5204.0.55.011 - Australian National Accounts: Distribution of Household Income, Consumption and Wealth, 2003-04 to 2014-15  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 26/11/2015   
   Page tools: Print Print Page Print all pages in this productPrint All

CHAPTER 4 - METHODOLOGY

This chapter presents a summary of the methodology used to create individual data points and some improvements to this methodology; a summary of the steps undertaken to enable the presentation of the data points in a time series, including a discussion on demographic shifts; and the strengths and weaknesses of the data presented in the publication.


OVERVIEW OF METHODOLOGY FOR INDIVIDUAL DATA POINT

The methodology used to compile the individual data points in this publication is based on the methodology described in the Information Paper: Australian National Accounts, Distribution of Household Income, Consumption and Wealth, 2009-10 (cat. no. 5204.0.55.009). A summary of the methodology is as follows:

Current price household estimates for income, consumption and wealth from the Australian System of National Accounts, 2014-15 (cat. no. 5204.0), for biennial years 2003-04 to 2013-14 were distributed for five household distributional indicators based on data from the ABS Survey of Income and Housing (biennial years starting from 2003-04 to 2013-14) and ABS Household Expenditure Survey (2003-04 and 2009-10).

Estimates for non-profit institutions serving households (NPISH) included in the household sector in the cat. no. 5204.0 were removed from the household national accounts in this release.

The household national accounts estimates for a particular year (macro) and the corresponding ABS household survey estimates (micro) were compared and coverage ratios (micro/macro) calculated. For some items, the macro and/or micro estimates were adjusted to derive the most relevant common scope for comparison. The corresponding micro household items were sub sectored into the following household groups: main source of income; equivalised income quintiles; household composition; age of household reference person; and equivalised net worth quintiles.

The Australian System of National Accounts (ASNA) household components and aggregates were distributed to the five household groups:

  • directly using the distribution of the equivalent micro component when the coverage ratio was considered adequate, for example, social assistance benefits;
  • indirectly by a related micro distribution when there was no direct micro distribution information for the national accounts item, for example the national accounts item non-life insurance claims were distributed using the micro distribution for total insurance premiums paid;
  • indirectly by creating a micro distribution ('synthesised') based on related micro distribution, for example, synthesised micro distribution was created for the national accounts item financial intermediation services indirectly measured (FISIM) for consumer loans; and
  • by the corresponding aggregate distribution for income (disposable income), consumption (final consumption expenditure), assets (total assets) and liabilities (total liabilities), when micro distributions either directly or indirectly are not available. For these national accounts items, the inclusion or exclusion of the item did not impact on the distribution of the national accounts aggregates.

The very remote communities (VRC) and people living in non-private dwellings (NPD) populations that were out of scope of the micro surveys were excluded from the ASNA estimates and distributed separately using data from the 2006 and 2011 ABS Census of Population and Housing. These distributions were then added to the ASNA distributions based on the micro surveys to obtain the final distribution of the ASNA household income, consumption and wealth estimates.

For detail information regarding the methodology described above, please refer to Information Paper: Australian National Accounts, Distribution of Household Income, Consumption and Wealth, 2009-10 (cat. no. 5204.0.55.009), Chapter 4 - "Data Sources and Methodology: Distribution Methodology for Household National Accounts Estimates".

The methodology to construct data point for 2014-15 is discussed below in the section, Time Series Presentation.


IMPROVEMENTS TO THE ORIGINAL METHODOLOGY

The following improvements were made to the original methodology described above:

Owner Occupied Imputed Rent

In the ABS Survey of Income and Housing (SIH) for 2013-14, a new experimental methodology for estimating gross imputed rent for owner occupied dwellings was implemented. The new methodology makes use of additional administrative data sources on the value of dwellings to improve these estimates, in particular for higher value dwellings. Estimates using the new methodology have been recompiled for each SIH from 2003-04. A detailed explanation of the new experimental methodology is expected to be published in December 2015 in Experimental Estimates of Imputed Rent, Australia, 2013-14 (cat. no. 6525.0).

