Since 1979 the Australian Bureau of Statistics has conducted a series of special supplementary surveys based on the monthly labour force survey. Although the 1986 Income Distribution Survey was the ninth special supplementary survey, it was only the second such survey to collect detailed income information. It collected a wide range of information on working arrangements, incomes and housing costs.

The main users of data from the survey are the social policy departments of the Australian Government and researchers in academic and private research centres. The survey was funded by the Australian government.

A public use unit record file of selected data from the survey is available for purchase from the Australian Bureau of Statistics. This file has been used to generate the LIS data file. Users are restricted to using this file for statistical purposes only, and may not use it to identify individuals. In addition, certain copyright restrictions apply.

Further information on the survey generally, and the conditions of usage of the unit record file in particular, can be obtained from:

The primary field work for the survey was conducted over the period 15 September to 5 December 1986 with income information collected both for the current period (e.g. most recent payment), and for the 1985-86 financial year. A second phase of fieldwork took place in February and March 1987 when call-backs were made to self-employed respondents to collect finalised tax return information.

The demographic data in the LIS dataset refers to the September to December period, whilst the income data used is that compiled by Bruce Bradbury of the Social Policy Research Centre, University of New South Wales. The author is grateful for the assistance provided by Jenny Doyle of the SPRC in the calculation of Tables 18.2 and 18.3, and the general assistance of Bob Dutton of the Australian Bureau of Statistics, but retains responsibility for any errors for the 1985-86 year.

A bibliography of the main official publications from this survey is provided in Section N. Unpublished documentation associated with the sample file is also available, and may be obtained from the above address.

In general, the methodology for this survey was very similar to that of the 1981-82 Income and Housing Survey also available in the LIS database (AS81). Important differences are noted below.

B POPULATION AND SAMPLE SIZE, SAMPLING METHODS back

1. The Sample Design and Sampling Frames

The survey was conducted throughout Australia on a multi-stage area sample of private dwellings (houses, flats, etc.) and a list sample of non-private dwellings (hotels, motels, etc.), excluding military establishments, and covered about one sixth of one per cent of the population of Australia. Sampling probabilities were based on population estimates from the census and other data sources.

The sampling fraction varied across the different states and territories of Australia in order to permit useful estimates for the smaller states.

Information was obtained by trained interviewers in a personal interview conducted with each person aged 15 years and over in the selected dwelling.

The following persons were excluded from the scope of the survey:

1. certain diplomatic personnel of overseas governments customarily excluded from census and estimated populations,
2. overseas residents in Australia,
3. members of non-Australian defence forces (and their dependants) stationed in Australia.

The following persons were asked a very limited range of questions:

4. persons who migrated to Australia after 30 June 1986,
5. students in boarding schools, patients in hospitals and sanatoria, and inmates of goals, reformatories etc.

Families containing persons in any of these categories have also been excluded from the LIS database. In addition, households containing persons who were not dependent children (see section G for definition) but fell into any of the following categories, were excluded from the LIS database:

• Females who changed marital status after 30 June 1985;
• Persons aged 15-20 years who attended school full-time for part of the 1985-86 financial year (but at the time of the survey were not at school);
• Persons who migrated to Australia after 1 July 1985;
• Australias overseas for more than 12 weeks during 1985-86, and working for a non-Australian business;
• Australians overseas for all of 1985-86.

Note that the first two of these exclusions were not applied to the AS81 LIS file. In particular, the first exclusion removes many sole parents. The extent of difference between the two surveys can be judged from Table B.5 (in the INTRMETH.* file), where both the original and LIS database sample sizes are given.

C. MEASURES OF DATA QUALITY back

1. Response Rates

The survey was conducted under the Census and Statistics Act, which mandates participation. As a consequence, response rates are very high. The response rates calculated from the initial fieldwork phase are summarised in Table AU85.1.

Table AU85.1
Response Rates

Source: Australian Bureau of Statistics, PSO Procedures, Special Supplementary Survey 9, Response and Sample Loss Tables.

The overall response rate of fully responding dwellings was 94.1 per cent. No analysis has been undertaken of the characteristics of non responding persons and households.

2. Reporting and Under-Reporting and Income Data Quality

The survey data has been checked against population census, National Accounts, income tax, Department of Social Security and Department of Veterans' Affairs aggregate statistics. In addition, extensive checks of the internal consistency of the data were carried out.

