Stats NZ has a new website.

For new releases go to

As we transition to our new site, you'll still find some Stats NZ information here on this archive site.

  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
What could you live without? Using the CPI basket to define the basics

The consumers price index (CPI) basket of goods and services can be used as a barometer of the changing tastes of New Zealanders. What we add and remove over time provides a window on society, reflecting the availability and affordability of different commodities.

100 years of CPI – basket changes highlights how the New Zealand CPI basket has evolved over the 100 years to the June 2014 quarter.

Options for weighting

The household expenditure patterns we use to weight the CPI basket also provide insights into consumer behaviour. Using our research into suitable weighting methods for the household living-costs price indexes (HLPIs) – the household-group-specific inflation rates that we’ll begin to publish in early 2016 – we can see which goods and services are more commonly purchased by ‘typical’ households than by more affluent households.

We are looking at which weighting method will best meet our customers’ needs. One option is to calculate weights using total expenditure for each group of households. This is the method used for the CPI, the household-sector weighted approach (in the CPI literature it is often called ‘plutocratic’ weighting).

Alternatively, we could calculate the expenditure weights for each household first and then take an average of these. This is the household weighted approach (often called ‘democratic’ weighting, since each household’s expenditure has an equal impact on the aggregate weights).

Using the second approach, household-group inflation would represent a ‘typical’ household rather than the household sector as a whole.

Table 1 summarises the weighting approaches that may be appropriate for different purposes.

Table 1

Weighting method options 
  Household-sector weighted (‘plutocratic’)  Household weighted (‘democratic’) 
Summary of approach We treat inflation experienced by the group as a single ‘super-household’  Inflation experienced by a ‘typical‘ household within the group 

May be conceptually appropriate for:

  • macroeconomic measurement
  • deflation of (aggregate) nominal expenditure
  • indexation of monetary payments, where the aim of indexation is to maintain the aggregate purchasing power for the group receiving payments  

May be conceptually appropriate for:

  • measuring changes in real median incomes
  • indexation of monetary payments, where the aim of indexation is to maintain the purchasing power for the ‘average household’ of the group receiving payments 
Practical implementation Most expenditure estimates are readily available from CPI weights. We can use the Household Economic Survey to allocate expenditure to population groups Requires micro-level calibration of household expenditure, before aggregating household-specific expenditure shares. This will use the CPI and Household Economic Survey in a more data-intensive way 

Separating necessities from luxuries

The differences between the two approaches provide insights into goods and services that are consumer ‘necessities’ and those that are ‘luxuries’ (see figure 1).

Goods and services with bigger weights using household-sector weighting, than using household weighting, make up a greater proportion of the expenditure of higher-expenditure households – this suggests they are ‘luxuries’. Goods and services that receive higher weights for households with more ‘typical’ expenditure we classify as ‘necessities’.

The data we used for figure 1 is the average class-level (of our classification) expenditure patterns for 2008 and 2011 for households in the middle income quintile.

Figure 1

A classification of commodities

Image, word clouds showing necessities versus luxuries, using CPI data.

The data shows us that the commodities we classify as necessities today are similar to those contained in the early CPI baskets – which had a clear focus on necessities. In 1914, price collection was limited to food, rent, and fuel and light.

Tobacco being classified as a necessity will surprise some, given its known negative effect on health, and current public policy aimed at reducing consumption. However, this classification simply means that a ‘typical’ household spends proportionally more on tobacco than does the average of all households. Tobacco was included within the food group in the 1914 basket!

Clothing being classified as a luxury illustrates how using class-level expenditure applies a broad brush. It overlooks distinctions within the class. Interestingly, the Report on the cost of living in New Zealand, 1891–1914  [PDF, 87 pages] discussed the treatment of clothing, stating “Clothing is an admitted necessity, but a large proportion of expenditure thereon may well be looked upon as luxury ... expenditure on clothing is to a great extent dependent on what surplus of income is available after the needs of housing and food are satisfied”. We have included clothing in the CPI since 1924.

Classifying petrol and telecommunication services as necessities reflects how society has changed over the past century. It was 1955 when we first included petrol and other private motoring costs, along with telephone services, in the CPI. Like clothing, classifying cars as a luxury does not reflect within-class variation in consumption patterns. As implied in our classification of petrol, some lower-value cars are necessities.

Figure 2

Graph, Comparison of weighting methods, superannuitants households, June 2011 quarter.

Figure 2 takes a more detailed look at the numbers – this time focusing on superannuitant households. (Those with one or more people aged 65 years or older who receive New Zealand Superannuation, and where a superannuitant had the highest household income).

The superannuitant household weights reflect this group’s specific expenditure patterns (which differ from those of ‘all households’). Also, there are technical differences in the definition of expenditure we use in the HLPIs compared with the CPI – we change from an ‘acquisition’ to a ‘payment’ conceptual approach.

Electricity, rent, and local authority rates are more highly weighted using the household weighted approach than the household sector approach. This means that expenditure on these goods and services is a proportionately more significant expense for ‘typical’ superannuitant households than for superannuitant households treated as a single group.

It is perhaps unsurprising to find that medical services, international flights, and new cars make up a greater share of expenditure by higher-expenditure superannuitant households.

Your views on the design of the household living-costs price indexes

We are designing the HLPIs to meet customer needs as well as we can, so we welcome your views.

Please send comment to us at (with ‘HLPI’ in the subject line).

The weighting approach is one of several design issues we are considering. Other issues include the household group definitions, and appropriate methods to measure quality-adjusted price change for interest payments.

See New measures of inflation for groups of households presented at the July 2015 NZ Association of Economists conference for more about these topics.

For more information on the HLPIs contact:

Alan Bentley
Wellington 04 931 4600 or 0508 525 525

  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+