The distribution of the national account items, Gross operating surplus (GOS) - dwelling owned by person and the household consumption item, imputed rent on owner occupiers were impacted back to 2003-04 by the new SIH methodology.

Distributional indicators for Households in Very Remote Communities and Non-Private Dwellings
  • Improved methodology from 2003-04 onwards for the distribution of income components for gross mixed income (GMI) and compensation of employees (COE) by household indicator, Main Source of Income (MSI). Previous methodology was based on demographic information on Census data only, the new methodology used SIH, ABS Household Expenditure Survey (HES) and the Census data to obtain the household distributional indicator for MSI.
  • Improved methodology from 2003-04 onwards for the distribution of income component, other current transfers, by household indicator, MSI. New methodology based on improved alignment between Government Finance Statistics and National Accounts estimates for other current transfers.
  • Improved methodology from 2003-04 onwards for the distribution of the memorandum item, superannuation benefits received, by household indicator, Age of reference person. New methodology based on an improved articulation of the SIH, which resulted in an improved distribution for the Age of reference person indicator.

Updated Data for Non-Profit Institutions Serving Households (NPISH)

All NPISH estimates removed from the household sector of the ASNA were revised from 2003-04 onwards due to the source data for these estimates, the Non-Profit Institution (NPI) Satellite Account (cat. no. 5256.0) for 2006-07 and 2012-13 being revised in August 2015. The NPI satellite accounts were revised due to the inclusion for the first time of volunteering services, sourced from unpublished results from the ABS General Social Survey, 2014 and due to other minor conceptual and methodological improvements.

For information regarding the improvements to the original methodology implemented in the 2014 release, please see Australian National Accounts, Distribution of Household Income, Consumption and Wealth, 2003-04 to 2011-12 (cat. no. 5204.0.55.011), see Methodology - Improvements to the Original Methodology.


TIME SERIES PRESENTATION

Periodicity

The time series presented in this release for distribution of the household income, consumption and wealth is biennial from 2003-04 to 2013-14 and for the latest year 2014 -15. The decision to start the time series in 2003-04, and compile the estimates biennially up to 2013-14 was based on:
  • the availability of income and wealth modules from the SIH for all six data points from 2003-04 for income, and 5 data points (excluding 2007-08) for wealth;
  • the availability of the improved income data collected through the SIH starting from 2003-04;
  • distributed data points being a minimum of two years and a maximum of three years from the 2006 and 2011 ABS Census;
  • distributed data points being a minimum of two years and a maximum of four years from the 2003-04 and 2009-10 HES; and
  • availability of the micro Social Transfer in Kind (STiK) data for four data points from 2003-04 (excluding 2005-06 and 2007-08).

The ABS received positive feedback from users regarding the analytical usefulness of the data presented in the 2014 release of Australian National Accounts, Distribution of Household Income, Consumption and Wealth, 2003-04 to 2011-12 (cat. no. 5204.0.55.011). However, many users noted that more timely household distributional data would be more useful to understand the current financial health of groups within the household sector, for example an assessment of debt levels of different households groups in the current low interest environment. The ABS responded to this user feedback and has included in this release extrapolated estimates for household distributional data for income, consumption and wealth for 2014-15. All distributional data points in the publication are benchmarked to the most recent issue (released 30 October 2015) of the Australian System of National Accounts, 2014-15 (cat. no 5204.0) and as a result the distributional data set for the 2014-15 data point is available approximately 5 months after the end of the reference year.

Models Used to Estimate Data Gaps in Micro Distributional Data

For this release, a review was undertaken on the linear interpolation (extrapolation) option chosen in the 2014 release to model the micro distributional household indicators for the years that micro (SIH, HES, Census and STiK) data was not available. For more information on why linear interpolation (extrapolation) was chosen in the 2014 release, please see, Australian National Accounts, Distribution of Household Income, Consumption and Wealth, 2003-04 to 2011-12 (cat. no. 5204.0.55.011), Methodology - Implementation of the Time Series.

