Today saw the release of the Q4 inflation date. Headline CPI was up just 0.2% q/q for an annual increase of +1.7% (down from +2.3% in Q3), this was a little weaker than the market had been expecting and falls outside the RBA’s 2-3% target range. However, the measures more closely watched by the RBA (namely the Trimmed Mean and Weighted Medians) were stronger and slightly above the market expectations at +0.7% q/q and +2.2% y/y and +2.3% y/y respectively (average +2.25% y/y). Whilst these increases were down from the previous quarter (+2.5% y/y average) they remain within the RBA’s target range. The stronger than expected “core” data is the focus for the markets and is the reason that the A$ has rallied almost a full US cent on the news, as a rate cut from the RBA looks rather less likely in the near term following the stronger than expected core data.
So why the big difference and what is “core” inflation as defined by the RBA? To answer those questions we need to understand a little about what the headline CPI measure is.
CPI in Australia measures quarterly price changes in a “basket” of goods and services that makes up a high proportion of the expenditure of the “CPI population group” – this is metropolitan households. Different weightings are then attached to the various groupings within the basket and an average calculated. Price changes measured in this way will contain two components. Persistent inflation and “noise”. The noise component refers to price changes that can be volatile but do not have an impact on persistent, trend inflation. Examples of this kind of noise might be changes in food prices caused by weather variables, or the infrequent setting of tax levels on goods or services (see the example of GST in the graph below). Of course some of this “noise” may end up being persistent but much will not.
What policy makers need is a measure of inflation that distinguishes between persistence and noise. In other words, a measure with a high “signal to noise” ratio. One way of doing that might simply be to remove the more volatile components from the CPI basket (the so-called exclusion method). These components tend to be food and energy. This method has two main drawbacks. The first being that some of these volatile component price changes will be persistent and yet are excluded. The second is that for any one period there may be other volatile components within the basket that are not excluded.
A method that has become popular amongst a number of Central Banks, including the Reserve Bank of Australia, is known as “trimmed mean”. Simply put this entails removing (or trimming) the extremes (both high and low) of price changes in any one period and then recalculating the weighted average. Which items are trimmed are not pre-selected, but are determined based purely on their relative position within the total spread of price changes. The RBA’s Trimmed Mean measure excludes both the top and bottom 15%. The sometimes quoted Weighted Median is simply a trimmed mean excluding both the top and bottom 25%.
Of course some prices vary consistently on a seasonal basis and this needs to be accounted for. For instance if a particular component’s price is always increased just once a year in March, then that component might be always trimmed since its one-off price rise would put it in the top 15% for that quarter. However, the measure needs to account for that component’s price change over the year. To get around this problem the Trimmed Mean measure uses seasonal adjustment. In the case above, although the nominal price rise might put that component into the top 15% (and therefore trimmed), once seasonally adjusted the increase becomes more “normal” and would fall out of the top 15% and therefore remain within the measure. So what does all this mean in terms of the inflation level?
Whilst the “headline” CPI number may well be the number splashed across the front pages of the newspapers, the Reserve Bank will be looking more closely at their Trimmed Mean and Weighted Median measures to try and get a better “signal” from the data as to actual inflationary trends.
The graph below gives some indication of the variation between the headline CPI and the RBA’s Trimmed Mean measure. As we can see, whilst headline CPI fell sharply during the economic slow-down in late 2008 and 2009, the Trimmed Mean measure remained resolutely high; outside the RBA’s inflation target level. Much of the slowdown in price growth at that time could therefore be seen as “noise”.
Since that time the Trimmed Mean has remained relatively stable within the RBA’s target range despite occasions when the headline CPI measure has either been above or below the target range (as now).
What we also notice from the graph is that the Cash Rate and inflation measures are currently at very similar levels (i.e. when inflation-adjusted by the average of the Trimmed Mean and Weighted Median, the “real” Cash Rate is virtually zero). As we can see this is historically a rather rare occurrence; today’s data sees the real Cash Rate at +0.25% and suggests that the Cash Rate is already at a very low level given the rate of core inflation. Having said that, the real Cash Rate has actually moved slightly more positive in recent quarters, having been negative for the three quarters from Q4 2013, suggesting a possibility of some (slight) room for Cash Rate cuts if the RBA deem them necessary in the future.