RDP 2025-06: An AI-powered Tool for Central Bank Business Liaisons: Quantitative Indicators and On-demand Insights from Firms 2. RBA's Liaison Program

In 2001, the RBA established its liaison program – a formal program of economic intelligence gathering. The program involves RBA staff verbally interviewing contacts who typically hold executive roles in organisations. Interviews are conducted on a continuous basis throughout the year; this contrasts with some business surveys that are conducted at static points, such as the end of the month or the quarter.

On average, around 75 contacts are spoken to a month across the nation and around 900 meetings are conducted per year. While most are interviewed for about an hour once per year, 10–15 per cent of contacts are spoken to at shorter intervals. Over a typical year, just over 650 individual contacts are interviewed. The RBA has conducted around 22,000 interviews as part of this program over the past 25 years. Through these interviews, the RBA collects four kinds of economic information.

1. Basic metadata: including the firm's name, the broad industry group the firm's activities fall within, the most relevant four-digit Standard Industrial Classification code, headcount, and geographic location of the Head Office or branch office where relevant.
2. Qualitative views: prose text of firms’ answers to economic questions posed by RBA staff, captured as ‘liaison summaries’. Most questions posed over the life of the program have been consistent over time. That is, most participants have been asked at each liaison interview the following ‘core’ questions: how demand for your goods or services, your investment, headcount, non-labour costs, wages and prices changed over the past year, and how are they expected to change over the coming year? The text captured often includes firms’ explanations of ‘why’ these economic conditions have changed (or not) and what reasons underpin their expectations for economic conditions over the year ahead. The prose captured also includes firms’ answers to occasional topical questions.
3. Quantitative economic outcomes: many firms also offer numerical percentage firm-level outcomes and expectations in responding to some of the above core questions (e.g. wages and prices growth for a firm).
4. Staff scores: Based on the responses provided in the interview and recorded in prose, RBA staff assign a quantitative indicator of the extent of change in the key economic variables being discussed. Specifically, staff assign scores on an ordinal scale from −5 to +5 based on the conditions reported by liaison contacts compared with one year ago and a score for firms’ expectations for the year ahead, where that information is provided. If the level of a variable was unchanged, the score would be zero. An extreme rise would be assigned a score of +5 and an extreme fall assigned a score of −5. While judgement is required for the scores in between, a score of +2 would generally be considered a ‘normal’ or ‘average’ increase from the previous year.

These records are not exact transcripts of the meeting but are detailed records, for example on average around 1,500 words per meeting have been recorded over the past five years. Records are completed on the basis of what was said from the perspective of the firm, but in recording the comments they are organised thematically and terminology is aligned where appropriate to be consistent with Australian official statistical concepts.

The interviews, and as such the information recorded, tends to be from medium- to larger-sized firms that are broadly representative of the industry and geographic structure of the economy. The liaison program has a pool of around 800 active contacts. Three-quarters of these are businesses, though an important part of the program is liaison with industry associations, government agencies and community associations that together provide a fuller picture of economic conditions. The known gaps in the sample, for example direct exposure to small firms (as opposed to the indirect exposure through industry associations) is a limitation of the dataset.^[4]

Footnote

For further discussion of the history and nature of the RBA's liaison program see Dwyer, McLoughlin and Walker (2022). [4]