3.4  Taxation

On the basis of each person’s gross income in the survey year, TAXMOD calculated income tax liabilities and assigned these to households, also taking into account household composition and other available information. The amount of tax calculated by TAXMOD may, however, differ from the amount of tax that a household actually paid. One reason is that, for business and investment income, tax payments may have depended on income earned in past years as well as in the survey year.

The total consumption tax take, consisting principally of Goods and Services Tax (GST) and excise duties, was obtained from the System of National Accounts and allocated to households in proportion to their expenditure over the survey period. Excise duties were not allocated according to expenditure on specific products (alcohol, cigarettes, petrol, etc).[6]

These methods were applied to both the 1987/88 and 1997/98 data.

3.5  Government spending and taxes not included

This study allocates those government benefits and taxes that can reasonably be attributed to households. It does not attempt to allocate “non-social” government expenditure such as roading, defence and debt repayment to households; nor does it attempt to allocate company tax to households. Around three-quarters of tax revenue and two-thirds of government spending are included in the analysis, however, which means that, on average, households will be seen as paying more in attributable tax than they receive in attributable benefits. This is an approach which is common to many other studies of final household incomes.

There is a tradition of studies, however, which do attempt to allocate all government expenditure and all government revenue to households (eg, Gillespie 1965, Musgrave, Case and Leonard 1974, Ruggles and O'Higgins 1981). A number of these types of studies were conducted in New Zealand in the 1980s (eg, Snively 1986, Income Distribution Group 1988, Department of Statistics 1990).[7] Other studies are also differentiated by including local government taxes and expenditure in their analysis of final income (O'Higgins and Ruggles 1981) or by including other forms of non-cash income such as imputed rent to owner-occupiers (Smeeding, Saunders, Coder, Jenkins, Fritzell, Hagenaars, Hauser and Wolfson 1993, DeNavas-Walt et al 2003).

This study does not include these wider benefits and taxes, chiefly because there is no clear conceptual basis for allocating many types of benefits and taxes to individual households. In other cases, while there might be a conceptual basis for allocation there is no data available from the HES or other sources to enable this to happen. Some taxes and benefits also fall on people who do not live in private households, such as publicly-funded long-term residential care for the elderly.

3.6  Population estimates

The steps outlined above provide information on final incomes for households in the HES, but this information needs to be translated into estimates for the whole population of New Zealand. This translation requires weighting the different types of households in the survey to reflect their prevalence in the population. In preference to earlier weightings used in the HES and in TAXMOD, a new set of “integrated weights”, developed by Statistics New Zealand, was applied to both the 1987/88 and 1997/98 results. Integrated weighting is described in detail in Statistics New Zealand (2001b).[8]

Population estimates are subject to sampling errors but these have not been calculated for all of the results reported in this paper. Using replicated sampling techniques, TAXMOD was used to construct a selection of confidence intervals for estimates from the 1997/98 HES. As expected, this analysis suggests that more reliance can be placed on changes in broad summary measures than on changes in particular deciles or household types. Results indicating relatively small differences between deciles or household types need to be interpreted with caution.

3.7  Other adjustments

3.7.1  Consistency with the Crown accounts

All government expenditure and revenue data used in the study for both 1987/88 and 1997/98 were reconciled to the Crown Accounts Analysis produced by Statistics New Zealand as part of the New Zealand System of National Accounts. Across the different types of expenditure or taxes, relative allocations per household were pro-rated upwards or downwards to make the population totals equal the official national figures.

The HES is a survey of private households and excludes individuals in non-private dwellings. Expenditure and taxation, however, covers the whole population. Census data from 1986 and 2001 were used to allocate expenditure and taxes to individuals living in non-private dwellings, according to their age, income and assumed eligibility for government services. This expenditure and tax was subtracted from the national totals before the study results for private households were pro-rated.[9]

3.7.2  Price adjustments

In order to get an accurate comparison across time, all money figures were adjusted to the December quarter of 1997, using the All Groups Consumer Price Index. Income and expenditure in 1987/88 was inflated by a factor of 1.364.

