COMPARISON OF THE NEW AND PREVIOUS EXPERIMENTAL METHODOLOGIES
The effectiveness of approaches taken to impute a market rent for individual owner-occupied dwellings is dependent on the availability of data on price determining property characteristics in the private rental market. Data are not available in SIH at the individual dwelling level to better define location in terms of attributes such as views or beach frontage and proximity to employment, transport, and shops/services. Data are also not available on the value of rented dwellings, nor for some important physical characteristics of the dwellings such as outer-wall construction, availability of garaged or off-street parking, size of block or number of bathrooms.
In the previous methodology, hedonic regression was used to estimate the market value of the rental equivalent for owner-occupied dwellings and other subsidised tenure types. Data from the SIH on reported rents paid by private market renters was regressed on the characteristics of their rented dwellings e.g. location and dwelling structure. The estimated coefficients were then applied to the corresponding characteristics of owner-occupied and other subsidised tenure dwellings to produce imputed values of the gross rental equivalence for these dwellings.
A shortcoming of the hedonic regression rental equivalence approach is that reliable results cannot be produced when rental markets are limited or do not exist (recognised in the Canberra Group Handbook on Household Income Statistics). In the SIH, rental dwellings are very limited for high value housing stock. An extrapolation method was used to partially compensate for this problem but further analysis indicated that it resulted in a substantial underestimate of the imputed rent for these owner-occupied dwellings.
The inclusion of dwelling price data in the new experimental methodology for owner-occupied dwellings overcomes this problem, as well as taking better account of the quality differences that are likely to exist between many owner-occupied and rental dwellings with other similar characteristics. For example, owner-occupied dwellings may generally have higher quality fittings or building materials, or be maintained to a higher standard than many rental dwellings and this would be expected to be reflected in the value of the dwelling and in the market rent it would be likely to attract.
EXPERIMENTAL ESTIMATES OF IMPUTED RENT FOR OWNER-OCCUPIED DEWLLINGS
The distribution of gross imputed rent has changed between the previous and new experimental methodologies. As shown in Graph 1, the new methodology has resulted in a more dispersed distribution.
In the previous methodology, hedonic regression could only be used when there were sufficient rental properties to use the regression model, which was at a much lower value than rents imputed using the new methodology.
Graph 1. Distribution of gross imputed rent for owner-occupied dwellings, new and previous experimental methodologies, 2011‒12
Table 1 shows the impact of the new experimental gross imputed rent methodology for selected percentiles for owner-occupied dwellings. The total average increase for all households between the new and previous methodologies is 16%, due mostly to the better estimation of market rents for higher value dwellings. The gross imputed rent at the 80
th and 90
th percentiles is more that 25% higher in the new methodology compared to the previous methodology.
Table 1. Gross imputed rent at top of selected percentiles for owner-occupied dwellings, new and previous experimental methodologies, 2011–12
|
 | New
methodology | Previous
methodology | Difference |
| Percentiles | $ per week | $ per week | $ per week | % |
|
| P10 | 192 | 239 | -47 | -20 |
| P20 | 249 | 267 | -18 | -7 |
| P30 | 290 | 293 | -3 | -1 |
| P40 | 328 | 319 | 9 | 3 |
| P50 | 367 | 343 | 24 | 7 |
| P60 | 414 | 370 | 45 | 12 |
| P70 | 476 | 402 | 74 | 18 |
| P80 | 566 | 446 | 120 | 27 |
| P90 | 742 | 558 | 184 | 33 |
| All households | 445 | 374 | 71 | 19 |
|
Table 2 shows the impact of the new experimental methodology on the equivalised disposable household income (EDHI) including imputed rent for owner-occupied dwellings. As the housing costs used to calculate net imputed rent have not changed in the new model, the impact is solely due to the change in gross imputed rent. In 2011‒12, the EDHI (including imputed rent) increased by an average of $43 per week (4%) for all households. The greatest impact was on the highest quintile (up by 6%).
Table 2. Distribution of equivalised disposable household income (incl. imputed rent), new and previous experimental methodologies, 2011–12
|
| | New
methodology | Previous
methodology | Difference |
| $ per week | $ per week | $ per week | % |
|
| Mean income | | | | |
| Lowest quintile | 448 | 446 | 2 | 0 |
| Second quintile | 713 | 694 | 18 | 3 |
| Third quintile | 923 | 893 | 29 | 3 |
| Fourth quintile | 1 202 | 1 156 | 46 | 4 |
| Highest quintile | 2 118 | 1 996 | 121 | 6 |
| All households | 1 081 | 1 037 | 43 | 4 |
| | | | |
| Income at top of selected percentiles | | | | |
| P10 | 496 | 500 | -4 | -1 |
| P20 | 611 | 600 | 10 | 2 |
| P30 | 710 | 695 | 15 | 2 |
| P40 | 812 | 792 | 20 | 3 |
| P50 | 920 | 889 | 31 | 3 |
| P60 | 1 041 | 1 005 | 36 | 4 |
| P70 | 1 189 | 1 149 | 40 | 3 |
| P80 | 1 419 | 1 358 | 61 | 4 |
| P90 | 1 773 | 1 702 | 72 | 4 |
|
Table 3 shows that the new experimental methodology for calculating gross imputed rent has a similar impact on both gross imputed rent and EDHI (including imputed rent) estimates for all SIH cycles from 2003–04 to 2011–12.
Table 3. Comparison of income estimates for owner-occupied dwellings, new and previous methodologies
|
| Mean gross imputed rent | | Mean equivalised disposable household income (incl. imputed rent) |
|
| |
|
| New methodology | Previous methodology | Difference | | New methodology | Previous methodology | Difference |
| Period | $ per week | $ per week | $ per week | % | | $ per week | $ per week | $ per week | % |
|
| 2003–04 | 317 | 264 | 53 | 20 | | 827 | 795 | 33 | 4 |
| 2005–06 | 334 | 286 | 48 | 17 | | 908 | 878 | 30 | 3 |
| 2007–08 | 382 | 317 | 65 | 21 | | 1 058 | 1 019 | 39 | 4 |
| 2009–10 | 418 | 357 | 61 | 17 | | 1 053 | 1 016 | 37 | 4 |
| 2011–12 | 445 | 374 | 71 | 19 | | 1 081 | 1 037 | 43 | 4 |
| 2013–14 | 491 | na | na | na | | 1 204 | na | na | na |
|
REFERENCES
UNECE (2011),
Canberra Group Handbook on Household Income Statistics, Second Edition, ECE/CES/11, Geneva.
Print Page
Print All