RDP 2003-07: Housing Wealth, Stock Market Wealth and Consumption: A Panel Analysis for Australia Appendix A: Data

The data set used in this paper consists of a panel of observations on seven variables. These variables are consumption, labour income, stock market wealth, dwelling wealth, other financial wealth and households housing debt and personal debt. The cross-section spans five Australian states: New South Wales (NSW), Victoria (VIC), Queensland (QLD), South Australia (SA) and Western Australia (WA)[25]. We also have a similar data set for the nationwide aggregate. All series are quarterly observations from 1984:Q4 to 2001 :Q4.

Consumption

Description: Household consumption expenditure at current prices in A$ millions, seasonally adjusted, from 1984:Q4 to 2001 :Q4. Available both by state and nationwide.

Source: ABS Cat No 5206.0

Income

Description: Household after tax labour income at current prices in A$ millions from 1984:Q4 to 2001 :Q4.

Sources: ABS Cat No 6302.0; ABS Cat No 6203.0; ABS Cat No 5206.0

Construction: We construct state-level and national labour income following Tan and Voss (2003). State i's after-tax labour income Yi is defined as:

Wages i are defined as wages i = (Inline Equation · awe i · employment i · scale i) where awe i are average weekly earnings, employment i is the number of wage and salary earners, scale i is a scaling factor that adjusts for self-employed workers (calculated as the ratio of hours worked by employed persons to hours worked by wage and salary earners)). Multiplication by Inline Equation converts average weekly earnings to a quarterly frequency. The data were obtained from ABS Cat No 6302.0 and ABS Cat No 6203.0 and are available quarterly from 1984:Q4.

Transfers i are calculated as total secondary income less social contributions for workers compensation for each state. Both series are available annually from 1989/90 onward in ABS Cat No 5206.0. We used aggregate-level data from the same source to backcast these series to 1984 and to infer a quarterly profile for them.

γi is the share of labour income in total household income, calculated as Wages i divided by total primary incomei. Total primary incomei is available annually from 1989/90 onward in ABS Cat No 5206.0. We used aggregate-level data from the same source to backcast this series to 1984 and to infer a quarterly profile.

Taxi is calculated as the sum of income tax payable and other current taxes on income, wealth, etc for each state. Both series are available annually from 1989/90 onward in ABS Cat No 5206.0. We used aggregate-level data from the same source to backcast these series to 1984 and to infer a quarterly profile for them.

Stock market wealth

Description: Household gross financial wealth invested in assets whose value varies directly with the stock market, in A$ millions, from 1984:Q4 to2001:Q4.

Sources: ABS Financial Accounts Cat No 5232.0; RBA Occasional Paper No 8.

Construction: The above sources provide data on a number of subcomponents of financial wealth. We group these into the following categories: currency and deposits, direct equity holdings, equity superannuation, non-equity superannuation, and other.

We use quarterly data from 1988:Q4 onward from ABS Cat No 5232.0 and annual data from RBA Occasional Paper No 8 before then. The quarterly profile before 1988:Q4 was inferred using the S&P/ASX 200 index of share prices for direct equity holdings and using a linear trend for the other categories of financial wealth.[26]

We used information on financial flows from the 1993/94 Household Expenditure Survey (HES) to distribute these aggregate financial wealth stocks across the five states[27]. Investment income was used for direct equity holdings and other, superannuation income was used for both superannuation wealth categories, and financial account interest income was used for currency and deposits. The process is described by the following equation where Inline Equation is the proportion of the total wealth in asset category c attributable to state i :

where Inline Equation is income flow from assets in category c to household n in state i, N i is the number of households from state i surveyed[28] and H i is the total number of households in state i.

We used two classifications of stock market wealth in our estimation. The first consists of the sum of direct equity holdings and equity superannuation. The second consists of direct equity holdings only.

Other financial wealth

Description: Household gross financial wealth not captured by the stock market wealth measure, in A$ millions, from 1984:Q4 to 2001 :Q4.

Sources: ABS Financial Accounts Cat No 5232.0; RBA.

Construction: We use two classifications of other financial wealth each of which corresponds to one of the above measures of stock market wealth. The first consists of the sum of non-equity superannuation, currency and deposits, and other. The second consists of the sum of equity superannuation, non-equity superannuation, deposits and currency and other.

Debt

Description: Total household debt in A$ millions, from 1984:Q4 to 2001 :Q4.

Source: ABS Cat No 5232.0.40.001; RBA.

Construction: The aggregate debt series was distributed amongst the states using data on housing loan repayments and personal loan repayments from the 1994/95 HES. The methodology is equivalent to that used to distribute aggregate stock market wealth amongst the states.

Housing wealth

Description: Household gross non-financial wealth invested in dwellings, in A$ millions, from 1984:Q4 to 2001 :Q4.

Sources: ABS Census of Population and Housing; ABS Cat No 8752.0; ABS

Cat No 3101.0; CBA/HIA dwelling price data.

Construction: The stock of dwellings by state is available from the census, for census years. This is converted into a quarterly series by linking the data for the census years using quarterly dwelling completions data from ABS Cat No 8752.0. Finally it is disaggregated into capital city dwellings and non-capital city dwellings using data on the number of households by capital city from ABS Cat No 3101.0.

The median price of established dwellings in cities and in non-metropolitan areas is available from the CBA/HIA. By multiplying this price measure by the corresponding quantity measure we are able to obtain capital city and non-capital city housing wealth measures for each of the five states. Summing the capital city and non-capital city housing wealth measures yields a total housing wealth measure for each state.

Footnotes

Tasmania, ACT and the Northern Territory are excluded due to data limitations. [25]

As we focus on long-run estimates, the precise quarterly profile should be less important. However, we also checked robustness by shortening the sample. [26]

The 1998 HES was also available. We preferred the 1994 HES because it is closer to the midpoint of our sample period. Using the 1998 HES affects the point estimates but does not change the overall result of our study. [27]

For superannuation we augmented this such that N i is the number of people in state i who reported earning superannuation income rather than the total number of people from state i in the survey. This is to account for the fact that most people with superannuation wealth do not receive superannuation income until they retire. [28]