The two options of linear interpolation (extrapolation) were considered for this release:
    (i) two and/or three data point linear trend interpolation (extrapolation); and
    (ii) three data point average growth rate of the total household micro estimate, with the average growth applied to the average share over three data points of the household indicator (e.g. quintiles).

Both option (i) and (ii) are simple methodologies, and work effectively if the historical data is stable over time and in particular if there are not large fluctuations between shares of the various indicators over time.

Option (i) was initially applied to model for the micro distributional household indicators for the missing years that the source micro (SIH, HES, Census and STiK) data was not available. In general this methodology smoothed out outliers and reduced the incidence of negative values. However, the extrapolated distributional data for 2013-14, used in our feasibility study (see Chapter 5) produced some negative values for some household indicators for income, wealth and STiK, an indication that for some items the distribution over time is not linear. A number of solutions to control for this was considered, such as when the observed values were close to zero, simply forcing it to zero, but doing so resulted in the household groupings not adding up to the total household estimate for a given item. Using option (ii) eliminated the negative vales and also ensured additivity across the household groupings.

Models Used for Current release

Option (i) was applied to interpolate (extrapolate) the micro distributional indicators for consumption for the years 2005-06, 2007-08, 2011-12, 2013-14 and 2014-15 from HES. In general, it is assumed that consumption patterns of household tend to be fairly stable over time, and this method has the advantage of capturing the systematic pattern (i.e. movements up or down) of distributional information over time. However, users need to be aware that a limitation of this method is that the further away from the actual survey data the extrapolated data is the more unstable the estimates are, as a result the 2011-12 estimates are expected to be more stable than 2013-14 estimate and 2014-15 estimates to be the least stable of the three estimates.

Option (ii) was applied to SIH to interpolate (extrapolate) the micro distributional indicators for income in 20014-15; wealth for the years 2007-08 and 2014-15; and STiK for the years 2005-06, 2007-08 and 2014-15. This method has the advantage of not generating negative vales, works adequately for when the distributed pattern is not linear over time and assumes that the shares of the distributional indicators are stable over time. The assumption regarding the stability of the shares is more than likely to be valid as the availability of the actual survey data used in the method is between 1-2 years from the interpolated (extrapolated) data point.

Option (i) was applied to interpolate (extrapolate) Census data that is used to construct the distributional data for the VRC and NPD populations. The total disposable income, total consumption expenditure and total net worth estimates for the NPD and VRC populations were approximately 1% of the total Australian aggregates of income, consumption and net worth for all data points. The impact of the methodology chosen to interpolate (extrapolate) would be minimal on the final Australia distributional estimates results and therefore a decision was made to keep the methodology the same as that applied in the 2014 release.

Feasibility Study: Income and Wealth estimates, 2013-14

To determine the robustness of the methodology used to extrapolate the 2014-15 estimates of the distributional data, a feasibility study was undertaken by comparing 2013-14 distributional data for income, wealth and STiK using actual SIH data from 2013-14 versus modelled micro distributional data using option (ii). Due to the lack of timely HES data, the robustness of the modelled consumption estimates was not tested.

A significant amount of analysis may be undertaken to test the robustness of the modelled estimates. For this study an analysis was undertaken on how the modelled data performed for each of the five household indicators (main source of income; equivalised income quintiles; household composition; age of household reference person; and equivalised net worth quintiles) for estimating:
    (i) total share of income and wealth components and aggregates; and
    (ii) growth rates of the levels of income and wealth components and aggregates. In addition to providing insight into the accuracy of growth rates of items between time periods, the analysis provides insight into the accuracy of the levels reported for each item and household group. When analysing the impact of the growth rates, the contribution of the individual household group to the estimate needs to be taken into account.

