The New South Wales Branch of the Statistical Society of Australia invites you to a presentation titled "An AI-powered Tool for Central Bank Business Liaisons" delivered by Finn Lattimore, Head of the Reserve Bank of Australia’s Data Science Hub.
Date: Wednesday 3 June 2026
Times (AEST):
- 5:30pm - 6:00pm - Light Refreshments
- 6:00pm - 7:00pm - Talks and Discussion
- 7:00pm onwards - Dinner (see further details below)
Venue: USyd, F07.Carslaw 173 Lecture Google Map
Format: In person
Note that attending the dinner requires a separate RSVP - please register prior to the event.
Abstract: An AI-powered Tool for Central Bank Business Liaisons
Central banks increasingly use qualitative information to complement traditional economic statistics, particularly in periods of heightened uncertainty. Business liaison is timely, captures firm sentiment and provides insights into the underlying drivers of firms’ decision making. However, as information is captured as text, it has historically been challenging to analyse systematically and at scale. This talk introduces a text analysis and retrieval system developed at the Reserve Bank of Australia to unlock 25 years and roughly 25,000 liaison meeting notes using modern natural language processing.
The system delivers three core capabilities: rapid full-history search across the liaison corpus; topic exposure and sentiment indicators derived from both transformer-based language models and keyword dictionaries; and machine extraction of precise numerical quantities such as firms self-reported wage and price growth. We present validation results showing these methods perform strongly against human benchmarks.
We then demonstrate the practical value of these indicators in an empirical nowcasting exercise for Australia’s Wage Price Index, comparing the Phillips curve baselines against ridge, lasso and elastic net specifications augmented with the liaison derived features.
Presenter Bio: Finn Lattimore
Finn is the Head of the Reserve Bank of Australia’s Data Science Hub, leading the application of advanced data science techniques to central banking challenges. Her work focuses on translating emerging methods into practical impact across policy, operations, and research. She holds a PhD in Computer Science from the Australian National University, where she specialised in causal inference in machine learning. Her research interests include causal inference, natural language processing, and applications of generative AI to central banking.
Any questions, please contact: secretary.nswbranch@statsoc.org.au
@StatSocAus
