2.1 Survey of Income and Housing, 2019-20

All SIH 2019-20 admin income estimates that are sourced from PS, ITR and ATO Superannuation data refer to 2018-19 financial year and are adjusted to 2019-20 dollars using changes in the Consumer Price Index. This adjustment reflects changes in purchasing power and living standards between these periods. Data from 2018-19 was used as the only alternative, waiting for more recent data (available approximately 18 months after the reference period) limits the timeliness of results produced. Previous financial year data, likely to be the data available in future applications, is used to demonstrate the quality attributes of timelier results.

DOMINO and STP items reference data from the 2019-20 financial year, where available.

Annual employee income selects the greater of PS_EMP_ANN or ITR_EMP_ANN.

Current employee income derives income for current government pension and allowance recipients (based on DOM_GOV_CUR>0) differently to non-recipients (DOM_GOV_CUR=0). This is due to the underlying variability in employee income that is typical for government pension and allowance recipients.

For recipients of a government pension or allowance in 2020, the derivation preferences STP_EMP_CUR_1M. If not available, the next preferences (in order of decreasing preference) are:  STP_EMP_CUR_3M, STP_EMP_CUR_6M, and the maximum of either PS_EMP_CUR or ITR_EMP_CUR. 

As STP data was unavailable in 2019, DOM_EMP_CUR was potentially used for current employee income for people participating in SIH during 2019. If this was not available, then the preference was for selecting PS_EMP_CUR over ITR_EMP_CUR. (Note: it was not possible to subject these estimates to job or employment checks using STP data.) To reduce instances of income being incorrectly assigned to people with a 2019 reference date, PS and ITR amounts were excluded from a person’s calculation where: 

• current superannuation pensions and annuity income (SAE_SUP_CUR or ITR_SUP_CUR) were present.
• they were aged less than 60 years receiving current income from government paid parental leave pay (ITR only).
• They were receiving the Age Pension (ITR only).

A residual number of records with values not available for DOM_EMP_CUR, PS_EMP_CUR or ITR_EMP_CUR, but who had reported in SIH that they were an employee, had an admin employee income imputed using hot-deck donor imputation. This assigned these records employee income information from a similar person based on age, sex and state or territory of usual residence and section of state. Additionally, where possible, the donors record would have the same marital status, family type, relationship in household, interview quarter and ranged survey weight as the record requiring imputation).

People who did not receive a current government pension or allowance received the maximum of STP_EMP_CUR_1M, _3M or _6M variants, PS_EMP_CUR or ITR_EMP_CUR. Records with a 2019 reference date used PS_EMP_CUR and STP_EMP_CUR without STP employment checks.

Annual own unincorporated business income is ITR_UBUS_ANN. Current own unincorporated business income is ITR_UBUS_CUR.

Annual government pensions and allowances use DOM_GOV_ANN if available, otherwise ITR_GOV_ANN is used. Similarly, current government pensions and allowances use DOM_GOV_CUR if available, then ITR_GOV_CUR.

To improve comparability with SIH estimates of income, amounts of coronavirus payments were modelled for eligible people and assigned to their current and annual income in the same manner as SIH.

Annual investment income is ITR_INV_ANN. Current investment income is ITR_INV_CUR.

Annual superannuation pensions and annuities income uses ITR_SUP_ANN if available, otherwise SAE_SUP_ANN is used. Similarly, current superannuation uses ITR_SUP_CUR if available, then SAE_SUP_CUR.

Annual other regular transfers income is ITR_OTH_ANN. Current other regular transfers income is ITR_OTH_CUR.

For each person, their estimate of total annual income is calculated by summing all annual estimates for the admin income type. Similarly, a person’s total current income is the sum of all current income estimates across income types.

The admin items selected in this section are summarised in the table below in preferential order (unless otherwise specified, see employee income).

For all 2,934 ’unlinked’ people missing income information (due to lack of linkage to the PLIDA spine), it was deemed appropriate to impute their admin income values. This supports the validity of comparing admin income statistics with those derived from reported incomes in SIH. 

Imputation in this setting requires an understanding of which variables contribute to the state of being unlinked. A list of potentially influential characteristics was nominated:  household and family relationship, age, sex, location, level of income as reported in SIH, and associated survey weight.  For each member of the group requiring imputation, these characteristics were compared with those of people with PLIDA linkage. A summary of comparisons is presented in the Appendix.

Based on these results, it was considered that a standard ABS imputation method, 'Hot-deck imputation', may be appropriate. This method is used to impute data for an individual from a similar ('donor') record that has complete information. A potential donor record is drawn from a pool of eligible donors and deemed eligible for use if it shares certain key characteristics with the record requiring imputation.

The implementation of Hot-deck imputation was informed by results from a logistic regression intending to recognise those covariates which were significant in determining whether a person had PLIDA spine ID. Regression results suggested key characteristics for donor selection. Further criteria were also used to guide appropriate donor selection.  An exact match was required for:

• Age (grouped in 10-year age groups from 15 to 44 years, and then from 45 to 64 years, and 65 years and over).
• Family composition of household.
• Relationship in household.
• Sex of person.

Additionally, donors were selected to match on as many of the following characteristics as possible:

• Marital status.
• Survey weight (ranged).
• Quarter of SIH interview.
• State or territory of usual residence and section of state.

The imputation process was mostly able to find suitable donors. A small remainder of records that could not be associated with suitable donors received mean imputation from similar donors based on age, family composition, or relationship in household. A comparison of the reported an imputed incomes for unlinked and linked people is available in the Appendix.

Household income values are prepared by summing total income estimates (either derived from data or as imputed as per 2.1.2) using SIH household structures.

The dataset is reviewed and edited to resolve any detected inconsistencies between income type estimates. Towards this, DOMINO and STP data are used to help ensure coherence across the income profiles of individuals and their household members. For instance, all non-government incomes for an individual and their household members are reviewed alongside government pension and allowance incomes. This includes assessing against the maximum eligible amount for at which a person becomes ineligible for certain payments.

Statistics are weighted for output. The results from this study use the same sample and weights that are calculated for SIH 2019-20 population estimates. 

More information on weights is available in the Survey of Income and Housing User Guide. The user guide also provides information to assist the interpretation of weighted income measures presented in the results.