A comparison of the survey income aggregates with those available from the Australian National Accounts is shown in Table AU85.2.3 Total household disposable income recorded by the survey comprises only 79 per cent of that recorded by the National Accounts, but most of this difference can be explained by the conceptual differences in the two data sources. Though such conceptual differences apply to all the items in this table, some key issues of data quality are identifiable.

Table AU85.2
Comparison with National Accounts Aggregates, 1985-86

Note.These calculations are based on the unit record file rather than the LIS database.
Source: 1986 Income Distribution Survey, Unit Record File, Australian National Accounts, National Income and Expenditure 1987-88, and SPRC calculations.

• income from unincorporated enterprises is apparently under-recorded in the survey. In part this reflects an explicit adjustment made in the National Accounts for expected under-recording of self-employment income in taxation statistics.
  
• Wage and salary incomes correspond very closely in the two sources. Note that this is different from the earlier survey (AS81) where some wage and salary income was included with self-employment income.
  
• Income from interest is very much under-recorded (though this is offset to a small extent by an over-estimate of dividend income).
  
• Only two thirds of personal benefits payments to residents are recorded in the survey. Much, but not all, of this difference stems from the inclusion of benefits such as medical and pharmaceutical benefits in the National Accounts measures.
  
• Personal income tax is slightly over-estimated. This may reflect the fact that the survey recorded income tax liability for the year, whereas the National Accounts estimates are for payments.
  

Further details comparing the survey estimates of selected aggregate income support benefits received during the year with Department of Social Security (DSS) expenditure statistics are shown in Table AU85.3.

Table AU85.3

Comparison with Department of Social Security Expenditure Aggregates

Note: * In the DSS data, expenditure on Wifes/Carers Pension is included with expenditure on Invalid and Age Pensions (depending upon which pension the spouse is receiving).

Sources: DSS Annual Report, 1985-86, DSS Ten Year Statistical Summary, 1976 to 1986, 1986 Income Distribution Survey, Unit Record File.

Whilst the exclusion of the institutionalised population from the scope of the survey makes these two sources of data not entirely compatible, some problems of under-recording are apparent. Income from Unemployment Benefits, for example, is apparently under-recorded by $939m, or 30 per cent. Similar rates of under-recording apply to supporting parents benefit and widows pension. (The exclusion of the institutionalised will mainly affect the comparison with age and invalid pension expenditure).

Similar rates of under-recording apply to the current data on numbers of persons in receipt of the different pensions or benefits. However it is not known whether the discrepancy is mainly a result of under-recording of income receipts, or under-sampling of the pensioner/beneficiary population.

3. Data Corrections and Adjustments

Where respondents to the survey could not supply precise details of annual income, income from the sources concerned was derived from other known sources (e.g pension rates) or was imputed using a "cold decking" methodology. That is, mean incomes of persons in the same sex/age/occupation/marital status/region cell were assigned.

In addition, very high incomes have been adjusted to ensure confidentiality of respondents. This has generally been done by selecting a number of persons with very high incomes (in a given demographic category) and recoding their incomes to their mean income (i.e. the mean income of the persons with high incomes).

4. General Assessment of Income Data Quality

The main strength of the survey is the very high response rate due to the compulsory nature of the survey. Fieldwork procedures were also very thorough. With the request for respondents to refer to records, one can expect this to result in income levels close to those recorded for taxation purposes.

None-the-less it appears that the survey suffers from under-recording of unearned income, as is typical of most other income surveys. It should be noted, for instance, that one reason for discrepancies between the National Accounts and survey estimates is that the National Accounts explicitly adjust for the under-reporting of income in taxation and similar data.

D. DATA COLLECTION AND ACQUISITION back

The data were collected via personal interviews with all persons aged 15 and over in the selected household. On the initial call the interviewer completed a household schedule, and then made arrangements (if necessary) to return to interview each person in the scope and coverage of the survey.

For each individual a detailed personal questionnaire was administered. Respondents were asked to refer to personal records such as taxation assessment or return forms, group certificates, pay slips, etc. where available. Persons with income from their own business who did not know their annual income were asked if the interviewers could call back when their records were available. Call-backs were made in February or March 1987. Around 80 per cent of respondents referred to records.

E. WEIGHTING PROCEDURES back

Each person record in the file has a weight assigned which takes account of probability of selection in the sample from their region, with an adjustment to account for under-enumeration at the age, sex and metropolitan/rest of state level. The weights were derived from the monthly population benchmarks showing the number of persons in each part-of-state, cross classified by age and sex. Because of a requirement to obtain stable estimates for the smaller states, probabilities of selection vary significantly from state to state. Hence the use of weights to obtain Australian estimates is essential.

