it is not presented as a direct evaluation of the SNB’s decisions or their economic outcomes.
“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, authors of the SNB working paper
Limits of the Text Analysis
The supplied summary does not report the exact yearly speech counts, average speech lengths, readability scores or theme frequencies, so the size of the measured changes cannot be stated here. It also does not specify how the authors validated ChatGPT’s outputs against human coding, which model version they used, or how results differed between the two methods. The authors themselves point to algorithmic pitfalls and open methodological questions.
The reported associations across historical periods do not establish why communication changed. The summary links the faster rise in speech frequency to the shift to flexible exchange rates but provides no causal test. It also does not show whether the thematic findings hold equally across individual speakers, speech types or other SNB communications.
Questions for Future NLP Research
The working paper identifies open questions for future research on using large language models to study central bank communication. The next research steps indicated by the authors are methodological: examining algorithmic limitations and clarifying how probabilistic models can work alongside reproducible, rule-based Python frameworks.
The supplied material does not announce a follow-up study, a policy response from the SNB or a schedule for further work. Readers seeking more detail on the dataset, methods and results can consult the full working paper published on the SNB website.
Key Questions
What does the working paper study?
It examines 1,063 speeches by high-ranking Swiss National Bank officials delivered from 1928 to 2024, focusing on frequency, length, readability and themes.
What changes did the authors identify?
The authors report more frequent speeches from the mid-1950s, an acceleration after the move to flexible exchange rates, shorter speeches, more accessible language since the 1960s, and greater emphasis on inflation in the flexible-rate era.
How did the researchers use ChatGPT?
They benchmarked ChatGPT against Python-based NLP text analysis to study the speeches and discuss methodological strengths, limitations and possible complementary uses.
Does the paper explain why SNB speeches changed?
The reported findings describe historical patterns, but the supplied summary does not establish their causes. It notes an association between the acceleration in speech frequency and the shift to flexible exchange rates, not proof that the change in exchange-rate regime caused it.
What remains unknown about the findings?
The summary does not give detailed yearly counts, numerical readability results or a full account of how ChatGPT’s analysis was validated. The authors also identify unresolved questions about algorithmic pitfalls in language-model research.
Source: primary
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