For details of the study, please refer to Chapter 4 "Feasibility Study: Income and Wealth estimates, 2013-14". A summary of the findings are provided below.

The feasibility study for the distribution of the five household indicators showed that for the aggregate adjusted disposable income (ADI) and net worth, there was good alignment between the SIH and modelled estimates, but for the components of ADI and net worth, the alignment was less than the aggregates. The study produced the following results:
  • modelled estimates for the share of aggregate ADI aligned closely to the SIH results in terms of the pattern of the distribution and size of the shares for household indicators;
  • modelled estimates of the shares of COE, property income receivable, social assistance benefits, interest payable and STIK aligned closely to the SIH results (pattern and size), whereas GOS-dwelling, GMI and taxes payable, the alignment was not as close, and for some household indicators the difference was significant compared to the SIH results;
  • modelled estimates for the share of aggregate net worth aligned closely to the SIH results in terms of the pattern of the distribution and size of the shares for all household indicators with the exception of main source of income;
  • modelled estimates for the share of residential dwelling and land, total financial assets, currency and deposits, insurance technical reserves and loan liabilities aligned closely to the SIH results in terms of the pattern of the distribution and size for all household indicators. However, alignment for the financial asset - shares and other equity was not as close for all household indicators;
  • modelled estimates for the growth rate of aggregate ADI and net worth, the alignment to the SIH results were mixed. However users are reminded to take the weights of the individual household groups into account when assessing the alignment of the household indicators; and
  • modelled estimates for the growth rate of the components of ADI and net worth, the alignment to the SIH results were mixed. However users are reminded to take the weights of the individual household groups into account when assessing the alignment of the household indicators.

The feasibility study provides users some insights into assessing the robustness of the modelled distributed income and wealth estimates in this release. In particular, it provides users some insights into the modelled estimates for the latest 2014-15 year. Users should bear in mind when making an assessment, that the latest available SIH results used in the model for 2014-15 were from 2013-14, a gap of one year. It is expected that the modelled estimates for 2014-15 would be more accurate due to the one year gap as opposed to the data in the feasibility study where the modelled estimates were two years away from the latest SIH results.


CHANGING DEMOGRAPHICS OVER TIME

When distributional data across different years is compared, it is important to note that the change in the estimate is impacted by (a) changes in the household's income (consumption and wealth) and (b) change due to more households in the distributional group.

In order to separate changes due to (a) and (b), the ABS applied a number of different methods to control for the demographic shift, that is item (b). For detail information regarding the different methods applied, please see, Australian National Accounts, Distribution of Household Income, Consumption and Wealth, 2003-04 to 2011-12 (cat. no. 5204.0.55.011), Methodology - Changing Demographics Over Time. In this release, dollars per household were applied to account for demographic shifts. It should be noted that the ideal methodology to capture demographic shifts accurately is to undertake longitudinal (panel) surveys such as the Melbourne Institutes 'Household, Income and Labour Dynamics in Australia ' (HILDA). However, it should be noted these types of surveys would be extremely expensive to undertake with the sample size required to produce robust detail distributional information such as what is produced by the current ABS micro surveys.

To assist users in understanding the size of the demographic shift, the tables below provide the proportion of households in each household distributional indicator.

Table 4.1: Proportion of households, by household indicator, 2003-04 to 2014-15

Income
2003-04
2005-06
2007-08
2009-10
2011-12
2013-14
2014-15

Total Number of Households
7 954 585
8 160 856
8 327 818
8 664 856
8 912 566
9 150 897
9 395 790