Weights for higher units of aggregation than the individual (i.e. for income units, families or households) were calculated as the harmonic mean of the person weights.

F. DETERMINATION OF SURVEY UNIT MEMBERSHIP back

The basic criteria for persons to be included were that they satisfied the scope criteria of the survey (see Section B) and were usual residents of the dwelling. However, usual residents of the dwelling were excluded if they were away from the dwelling for the whole of the field-work period, whilst visitors were excluded if they stayed at their usual residence (i.e. elsewhere) for any part of the period. Visitors were included in the survey if they did not usually live in a private dwelling. Visitors who were included in the survey at selected dwellings were allocated separate dwelling numbers on the public use unit tape.

To determine family relationships, respondents were asked to nominate a person as head of the household, with relationships being recorded relative to that person. Within households, respondents were classified into families and "income units".

Income units can be one of the following types: single adults, sole parents with dependent children, married couples or married couples with dependent children. Couples in "de-facto" (cohabiting) relationships are coded identically to those who are legally married.

Families comprise persons related by blood, marriage (legal or de-facto) or adoption with the proviso that there can only be one couple or sole parent per family. Thus a household comprising a married couple and their sole parent daughter is coded as two families, as is a household comprising a husband and wife together with the wife's mother and her husband.

The family head on the unit record file, however, is not necessarily the person nominated as household head by the initial respondent.

The family head is defined as the head of the primary income unit of the family. If this income unit is a couple, the husband is the head. The primary income unit is defined according to a pre-defined hierarchy, with married couples, for example, coming before individuals. In the LIS database, information at both the family and household level can be accessed (see variable D5), as well as at the person level.

G. CHILDREN AND SPOUSES back

Dependent children are defined as persons aged under 15 years, or aged 15 to 20 years and a full-time student, who has a parent/guardian in the income unit and is neither a spouse nor parent of anyone in the income unit. Since all persons aged 15 and over are included in the unit record file, alternative definitions of children are possible. This information is used to calculate the LIS variable D27.

As noted above, couples in de-facto (cohabiting) relationships were coded as if married. De-facto relationships were defined as existing where a married couple live together in a married situation, but are not legally married. For example, responses such as "fiancee" and "living with my girlfriend/ boyfriend" are coded as de-facto.

H. AVAILABILITY OF BASIC SOCIAL AND DEMOGRAPHIC INFORMATION back

A wide range of social and demographic information is available from the survey, and most of this is also made available on the unit record file. Some variables, however, have had categories collapsed in order to protect the confidentiality of respondents. A summary of available variables is shown in Table AU85.4. These variables are available for all persons interviewed. Information on dependents aged under 15, however, is much more limited, consisting only of the number of dependents in different age ranges, and the age of the youngest child.

Table AU85.4
Basic Social and Demographic Information (available on public use unit record file)

I. AVAILABILITY OF LABOUR MARKET INFORMATION back

A wide range of labour force information was collected for all persons included on the public use unit tape. These variables relate both to the persons' status at the time of interview, and their experience during the 1985-86 financial year. Table AU85.5 summarises the information available from the public use unit record file.

Table AU85.5
Labour Market Information

J. AVAILABILITY OF GEOGRAPHIC INFORMATION back

State of residence is identified (The Australian Capital Territory and Northern Territory are combined).

K. SOURCES AND AMOUNTS OF CASH INCOME back

Both current and 1985-86 financial year incomes were recorded from a wide range of cash income sources. This information was collected from all respondents except for persons overseas for the whole of 1985-86. The incomes recorded in the LIS database are for the 1985-86 fiscal year. Negative incomes from own businesses and partnerships were recorded.

L. TAXES back

Personal income tax liability for the 1985-86 year was asked in the survey. However despite the call back procedure this information was unavailable for around 7.5 per cent of respondents. In the LIS database, income tax has been imputed for these cases.

In addition the survey recorded some information on payments for local government and utility taxes (as part of the housing cost component of the survey).

M. BIBLIOGRAPHY OF MAIN PUBLICATIONS BASED ON THE CSP PANEL back

1986 Income Distribution Survey Australia, Preliminary - Results Catalogue No. 6545.0

1986 Income Distribution Survey Australia, Income Units Catalogue No. 6523.0.

1986 Income Distribution Survey Australia, Persons with Earned Income Catalogue No. 6546.0.