This document was added 30/07/2010.

A VIEW OF HOUSING DENSITY IN PERTH, 2005-2009


INTRODUCTION

Over the four years to June 2009, Western Australia was the fastest growing state by population, with Perth being the fastest growing capital city (Regional Population Growth, cat. no.3218.0). As a result, there has been much debate and interest in strategic planning initiatives to ensure that adequate land, infrastructure and services will be available to accommodate the expanding population.

Under a national objective, aimed at ensuring that 'Australian cities are globally competitive, productive, sustainable, liveable and socially inclusive and are well placed to meet future challenges and growth', the Council of Australian Governments (COAG) has agreed to a series of criteria relating to land-release policies; urban design; integration of nationally significant infrastructure; the clarity of future planning measures; and policies to encourage investment. As a corollary to the COAG criteria, the Western Australian government has also been addressing residential density, specifically for Perth.

This study explores the extent to which residential densities in Perth may be changing, as indicated by the relative proportions of building approvals for low, medium and high density housing, in recent years compared with a decade ago.


BACKGROUND

The ABS publishes monthly building approvals data, collected primarily from local government authorities, for all states and territories. The Functional Classification of Buildings (FCB) classes residential dwellings into three broad types: houses; semi-detached, row or terrace houses, townhouses, etc; and flats, units or apartments (see ABS Functional Classification of Buildings, 1999). Although plot sizes would provide additional useful information pertaining to the proposed number of dwellings per hectare, these three dwelling types may be roughly interpreted to align with low, medium and high density housing respectively.

In 2001, the ABS published a study investigating the mix of dwelling types for the Perth metropolitan area, using building approvals data for three financial years from 1998-99 to 2000-01. With a focus on trends in medium density development, the study employed a modified grouping to identify Clustered dwellings. Clustered dwellings included Grouped (two or more) detached houses on a single parcel of land, as well as Semi-detached, row or terrace houses and townhouses. Under the FCB, these Grouped houses remain under the Houses category by virtue of their detached status.

Grouped houses, however, generally have smaller lot sizes, often have shared driveways and gardens and potentially lower average values, thereby demonstrating many similarities with the semi-detached building category. Combining the two types to form the category Clustered dwellings provides a different representation of medium density housing approvals in Perth and offers the potential to better inform decision-makers with regard to planning and infrastructure development for the community (see A View of Housing Density in Perth, Dec 2000).

Renewed interest among WA agencies concerned with future housing development in the Perth metropolitan area indicated the need to investigate present trends in dwelling type approvals and changes that may have occurred since 2001. This article presents the findings of the follow-up investigation.


BUILDING APPROVALS IN PERTH, 2005-2009.

The table below presents the total number of dwelling units approved in the Perth Statistical Division (SD), according to building type as classified by the FCB over the four financial years 2005-06 to 2008-09.

NUMBER OF DWELLING UNITS APPROVED, Perth Statistical Division
Building type (FCB)	2005-06	2006-07	2007-08	2008-09	Total
Houses	15 388	13 400	11 732	11 104	51 624
Semi-detached, etc(a)	1 606	1 890	2 158	1 629	7 283
Flats, units and apartments	1 468	2 126	3 033	1 053	7 680
Total	18 462	17 416	16 923	13 786	66 587

(a) Includes row or terrace houses and townhouses.
Source: ABS data available on request, Building Approvals, Australia (cat. no. 8731.0).

There was a steady decline in total dwelling units approved for the Perth SD over the four years, with the number in 2008-09 (approximately 13,800) being 25% lower than the number in 2005-06. The drop in building approvals in 2008-09 was largely driven by the high level of approvals for Flats, units and apartments in the preceding year which was followed by a significant reduction in 2008-09. The total decline in approvals for that year also reflects the loss of home buyer confidence as a result of the global financial crisis.

Approvals for Houses made up 78% of all dwelling unit approvals over the four years, with a further 11% of approvals being for Semi-detached, row or terrace houses and townhouses and 12% for Flats, units and apartments. These proportions were fairly consistent across the four-year period and consistent with the years analysed in the 2001 study. An exception was 2007-08, when the proportion of Houses fell considerably and the proportion of Flats units and apartments increased.

The proportions of approvals by building type also mirror the distribution of housing stock as reported in the 2006 Census of Population and Housing. Records for the Perth SD from the 2006 Census show 78% of dwellings as Separate houses, 12% as Semi-detached, row or terrace houses and townhouses, and 10% as Flats, units or apartments.


AN ALTERNATIVE PERSPECTIVE ON HOUSING DENSITY

Methodology

This analysis used the same methodology as the 2001 study to identify Clustered dwellings and applied it to building approvals data from the four financial years from 2005-06 to 2008-09 (the most recent data at the time of writing). As in the previous study, the analysis was largely based on aggregated data to 'smooth out' potential aberrations in annual figures. Because of the significant effects of the global financial crisis on residential building approvals in recent years, the decision was made to include four, rather than three, financial years.