Main Source of Income (%)
Wages and salaries
56.0
57.6
60.1
59.5
59.8
59.8
59.8
Income from Unincorporated Business
8.1
8.0
8.2
7.9
7.3
7.3
7.5
Property income & Superannuation
8.6
8.8
8.4
7.9
8.5
8.5
8.3
Government pensions & allowances
26.0
24.3
22.0
23.7
23.3
23.2
23.4
Other
1.3
1.3
1.3
1.1
1.1
1.1
1.1
Household Composition (%)
Lone person under 65
16.4
16.6
16.1
16.1
15.7
15.7
15.8
Lone person 65 and over
10.7
10.8
10.6
10.3
10.8
10.7
10.5
One parent with dependent children
6.7
6.7
6.0
6.2
5.7
5.7
5.9
Couple only, reference person under 65
17.2
16.8
17.2
16.9
16.2
16.2
16.4
Couple only, reference person 65 and over
8.3
8.3
8.6
8.6
9.0
9.0
8.8
Two adults or more with dependent children
27.2
26.4
26.8
26.8
26.8
26.8
26.8
Other
13.6
14.4
14.7
15.1
15.9
15.9
15.7
Age of Reference Person (years) (%)
15-24
4.5
5.0
4.6
4.2
4.1
4.1
4.1
25-34
17.8
17.2
16.3
16.4
16.0
16.0
16.2
35-44
21.7
21.1
20.1
20.2
19.5
19.5
19.7
45-54
20.2
19.9
20.2
19.9
19.9
19.9
19.9
55-64
14.7
15.6
16.4
17.1
17.5
17.5
17.4
65+
21.1
21.2
22.3
22.2
23.0
23.0
22.7
Equivalised Disposable Income Quintiles (%)
Lowest
20.9
20.9
20.9
20.9
21.0
21.0
21.0
Second
20.2
20.2
20.1
20.1
20.0
20.0
20.0
Third
19.7
19.7
19.7
19.7
19.7
19.7
19.7
Fourth
19.6
19.6
19.6
19.6
19.6
19.6
19.6
Highest
19.6
19.6
19.7
19.7
19.7
19.7
19.7
Equivalised Disposable Net Worth Quintiles(%)
Lowest
20.9
20.9
20.9
20.9
21.0
21.0
21.0
Second
20.2
20.2
20.1
20.1
20.0
20.0
20.0
Third
19.7
19.7
19.7
19.7
19.7
19.7
19.7
Fourth
19.6
19.6
19.6
19.6
19.6
19.6
19.6
Highest
19.6
19.6
19.7
19.7
19.7
19.7
19.7


The number of households in the equivalised income and net worth quintiles when distributed to the in scope micro survey population was 20% of households for each quintile. However, when the VRC and NPD populations (out of scope for the micro surveys) are included, the proportion of households in each quintile marginally moves away from 20% as illustrated in the table above.


TIME SERIES ANALYSIS INCLUDED IN THE RELEASE

The release includes the following tables to:
  • analyse how household distributional groups have contributed to total growth of income, consumption and wealth - electronic table 9. This table includes demographic shifts such as the increase in the number of households in a particular household group, as they are an important driver to total growth;
  • remove demographic shifts, the method used in this release is dollars per household - electronic table 3 and 4;
  • analyse household groups per household growth in income, consumption and wealth - electronic table 10;
  • analyse contributions by component (income, consumption and net worth) to a household group's growth per household of gross disposable income, house final consumption expenditure, net worth and actual final consumption - electronic table 5 to 8; and
  • analyse the impacts of redistribution policies such as income tax, social assistance benefits and social transfers in kind; and the effectiveness of the policies over time- electronic table 11.


STRENGTHS AND WEAKNESSES OF TIME SERIES

To enable users to interpret the data of the time series, the strengths and weaknesses of the time series are presented below.

