An Australian inflation fixation

I want to canvass three thoughts today around the RBA's inflation target, the paradox of effective monetary policy, and fudging the inflation numbers.
 
First, it is interesting to recall that the RBA's inflation target of 2-3 per cent per annum is higher than pretty much all of its peers around the world. The ECB, RBNZ, Bank of Canada, Bank of England, and Federal Reserve all have materially lower inflation targets of 2 per cent per annum, or slightly less than 2 per cent.
 
This would not be a concern had the RBA kept inflation within the middle of its target band. But, as I explain below, average inflation in Australia over the last decade has been well above the RBA's implied 2.5 per cent target. This naturally begs the question of whether the RBA's higher target has induced an upward bias into Australian inflation outcomes (and hence expectations).
 
Second, as I argued frequently last year, I have never really been comfortable with the popular belief that central bankers should set interest rates by responding to past inflation innovations (for example, last quarter's result). If you are changing the cost of credit today to influence consumer prices over the next two years, which is what the RBA thinks is its impact horizon, then you must, by definition, do so on the basis of your projections of future, not past, inflation (obviously there is some persistence in inflation over time, and in this respect the past is informative of the future).
 
In turn, if your forecasts of future inflation are accurate and your policy settings in response to these projections similarly correct, you should be able to keep changes in consumer prices within your chosen target range absent any utterly unpredictable external events (such as the September 11 attacks), or innovations in prices that are demonstrably temporary in nature, like the imposition of a tax.
 
So the common financial markets idea that you should wait until inflation – be that the statistical 'core' or true headline – pops up beyond, or towards the top of your target band, prior to putting in place the right policy settings, seems to me to be quite misplaced. Indeed, self-evidently so. It would be no different to investing in shares purely on the basis of their past price movements.
 
What this tells us is that a largely error-free central bank will rarely permit inflation to consistently exceed its target range. Yet this efficacy in and of itself creates a communications challenge: while interest rates might move strikingly in accordance with the preferences of the central banker, consumer price inflation should generally be stable. That is, you would never have inflation problems (think of Australia during the great moderation between the mid 1990s and mid 2000s).
 
And so the paradox of good monetary policy is that you should not have a counterfactual of bad monetary policy against which to rationalise your remedial actions. Of course, Australia does have an uncomfortable number of said counterfactuals: the experience up to the early 1990s and the period prior to the GFC are two that spring to mind.
 
My third thought concerns the disconnect between the inflation measures that are popularly used to benchmark central bank performance, the inflation numbers that actually count for consumers, and the inflation outcomes that central banks are legally mandated to focus on. Critically, the first is quite different from the latter two.
 
Think about it: on Monday a monthly gauge of inflation reported that Australian consumer prices had risen by a very significant 0.4 per cent month-on-month (or nearly 5 per cent annualised). But a sanitised 'core' version of this gauge implied that consumer prices had, in fact, fallen over the month, since it removed the influence of fruit and vegetables, amongst other things. I am not sure about you, but my family eats fruit and vegetables, and we probably consume many of the other expenditure classes that are overlooked by this 'core' measure!
 
In the real world, headline inflation is what matters for our cost of living. And even setting aside this fact, we should be focused on actual inflation since this also influences consumers' future inflation expectations, which ultimately determines their wage claims.
 
A second problem here is that central banks like the RBA are not paid to keep statistical abstractions such as 'core inflation' within their target range, which in Australia is 2-3 per cent. The RBA itself has noted this. They are paid to manage headline inflation – not a cleansed version of the data that conveniently ignores the highest growth spending categories.
 
In this context, year-ended consumer price inflation in Australia has averaged above the RBA's target 2-3 per cent per annum band over the last decade (specifically, since January 2000). More pointedly, it has averaged 3.1 per cent per annum, which is well above the RBA's assumed through-the-cycle target of 2.5 per cent per annum. (For what it is worth, the median and modal outcomes are also above this target.) Even if we stretch the time horizon back to 1993, headline inflation has averaged near the top of the RBA's target band.
 
A final question I'd like to pose concerns the cleansing techniques the RBA and ABS use to come up with these core abstractions. In short, they remove all those spending categories that are ranked in the top 15 per cent of price changes in the quarter. They also strip-out the bottom 15 per cent of spending categories. The 'trimmed mean' inflation proxy is then the middle 70 per cent of categories weighted by dollars spent.
 
The problem with this approach is that you can have important expenditure classes that are driving up consumer price inflation, and which consistently rank in the top 15 per cent of price movers, that are completely eliminated from your inflation measure (think of any group of consumer goods that happen to be on a price run).
 
You can also have a skewed distribution of price changes, which, for instance, might mean that the rate of change in the top 15 per cent of spending categories is greater than the bottom 15 per cent. The symmetrical truncation of the top and bottom tails of the distribution might therefore remove more high inflation than low inflation.
 
The second 'core' inflation benchmark the ABS and RBA use is the so-called 'weighted-median', which is even more rudimentary than its colleague the trimmed mean. To get the weighted median, the ABS/RBA lop off the top 15 per cent and bottom 15 per cent of price movers, and rather than averaging the remaining 70 per cent of categories, simply select the middle or median category (ie, a single class of goods).
 
Now, I think it is fine – indeed valuable – to use these alternative inflation proxies as analytical tools. But that is not what happens. There seems to be a flawed pattern of trying to judge the performance of monetary policy using core rather than headline inflation. This is both practically for consumers, and legally under the standard interpretation of the 1959 Reserve Bank Act, inappropriate.
 
These thoughts were germinating when I stumbled across some similar sentiments expressed by the European Central Bank in the Financial Times. And I could not have put it better myself:
 
"The ECB takes little comfort from "core” inflation measures, a concept which Jürgen Stark, executive board member, described this week as "well suited for central bankers who don't eat or drive”. Speaking on Thursday in Bologna, Italy, Lorenzo Bini Smaghi, a board colleague, argued that imported inflation could not be ignored. To avoid knock-on, or "second round” effects, prices in the eurozone, he said, had "to be significantly more constrained than those of emerging countries.”