3.8  Income deciles and equivalence scales

This study reports a number of results by income decile. A single ranking-and-dividing system is required so that “decile 3”, for example, refers to the same group of people and households, regardless of whether the analysis is of market income, government benefits, tax payments, gross income, disposable income or final income. For this purpose, the study used equivalent disposable income as the measure by which deciles were constructed. Disposable income is a household’s income from all sources after income tax has been deducted—it is a household’s “cash in hand”. Equivalising this income takes account of differences in the demands on households’ resources. This study uses the LIS(0.5) equivalence scale, which scales down household disposable income according to the square root of the number of individuals in the household, regardless of whether they are adults or children (Atkinson, Rainwater and Smeeding 1995).[10] Equivalent disposable income roughly corresponds to common ideas in the community about who is better off and worse off and is therefore a proxy for people’s level of welfare.

In constructing deciles of equivalent disposable income, households were weighted by their number of occupants, so that a household of six people “counts” six times as much as a single-person household. Individuals were ranked by their household’s income (where all members of a household have the same rank) and divided into ten even groups. Decile 1 therefore contains the least well-off 10% of the population and decile 10 contains the most well-off 10% of the population. Deciles of equivalised disposable income are used in all the tables and figures in the paper (apart from when calculating Gini coefficients in Table 3). Weighting households by size means that the economic welfare of each individual in society counts equally when measuring income distribution (Danziger and Taussig 1979). The alternative, which is to give each household equal weight, reveals something about the economic differences between households but begs the question of the number of people affected by those differences.[11] Weighting households by size also makes for consistent comparisons over time, since the size of households in different deciles has changed between 1987/88 and 1997/98 (see Appendix Tables 6 and 7).

Equivalised income is used only to rank people for the purposes of creating deciles. All of the monetary values shown in this paper are actual dollars and are not equivalised. This is because the LIS(0.5), and other equivalence scales, have been designed to adjust for the demands on households’ disposable incomes—the housing, clothing, food, and other needs of different types of households. Households’ disposable incomes do not have to pay for publicly-funded education and health services, and therefore the education and medical needs of different types of households are not (at least in theory) reflected in ordinary equivalence scales. Therefore it may not be appropriate to use ordinary equivalence scales to adjust household incomes that include notional income from the receipt of government education and health benefits (Radner 1997). In other words, just as the relative incomes of different groups of households change when health and education benefits are included in the definition of income, the relative needs of different groups of households also change and equivalence scales should change. It is beyond the capacity of this study to develop an equivalence scale for final income, and therefore the dollar figures for income are left unadjusted for household needs.[12] However, while household incomes are not equivalised, Appendix 2 presents the results of the study by various different categories of households at different stages of the life-cycle, and thus provides a partial equivalisation analysis.

Other final income studies use a variety of ways of weighting, measuring and rank-ordering household incomes. The various possibilities are discussed in O'Higgins, Schmaus and Stephenson (1989). Ranking according to equivalised disposable income, but reporting actual final income, as this study does, follows the same methodology as the annual studies conducted in the United Kingdom by the Office for National Statistics (Lakin 2004).

3.9  Household types

Appendix 2 contains a series of tables showing the components of final household income by categories of households. These “life-cycle” household types are those previously included in the Department of Statistics (1990) study with minor changes to reflect an increase in the age of eligibility for Superannuation: the 1987/88 analyses refer to people aged 60 or more but this has been changed to age 63 or more in the 1997/98 analyses. In this classification, “other family group” refers to households which include a family together with one or more other people, and “non-family household” refers to households in which none of the occupants of the household are related to each other.[13] “Children” means dependent children who are under 18 years of age, are not receiving a benefit, and are not in full-time work. The number of households in each category in 1987/88 and 1997/98 is shown in Appendix Tables 2 and 3.