Strengths

The time series of the distributed household income, consumption and wealth data:
  • is benchmarked to the aggregates published in Australian System of National Accounts, 2014-2015, (cat. no. 5204.0), which enables users to interpret household distributional data within the broader context of published estimates on the Australian economy such as gross domestic product (GDP);
  • includes household distributional data for 2014-15, available five months from the end of the reference year, and therefore is the most up to date distributional data on households publicly available;
  • is complete and consistent for the years presented, where data was missing linear interpolation and extrapolation modelling techniques have been implemented;
  • is based on and expands upon work undertaken in an Organisation for Economic Cooperation and Development (OECD) and Eurostat (European Union statistical commission) expert group for measuring disparities in a national accounts framework. As a result, the data set produced for this publication should be comparable with any time series analysis on the distribution of household income, consumption and wealth in a national accounts framework performed by members of this expert group in the future; and
  • may be extended with future data points and revised with new source data, enabling a more accurate and longer time series. The time series is based on robust methodology formulated through an international expert group, with the availability of new source data from ABS micro surveys (Survey of Income and Housing and Household Expenditure Survey) and revised aggregate data from the ASNA , the data presented in this release may be easily revised and updated with future data points.

Weakness

A major weakness of this data set are the relatively long intervals between collection years for the Household Expenditure Survey (HES) and, to a lesser extent, the Census of Population and Housing (Census). As a result:
  • due to the availability of only two time periods for the HES (2003-04 and 2009-10) and the unavailability of proxies for the time periods in between, the missing source data had to be interpolated or extrapolated. While linear movements were assumed for the purpose of interpolation and extrapolation, care should be taken when looking at the distributions of household consumption items for 2005-06, 2007-08, 2011-12, 2013-14 and 2014-15, as this assumption may not be accurate. An additional limitation on extrapolation is the estimates are based on past observations, increasing the possibility of error caused by the actual movement deviating from the past movement; and
  • due to the unavailability of data on the distribution of household income, consumption and wealth items for populations living in Very Remote Communities and in Non-Private Dwellings, estimates had to be made for these items using ABS Census data. However, as this issue only affected about 2% of households the impact of the missing source data is assumed to be very small.

The unavailability of the 2007-08 (wealth only) and 2014-15 SIH data, the missing micro data points are modelled. Please see the section above on "Time Series Presentation" for the model used and Chapter 5 - "Feasibility Study: Income and Wealth, 2013-14", for an assessment of this model. Similar to the model used for the missing years of HES data, the models to estimate the SIH data are based on past observations, increasing the possibility of error caused by the actual movement deviating from the past movement.

Due to the inclusion of the VRC and NPD populations into the distributed estimates from the surveyed population, the number of households in each of the income and net worth quintiles does not equal 20%. After the addition of the distributed estimates for VRC and NPD, the quintile proportion ranges from 19.6% to 21.0%. Please see Table 1 above for more detail.

It could be argued that the time series (electronic Table 1, 2, 9 and 11) represents static points of cross sectional data and therefore falls short of a strictly defined time series such as a time series from a longitudinal study (e.g. Melbourne Institute HILDA survey). That is, the data in these electronic tables do not account for demographic shifts in the time series which a longitudinal study would capture. Please refer to section above on "Changing Demographic Over Time " for the method used to capture demographic shifts in this release.

Finally, users are encouraged to read Chapter 4, "Methodological Issues", in the Information Paper: Australian National Accounts, Distribution of Household Income, Consumption and Wealth, 2009-10 (cat. no. 5204.0.55.009), to understand the issues in constructing a household distributional data set for a single time point.


REVISIONS

Revisions of estimates from the previous release are due to:
  • the implementation of revised national accounts aggregates from Australian System of National Accounts, 2014-2015, (cat. no. 5204.0);
  • the implementation of revised national accounts data from Australian National Accounts: Finance and Wealth, June Quarter 2015, used to construct detail components for the national accounts net worth aggregates;
  • the implementation of the revised estimates for NPISH from the revised Non-Profit Institution (NPI) Satellite Account (cat. no. 5256.0) for 2006-07 and 2012-13;
  • improved distributional estimates for VRC and NPD population; and
  • new micro distributional estimates for imputed rent for owner occupied dwellings.