To enable comparisons between the findings of the two studies and measure some aspects of change in housing density since 2001, as portrayed by building approvals, the findings are presented in a generally similar manner. However, the measure of Gross Site Area (GSA), a significant indicator of housing density included in the 2001 analysis, is not included here because this information was reported for only 13% of building approvals between 2005 and 2009 and was deemed unsuitable for dissemination. Although the floor area of the dwelling, which is better reported by local councils, is a less than perfect surrogate for gross site area, some analysis of building approvals by floor area was undertaken in this study to provide further insight into the densification question. Building approvals were also analysed according to dwelling values to see how these relate to dwelling density across the Perth Statistical Division (SD).

As with the previous study, this analysis was undertaken with the knowledge that building approvals statistics were not specifically designed to measure trends in dwelling density. Local observation suggest there has been an increasing trend of single detached houses being demolished and replaced by two or three grouped houses. The absence of data about demolitions represents a significant gap in the data.

Other Data Limitations In considering how well dwelling approval records may inform about likely outcomes in housing density, and whether apparent trends in density may either be missed or overstated, the following caveats are provided to explain potential data quality issues. It is also likely that approving bodies or councils differ in the way they report approvals. For greenfields, or large in-fill developments undertaken by a single developer, several houses may appear with essentially the same address though they are being built on separate lots. An individual approval may include multiple houses being built in an area by one developer but at separate addresses. An approval relating to more than one dwelling does not necessarily mean that the dwellings are clustered or collocated to an address. In-fill resulting in additional dwellings on a lot with already established dwelling(s) is not identifiable. This includes cases where a block has been subdivided as well as direct in-fill on the same lot. Multiple approvals may be received for attached dwellings being built to the same lot.

With the above data limitations in mind, the following analysis provides an alternative view of low medium and high density housing based on residential dwelling approvals, in which:

• Single Houses are detached dwellings on a single parcel of land and represent Low density dwellings.
• Clustered dwellings are the sum of Semi-detached, row or terrace houses and townhouses and Grouped houses on a single parcel of land and represent Medium density dwellings.
• Flats, units and apartments remain their own group and represent High density dwellings.

The following chart presents this alternative view of housing density relating to building approvals in the Perth SD between 2005 and 2009.
(a) FCB classification codes 111 to 113.
(b) FCB classification codes 121 and 122.
(c) FCB classification codes 131 to 134.


Based on the FCB, there were 51,600 Houses approved over the four financial years reported, representing 78% of total dwelling units approved. Of those, 6,900 (13%) were identified as consisting of more than one house built on a single parcel of land (Grouped houses). The proportion of these Grouped houses to total houses varied from a high of 15% in 2006-07 to a low of 11% in 2008-09. Removing Grouped houses from the Houses category resulted in Single houses approvals making up 67% of total approvals. This proportion was at its lowest in 2007-08 (61%) and its highest in 2008-09 (72%).

In comparison, based on the FCB, there were 7,300 approvals for Semi-detached, row or terrace houses and townhouses approved over the same four year period, representing 11% of total dwelling units approved. Combining these with approvals for Grouped houses resulted in 14,200 units identified as Clustered dwellings. These Clustered dwellings comprised 21% of total approvals, a proportion that remained relatively unchanged over the four financial years.

Differences in the distribution of building approvals coded to the FCB and the modified grouping for approvals are shown in the graphs below.

Alternative Views of Building Approvals by Housing Density, 2005-2009(a)

The findings from this study indicate that there was little change in the broad mix of approvals for Single houses and Clustered dwellings from 2001. The 2001 study reported that Single houses accounted for 68% of all residential building approvals, as compared with 67% in the present study. Similarly, the 2001 study showed that Clustered dwellings accounted for 23% of all approvals, a slightly higher proportion than was found in the current study (21%).

Although the modified groupings used in the analysis had no impact on the category of Flats, units and apartments, approvals for this accommodation type increased as a proportion of all approvals from 9% to 12% between the two study periods. Thus the three percentage point increase in approvals for this High density accommodation was associated with a 1% fall in approvals for Low density (Single houses) and a 2% fall in approvals for Medium density (Clustered dwellings).

The following section presents more detail about the types of clustered dwellings that were approved in the 2005-09 period and compares with results from the 2001 study.


CLUSTERED DWELLINGS

The extent to which dwelling density in urban areas is increased by building Clustered dwellings, rather than Single houses, partly relates to the number of dwelling units associated with each Clustered dwelling approval. The following table shows approvals for Clustered dwellings in the Perth SD for the four years to 2009 according to the number of dwelling units associated with these approvals.

