There will always be items in the CPI basket with price inflation a long way from the RBA’s 2–3% target. That is no reason to dismiss measures of underlying inflation, such as the trimmed mean, as an indicator of current inflation pressures.
The RBA’s mandate involves an inflation target of 2–3%, always aiming for the 2½% midpoint, while recognising that there will be times when flexibility is needed. The overall CPI inflation result can be pushed around by one-off movements in particular components that say nothing about future trends. That is why the RBA also focuses on various measures of ‘underlying’ or trend inflation. They particularly focus on the trimmed mean measure, which is calculated by excluding the largest increases and decreases that quarter, then taking the average over what remains. Trimmed mean inflation is not the target, but it is the best available single indicator of current momentum in inflation. It is also the measure the RBA uses in most of its forecasting models.
Within both total and trimmed mean inflation, particular subcategories can have inflation rates that are very different from the 2–3% target. Some of these are the results of transitory supply shocks, like the 17% increase in lamb and goat meat prices over 2024, or policy changes, like the 12% increase in tobacco prices over the same period. These shifts typically get excluded from the trimmed mean calculation. As such, there is no real need to remove selected items and then perform the trimmed mean calculation.
(Be sceptical, too, when people advocate removing one policy-affected item, such as electricity, but not another policy-affected item moving in the opposite direction, such as tobacco. In any case, Westpac Senior Economist Justin Smirk has previously calculated that removing electricity before calculating the trimmed mean makes very little difference to the result.)
There are, however, plenty of categories where the trend rate of inflation is materially different from the overall inflation rate. For example, insurance prices have risen at an annual rate of more than 5% over the past 20 years. On the other side, prices of garments, adults’ shoes and household appliances have fallen overall over the same period. Not all these categories will be trimmed out in a particular quarter. It depends on how extreme those trend price changes are relative to the overall distribution of inflation by component.
The important point here is that the RBA’s mandate is not to get prices of every single category in the CPI basket increasing at 2–3%. As long as the average is in the range – and, per the latest Statement on the Conduct of Monetary Policy, heading towards its midpoint – then the mandate is being fulfilled. There will always be items with price inflation a long way from that range. Changes in relative prices are a natural and desirable part of a well-functioning market economy. They are central to the way buyers and sellers – in this case, consumers and producers – respond to real-world changes.
This is why one should be wary of arguments that the RBA should set monetary policy in a specific way because some subset of the inflation basket is showing price growth above the 2–3% target range, even though overall inflation is at or near target. It depends on the context.
For example, for much of the inflation-targeting period since the mid-1990s, services inflation has run faster than goods inflation. Is that difference a reason to run tight policy? Not necessarily. It depends on whether the low-inflation category is likely to stay that way.
That question became particularly salient in the post-pandemic period. Goods inflation increased considerably as supply chains were disrupted. If goods inflation normalised (to below-target rates) as supply chains normalised, then services inflation could also normalise to its above-target average rate, and overall inflation could settle within the target as intended.
The wrinkle would be if goods prices reverted some way back to their pre-pandemic norms. That would mean goods inflation would spend some time below normal, possibly involving some outright price declines, before presumably reverting to average. If overall inflation had returned to target on the back of a temporary period of below-normal goods inflation, then clearly above-trend services inflation would not be compatible with keeping inflation sustainably in the target range once goods inflation normalised.
This has been part of the concern in the US. Like Australia, services inflation typically runs faster than goods inflation there. Rolling 10-year average inflation rates have generally been 1ppt or so higher for US services than US goods, similar to the post-2000 picture in Australia. But unlike the situation in Australia, goods inflation fell noticeably below its pre-pandemic average recently, flattering the overall rate of inflation. Given recent tariff announcements, it seems unlikely that goods prices will continue falling for an extended period. Indeed, a period of US goods inflation above historical averages is probably on the cards. Falling housing-related inflation may help offset this, but one can’t help thinking that the sustainability of US inflation near target is more fragile than is the case for Australia.
Some may object that current rates of inflation for services in Australia are still well above pre-pandemic rates. Recall, though, that overall inflation undershot the RBA’s target for several years pre-pandemic, partly because services inflation was running at a persistently below-average pace. So that is not the best benchmark period for comparison. Current services inflation is also a bit above the average of the first decade of the 21st century, but not drastically so. And the ongoing unwind in rents and insurance inflation, as pandemic effects continue to wash out, should help narrow the gap.
Perhaps we should be more concerned that, despite weak household demand, goods inflation didn’t see the same period of decline as in the US. Part of this might be the pause in the long-running trend decline in prices of audiovisual and computing equipment in Australia. We also observe, though, that homebuilding costs are included in the CPI in Australia but not in the US, and that the pandemic surge in these costs is unwinding.
Whatever the main driver of the difference, the main point here is that goods inflation is not temporarily flattering the total inflation result in Australia. And since electricity prices are currently excluded from the trimmed mean inflation rate, they are not flattering underlying inflation either. Economy-watchers, including the RBA, can therefore be confident that the 3.2% result for trimmed mean inflation over 2024 (and an annualised rate of 2.7% over the second half of the year) is indeed giving a sufficiently accurate picture of current inflation pressures in Australia, and act accordingly.
