Saturday, February 3, 2018

CPI a more accurate measure of living costs than we imagine

Ask any pollie, pollster or punter in the pub and they'll all tell you there are no political issues hotter than the soaring cost of living. But this week the Australian Bureau of Statistics issued its consumer price index for the December quarter.

Oh no. It showed prices rising by 0.6 per cent in the quarter and a mere 1.9 per cent over the year to December.

That's a soaring cost of living? What are these guys smoking? Has the government got to the statisticians? Or do the bureaucrats sit in some office in Canberra making up the numbers?

None of the above. In truth, the bureau puts an enormous amount of expertise, care and effort into making the CPI as accurate as possible. Which is not to say the indicator is without its limitations – nothing in the real world is.

The care is shown in an explanatory paper the bureau issued this week to accompany its latest six-yearly updating of the index.

The CPI is purpose-built to measure changes in the price of a fixed quantity of goods and services bought by people living in metropolitan households.

"Metropolitan" means the eight capital cities, and the households include wage-earners, the self-employed, self-funded retirees, age pensioners and social welfare beneficiaries. That covers almost two-thirds of all Australian households, leaving out only those in regional areas.

The index measures the change in the price of a metaphorical basket containing fixed quantities of goods and services bought in each of the capital cities. It looks at thousands of prices of individual items, divided into 87 expenditure classes, 33 sub-groups and 11 major groups.

These are: food and beverages (accounting for 16 per cent of the total basket), alcohol and tobacco (7 per cent), clothing and footwear (4 per cent), housing (23 per cent), furnishings, household equipment and services (9 per cent), health (5 per cent), transport (10 per cent), communication (3 per cent), recreation and culture (13 per cent), education (4 per cent), and insurance and financial services (6 per cent).

How does the bureau know which particular goods and services to include in the basket and, more especially, what "weight" (relative importance) to give each class of expenditure?

Every six years it conducts a survey of more than 10,000 households, asking them to keep diaries of the spending they do. As spending patterns change over time, it updates the contents and the weights given to the items in the basket.

This week it applied new weights derived from the household expenditure survey it conducted in 2015-16.  From now on, however, the weights will be updated yearly.

The bureau checks the prices consumers are being charged by regularly visiting shops and offices, by phoning businesses, and, increasingly, by checking online supermarket sites and records of scanner transactions in stores.

It checks the prices of items at least once a quarter, but more frequently if prices – petrol, for example – keep changing. It aims to show the average price charged during the quarter.

It measures the retail prices we actually pay, so prices include the goods and services tax, and excise taxes, embedded in them, but also any government price subsidies for items such as private health insurance or childcare.

It takes account of widespread "specials", provided the items are of normal quality. It seeks to measure "pure" price changes, meaning it tries to exclude price changes attributable to a change in the quality or quantity of the latest version.

If some producer tries to disguise a price increase by leaving the price of a can of baked beans unchanged, but reducing the amount of beans, the bureau uses the actual price increase per gram.

When the latest laptop or mobile phone is more powerful than the previous model, or does more tricks, the bureau tries to take account of this quality improvement by calculating the underlying or "pure" price change – often a price fall.

But if the bureau takes so much care to measure price changes accurately, why do its figures invariably seem much lower than our impression of the price rises we've experienced?

Short answer: because we don't take nearly as much care as it does. We don't keep meticulous records, but form impressions. And, as behavioural economists tell us, our memories of prices changes are subject to predictable biases.

Price changes we don't like stick in our minds, while those we don't mind are soon forgotten. We remember clearly a few big price increases – the shock we got when we saw our quarterly electricity bill – but don't remember price falls (of which there are far more in these days of digital disruption). And it never occurs to us to take account of all the many items whose prices hardly change.

As a statistician would say, we don't attach the right weights to the price changes (including zero changes) that come our way.

So, for instance, we carry on (justifiably) about ever-rising power prices, but forget that electricity accounts for just 2.2 per cent of the average household's total consumer spending.

Of course, no particular household's experience is likely to be perfectly represented by such a broad average. The index lumps together people in different cities, smokers and non-smokers, drinkers and non-drinkers, renters, mortgagees and outright home owners.

The bureau tries to reduce this problem by also publishing special living cost indexes for certain types of households. Over the year to September, in which the CPI rose by 1.8 per cent, living costs rose by 1.5 per cent for employee households, 1.6 per cent for self-funded retirees, 1.7 per cent for age pensioners, and by 2.1 per cent for unemployed households.

Sorry, but the notion that the prices I pay rose way more than other people's did is just another of our happy self-delusions.