In last week’s inflation release, headline inflation came in at 2.4% – the second consecutive print within the RBA’s target range of 2–3%. Meanwhile, the RBA has said it will cut rates when inflation is on track to be sustainably within the target range. Why do they judge this to have not happened already?
A key reason is that central banks look more at ‘underlying inflation’ than headline inflation. Both headline and underlying inflation are weighted averages of various granular price series for different categories of goods and services (like bread, rents, childcare, etc). While headline inflation weights these series to best represent what consumers spend on, underlying inflation uses different weights that aim to measure the underlying inflationary pressure in the economy.
The RBA’s preferred measure of underlying inflation is the trimmed mean, but it also looks at measures called weighted median and ‘excluding volatiles’. These are each different methods for removing the biggest price movements and summarising the rest. They’re all based on the assumption that the prices that move the most are the least representative of the economic forces that generate inflationary pressure.
In last week’s inflation release, trimmed mean came in at 3.2% – a little below the latest RBA forecast of 3.4%, and still above the RBA’s target band. Weighted median and excluding volatiles came in at 3.4% and 2.7%, respectively.
However, there exists another measure based on different assumptions that is used by other central banks, including in the United States, Europe, Canada and New Zealand. This measure uses a statistical method called factor modelling. When applied to inflation, factor models are designed to upweight price series that in the past have moved together, and downweight price series that move idiosyncratically – that is, series whose movements are not shared with other categories of goods and services. The idea is that if some categories move in their own unusual ways, they’re less likely to be driven by broad economic pressures.
Factor models are therefore well suited to measuring the underlying inflationary pressure in the economy, which is why other central banks use them. Another big advantage is that these models do not assume that price categories with large movements are necessarily less representative of inflationary pressure, so they provide an alternative way of understanding price movements that complements the RBA’s existing suite of measures.
Our factor model measure of inflation comes in at 2.7% for the year to December 2024 – well below the trimmed mean measure. It also displays a lower peak in early 2023 compared to trimmed mean, and a more consistent decline since then.
One potential downside of factor model measures – depending on your perspective – comes from their use of past data to weight the price categories. As new price data are released each quarter, the new information is used to reweight the categories, which means the past values of the factor model measure also get revised – not unlike the revisions that happen to past GDP data. This is important to acknowledge when looking back at policy decisions, but is no reason to pay less attention to the information it provides.
For example, the fact that inflationary pressures did not decline much during the pandemic (see figure below) would have been less evident from the measure at the time (solid line) than it is now (dotted line). Something we leave for future discussion is that the factor-model weights provide additional rich information on price co-movement patterns, which at that time would have pointed to a high likelihood of future revisions. The trimmed mean and weighted median measures declined by noticeably less, which demonstrates the diversity value of using a variety of measures.
The RBA Board has several measures of inflation to consider in its crucial decision on 18 February. Its importance warrants a diverse set of evidence. This new measure of inflation – using factor modelling – complements the RBA’s suite of underlying inflation measures and more strongly supports the interpretation that inflation is falling relatively quickly. Perhaps most importantly, it measures underlying inflation as already sustainably within the RBA’s target band.
Nick Garvin
Dr Nick Garvin specialises in microeconometric research for Australian economic and financial policy. His experience includes research at e61, RBA and APRA on topics related to housing, mobility, superannuation, financial stability, financial markets, and payments, and as an econometrics lecturer at UNSW. Nick holds a PhD in Economics from Pompeu Fabra University, a Masters in Economics from London School of Economics, and Bachelor Degrees in Economics and Psychology from Macquarie University.
Luke Hartigan
Dr Luke Hartigan is a Lecturer in economics at The University of Sydney. Luke Hartigan received his Ph.D. in Economics from The University of New South Wales in 2017. His research interests relate to macro-econometrics, focusing on methods useful for the empirical analysis of business cycles and summarising and nowcasting economic and financial conditions. Prior to joining the School of Economics, Luke was an Economist at the Reserve Bank of Australia.