	NUMBER OF DWELLINGS
Dwelling units per approval	2005-06	2006-07	2007-08	2008-09	Total	% of total
2	792	672	736	608	2 808	19.8
3	876	939	837	705	3 357	23.6
4	480	456	484	320	1 740	12.2
5 or more	1 644	1 857	1 567	1 206	6 274	44.2
Other (b)	10	14	3	2	29	0.2
Total	3 802	3 938	3 627	2 841	14 208	100.0

(a) Grouped houses on a single parcel of land plus Semi-detached, row or terrace houses, townhouses.
(b) Unable to establish cluster category.
Source: ABS data available on request, Building Approvals, Australia (cat. no. 8731.0).

Over the four year period, the highest proportion of Clustered dwellings were for approvals of five or more dwellings units (44%), a slight increase on the proportion presented in the 2001 study (41%). These approvals are likely to include quite large developments such as retirement villages.

Clustered dwellings of three units made up 24% of total clustered dwelling approvals, which was a considerably higher proportion than reported in 2001 (16%). In contrast, Clustered dwellings with two units per approval comprised only 20% of total clustered approvals, down from 30% reported in the previous study.

The recent trend away from building two Clustered dwellings in favour of three Clustered dwellings may reflect a shift towards higher density grouped developments to allow for greater utilisation of available land resources. Such a trend may be seen as addressing, to some degree, the need for greater densification to accommodate Perth's expanding population.


GROSS SITE AREA, FLOOR AREA AND VALUE

Looking at changes in building approvals by dwelling type gives a partial view of trends in housing density. It is also important to consider the size of the site area or plot on which the dwellings will be built. While the 2001 article analysed Gross Site Area (GSA) attaching to dwelling approvals, this data item was reported for less than 13% of dwelling approvals over the last four financial years and was therefore deemed unsuitable for use in this study.

Although floor area relates only to the dwelling itself and not to the land it is built on, it does provide a useful indication of building size and perhaps sheds some light on associated patterns of housing density. Floor area, as reported in building approvals data, was therefore investigated in this study to test the notion that grouped houses would have a lower average floor area than single houses on their own parcel of land.

Average Floor Area

For the period between July 2005 and June 2009, floor area was reported for 70% of Houses approved in the Perth SD. This reporting rate allows for analysis of floor size for this dwelling type. During the same period, however, reporting rates for floor area of Semi-detached dwellings and for Flats, units and apartments were only around 35%: this level of reporting precludes anything but the broadest general comment about sizes of these dwelling types. Across all dwelling types, non-reporting of floor area was concentrated in certain areas; for example in 2008-09, nearly 2,500 approvals issued in the cities of Stirling, Gosnells, Nedlands and Perth omitted this value.

Floor area is reported in square metres and relates to any part of the building bounded by external walls, excluding carports and pergolas. Average floor area data represents the average only of records for which floor area has been reported.

The table below shows average floor area for Single houses, Grouped houses and all Houses over the four years of the study period.

AVERAGE FLOOR AREA (a), House Approvals - Perth Statistical Division
	2005-06	2006-07	2007-08	2008-09	Total
	Square metres (m2)
Single houses (b)	262	268	273	254	264
Grouped houses (b)	164	166	170	166	166
All Houses	253	258	264	247	256

(a) Floor area information reported for only 70.3% of house approvals.
(b) On a single parcel of land (modified grouping).
Source: ABS data available on request, Building Approvals, Australia (cat. no. 8731.0).

In the years 2005-09, average floor area for dwelling approvals in the Perth SD was consistently larger for Single houses on their own parcel of land than for Grouped houses although, in both cases, average floor area increased over the first three years of the study period and then decreased in the final year. It is likely that smaller average floor areas in 2008-09 can be attributed to the stimulating effect of the extra boost to the First Home Owners Grant. This scheme brought about an increase in the number of first home buyers choosing to build, many of whom would have built smaller, entry-level, houses.

Based on aggregated approvals data for the 2005-2009 period, Grouped houses had an average floor area of 166 square metres, nearly 100 square metres, or 59%, less than that of Single houses (264 square metres). This smaller average floor area for Grouped houses serves to 'deflate' the overall average for Houses (as classified by the FCB). In addition, average floor area for Grouped houses was consistently smaller where there were more dwellings in the cluster. Very generally, where a cluster approval was for two dwellings, floor area for each averaged around 180 square metres. Where a cluster approval was for five dwellings, average floor area was around 150 square metres.

By way of comparison, and although poorly reported, average floor area over the four years to 2009 was 120 square metres for Flats, units and apartments, and 164 square metres for Semi-detached, row or terrace houses, and townhouses

These data tend to support the hypothesis that grouped houses with smaller floor areas are currently making some contribution to the process of urban densification in the Perth SD. However, a reliable measure of the Gross Site Area, or the number of dwellings per hectare in each approval, would be necessary to provide positive proof of this hypothesis.

