2  Construction of the data and the cohort approach

In this section we first describe the data, and how we have constructed measures of saving and synthetic cohorts. We then present a first glimpse of household saving patterns based on these cohorts.

2.1  The Data

We have used the income and expenditure data from the HES to estimate saving as a residual. We do this fully cognizant of the limitations of the HES.[1] Our defence rests largely on the fact that there is no other source of micro-data for examining household saving behaviour. Moreover, analysts in other countries have used similar data sources, particularly the Family Expenditure Survey in the United Kingdom (Attanasio and Banks, 1998) and the Consumer Expenditure Survey in the United States (Attanasio, 1998). We have tried to eliminate some outliers, and we have a large sample, which might arguably compensate for the underlying deficiencies. But we accept that our results are only as good as the survey data from which they are derived.

The definition of consumption that we have used when deriving the saving rate excludes items that are more properly considered as forms of investment and hence are a type of saving. In particular, to obtain the estimate of “current” consumption expenditure we removed from HES total expenditure, expenses on education, life and health insurance, purchases of durable goods, medical expenses, repayments of mortgage principal, and contributions to savings. In other words we attempt to construct a measure of expenditure, which would result in an "economic” view of saving.[2] Consequently, our consumption and saving variables differ from those that may have been available for previous studies and from the definitions used for national aggregates.

The data cover the years 1983-48 to 1997-98. We refer to these years by the latter year; ie, 1984 and 1998. A total sample of 50,624 households was available over these 15 years. We have removed all observations where household disposable income was reported as negative (some 330 households)[3]. Further we have truncated the sample to remove all observations where the age of the household head was reported as less than 19 or greater than 74 at the time of the survey. This left us with a sample of 46,269 households.

Our justification for restricting the sample to the age range 19-74 is that those less than 19 were not considered important for studying lifetime saving patterns, while amongst the elderly, the HES does not cover institutions so those living in rest homes are not included. This means that in the upper age groups we have an incomplete sample based only on those living on their own or as part of another household and this group may not be representative of the full population of the elderly.[4]

Rather than taking the ratio of saving to disposable income, we have chosen to follow Attanasio (1998) and calculate saving rates by the ratio of saving to consumption. This has the advantage of being defined even when reported disposable income is zero[5].

Ideally, a complete accounting for income and saving requires both flow measures and a household balance sheet to track stock changes. The HES provides no measures of the stock of household wealth. As a consequence, any contributions made by an employer to a private pension fund are not recorded as saving; and any withdrawals from a pension fund are counted as income in the year received rather than “dissaving”. Fortunately, pension schemes (outside the public superannuation scheme) play a relatively minor role in New Zealand[6].

Should the unit of analysis it be the individual or the household? There is no clear answer to this; both have advantages and drawbacks. The HES reports income for each individual in the household, and expenditure on a household-wide basis. This means that to compute saving, one needs to either:

1. allocate expenditure to individuals and then subtract from reported incomes to find individual saving levels; or
2. combine the incomes of individuals to a total household income and subtract reported expenditure.

We have chosen the second option, believing that many saving decisions are taken on a household basis, and considering that allocating expenditure to individuals would have created some spurious saving estimates, especially for those household members who are not participating in the labour force. It must however be recognised that in multi-generational households the saving of working age members could be offset by the dissaving of younger and elderly members of the same household. The net saving rate in such a household could then differ quite significantly from the saving rates of individual members.

We have defined the age, gender, labour market status and ethnicity of the household based on the reported characteristics of the head. However we also report results based on household shares (eg the share who are working, who are male, etc).

Table 1 reports the number of observations in each year together with the saving rates at the mean, median, 25^th and 75^th percentiles. The final column reports the ratio of the averages of saving to consumption, as distinct from the average (or quantile) of the ratios. The mean is clearly influenced by extreme values, so frequently it will be helpful to focus on the median saving rate.

Table 1 – Sample size and saving rates by survey year

	Total Sample	Savings Rate (S/X)
		Mean	25th Percentile	Median	75th Percentile
1984	3331	0.376	-0.058	0.227	0.613	0.189
1985	3295	0.287	-0.106	0.168	0.498	0.139
1986	3174	0.318	-0.078	0.201	0.551	0.177
1987	3210	0.341	-0.084	0.209	0.581	0.204
1988	4021	0.347	-0.043	0.212	0.563	0.200
1989	3142	0.358	-0.080	0.207	0.601	0.200
1990	3047	0.313	-0.110	0.188	0.560	0.166
1991	2674	0.340	-0.099	0.193	0.575	0.227
1992	2712	0.380	-0.062	0.217	0.609	0.251
1993	4244	0.415	-0.057	0.222	0.621	0.270
1994	2839	0.338	-0.096	0.197	0.546	0.235
1995	2695	0.336	-0.110	0.191	0.607	0.234
1996	2621	0.355	-0.111	0.186	0.572	0.246
1997	2642	0.359	-0.091	0.193	0.578	0.259
1998	2622	0.428	-0.081	0.238	0.676	0.320
Total	46269	0.353	-0.086	0.202	0.584	0.222

It should be stressed that there is enormous underlying variability in the data. Household savings vary between -$1.56 million and +$0.78m. Among any one group with the same rate of saving (e.g. 0.20 to 0.29) consumption expenditures vary from $2,950 to $125,474, and their absolute level of saving varies from $879 to $32,241. Nearly 32 percent of the sample report negative saving. Inevitably some of this may be due to under-reporting of income. Evidence from the USA suggests that when the differential under-reporting of both income and consumption is allowed for, up to one third of the apparent fall in the saving rate between 1972-73 and 1983-84 may be due to misreporting (Bosworth, Burtless and Sabelhaus (1991)).