Inflation figures just don't add up

Fortunately, the Reserve Bank did not raise interest rates on Tuesday. Had it done so, it would have been because the June-quarter inflation figures, at face value, show that underlying inflation in the six months to June 2011 had climbed outside its inflation target.

But as the Reserve surely knows, those figures cannot be taken at face value. They are almost certainly overstated. And we know that because the Australian Bureau of Statistics, which compiles them, has told us so.

In a post-mortem of the previous cycle of CPI data, which ran from 2000 to 2005, the bureau found that its methodology over time created an ''upward bias'' in the CPI figures. The methodology assumes that consumers keep buying the same basket of goods and services, regardless of their price. And over time, that increases the relative weighting of items whose prices are rising rapidly, and reduces the weighting of those whose prices are falling or relatively stable.

Over time, the basket of goods and services that comprises the CPI becomes increasingly unrepresentative of the real world of consumer purchasing, because it assumes that consumers pay no heed to price signals. Without correct weightings for each of the 90 categories in its basket, the CPI figures are inaccurate, and so are their derivatives, such as measures of underlying inflation.

This is serious. There is a flaw in the inflation data that we should have known about, but didn't. And it almost cost us dearly.

The bureau's post-mortem estimated that between 2000 and 2005, this upward bias had overstated the level of inflation by a cumulative 1.2 percentage points. In the year to June 2005 alone, inflation was overstated by 0.4 percentage points. That was the fifth and final year of the cycle. Inflation was reported as 2.5 per cent, but when the bureau surveyed what households were actually buying it found the true inflation rate was 2.1 per cent.

What does that tell us about the June 2011 figures? They came at the end of year six of this data cycle - and it is clear that the same pattern has overstated the figures again.

The true level of inflation is lower than the CPI figures report. And even in the first half of 2011, underlying inflation is lower than the 3.5 per cent reported last week.

Indeed, on the original 2005 weights, both the trimmed mean and the weighted median, in the first half of 2011, showed underlying inflation running at an annualised 2.6 per cent. Both of them rose 0.8 per cent in the March quarter, but by just 0.5 per cent in the June quarter. No crisis whatever in sight.

I don't pretend that that is the correct inflation figure. I chose a methodology that assumed that our relative spending distribution between different items is fixed, which is just as unrealistic as the CPI's assumption that we keep buying the same items regardless of their price. The true inflation rate is somewhere between them - and sometime next year, when the bureau has time to go through the new spending data it is about to publish, it will work out a reconciliation.

But we ought to know about all this - and we don't. Unfortunately, the bureau issued its warning in December in a long paper titled Outcome of the 16th series Australian Consumer Price Index Review. The paper detailed the changes the bureau plans to make to the consumer price index (CPI) from the September quarter - and other changes it would like to make, if only the government would give it the money to do so.

One of the changes on its wish list is to reduce the length of each CPI series - the period between each revision of the basket - from six years to four years, because, it warned, the data increasingly overstates the inflation rate the longer it is left unchanged.

''As consumer expenditure patterns change over time, a fixed set of weights used in the CPI runs the risk of becoming unrepresentative, and can lead to potential item (or upper level) substitution bias,'' it concluded. ''Item substitution occurs when households react to changes in relative prices by choosing to reduce purchases of goods and services showing higher relative price change, and instead buy more of those showing lower price change.''

Take bananas. The CPI methodology assumes that we are still buying just as many bananas at $12 a kilogram as we did when they were $2 a kilogram. Its numbers assume that $1.64 of every $100 we spend now goes to buying fruit, up from just 95¢ in 2005. That's rubbish, but it means that rises in fruit prices now count for almost twice as much in the CPI, which inflates its bottom line.

The same is true in reverse. With each new model of TVs, computers, sound systems, mobile phones and even cars offering us far more oomph for the same price as before, the CPI assumes that instead we are still buying 2005 models of all of them, at very much lower prices.

That is patently incorrect, but it means the the CPI is calculated on the assumption that we now spend only 53¢ in every $100 on buying computers and audio-visual equipment, down from $1.53 in 2005. And that means each fall in their prices now counts for only a third as much as in 2005.

So you see how the upwards bias creeps into these figures. As the bureau put it bluntly: ''A fixed-price Laspeyres index will overstate the price change of the whole basket as it is unable to take account of changes in the substitutions that consumers make in response to relative price changes. ''The upward bias appears more marked after four years … The longer the period between re-weights, the larger the potential bias.''

So why can't the bureau just re-weight the index every year? Because its data comes from a huge survey of household spending, which costs tens of millions of dollars. That's why it has been done just once every six years, but the bureau now wants to do one every four years.

The government is still chewing over this one, but it's surely a no-brainer. A second initiative the bureau is exploring is the potential to gather scanner data from supermarkets, retail stores and other outlets to get more up-to-date pictures on what people are really spending their money on. That has a long way to go.

Immediate relief, however, is coming. In September the bureau will release the first results of the 2009-10 household expenditure survey. It will then use these to create a new basket of goods and services for the CPI, which will make its debut when the September-quarter figures come out in late October. Then we will have some accurate data, which will show the real inflation trends. We haven't had that for quite a while.

Read more: http://www.smh.com.au/business/inflatio ... z1U1Po7vrC

second initiative the bureau is exploring is the potential to gather scanner data from supermarkets, retail stores and other outlets to get more up-to-date pictures on what people are really spending their money on. That has a long way to go.