Average Value

One factor that is likely to impact greatly on demand for medium density dwellings in the Perth SD, as in other Australian capital cities, is the high cost of housing. Comparing the average value of building work approved for different dwelling types gives an insight into the impact of value as a motivator behind trends in housing density. Particularly in light of recent economic conditions, the desire for lower value dwellings and better returns for investors and developers may be an important factor when deciding upon dwelling/density type.

Average values are derived from estimated value of construction work as reported on building approval documents provided to local councils. These data exclude the value of land and landscaping but include site preparation costs. These estimates are usually a reliable indicator of the completed value of Houses. However, for Other residential buildings they can differ significantly from the completed value as final costs and contracts may not been established before council approval is sought and gained.

The following table shows average values for approvals across all housing types provided under the Functional Classification of Building as well as under the modified grouping.

AVERAGE VALUE(a) BY BUILDING APPROVAL TYPE, Perth Statistical Division
	2005-06	2006-07	2007-08	2008-09	Total
	$'000
Functional Classification of Buildings
Houses	196.5	234.1	260.9	254.5	233.4
Semi-detached, etc	160.4	177.6	200.6	195.5	184.6
Flats, units, apartments	245.2	300.0	436.7	348.6	350.2
Modified grouping
Single houses	205.7	246.1	270.3	260.2	242.8
Clustered dwellings	149.4	172.6	198.2	201.1	178.6

(a) Excludes the value of land and landscaping but includes site preparation costs.
Source: Building Approvals, Australia (cat. no. 8731.0).

Approvals for Single houses on a single parcel of land, perhaps not surprisingly, had a consistently higher average value than approvals for Houses classified to the FCB in the years 2005 to 2009. This is due to the lower values attaching to Grouped houses on a single parcel of land, deflating the average value of Houses. Overall, approvals for Single houses had an average value of $243,000 whereas the average approval value of Clustered dwellings was $179,000 ($64,000, or 26%, less).

Within Clustered dwellings, average values were lower where cluster size was higher, although value increased across all cluster types over the four year period. For higher density flats, units and apartments, the very high average value ($350,000) reflects the higher cost of construction for multiple storey buildings as well as the often luxurious nature and high quality design found in many new developments of this dwelling type, particularly in inner-city areas.



Over the four financial years 2005-2009, the average value of new Single houses on single parcels of land approved in the Perth SD increased by 26.5%. In this time period, average values were highest in 2007-08 before the global financial crisis impacted on the Perth housing market. For Flats, units and apartments, there was a similar peak in average values in 2007-08. However the average value of approvals for Clustered dwellings continued to rise in each of the four years.


REGIONAL ANALYSIS

The following maps display the spatial distribution of Building Approvals 2005-2009 within the Perth SD, aligned to the modified grouping for approvals. Pre-existing housing stock is not a consideration in this analysis. The distribution of Clustered dwellings, Single houses and Flats, units and apartments is shown for Statistical Local Areas (SLAs), which align very closely with Local Government Areas across the Perth SD.

Not unexpectedly, approvals for medium and high density dwellings tend to be concentrated in the older inner suburban areas of the city, while approvals for low density housing dominate in the outer metropolitan areas.

Approvals for Clustered Dwellings

Overall, more than one-fifth (21%) of residential building approvals in the Perth SD between 2005 and 2009 were for clustered dwellings. This proportion varied significantly across SLAs. Stirling - Central had both the highest proportion of approvals for Clustered dwellings (66% or two thirds of total approvals for that SLA) as well as the highest number of approvals for Clustered dwellings (3017). The next highest proportions of Clustered dwellings were in Stirling - South-Eastern (55%), Belmont (51%) and Canning (47%). Areas with low proportions of clustered dwellings included the outer metropolitan SLAs of Rockingham (7%), Mundaring (5%), Kwinana (2%) and Sepentine-Jarrahdale (1%).

In almost 40% of the 37 SLAs in the Perth SD, approvals for clustered dwellings amounted to one-fifth (20%) or more of all approvals. In six of these SLAs, the proportion of approvals for Clustered dwellings accounted for 35% or more of all approvals.

While the average value for Clustered dwelling approvals across the Perth SD, between 2005 and 2009, was only $179,000, much higher average values for this dwelling type were recorded in some areas. Highest average values for Clustered dwelling approvals were in the inner western suburbs of Cottesloe, ($740,000), Claremont ($609,000) and Peppermint Grove ($600,000). Lower average values for Clustered dwelling approvals were in Gosnells ($134,000) and Wanneroo - North East ($141,000). Swan and Armadale also had average values for this type of dwelling approval below $150,000.





Approvals for Single Houses

Across the Perth SD, in the four years to 2009, 67% of approvals were for Single houses on their own parcel of land. Approvals for this dwelling type comprised the large majority of building approvals in some SLAs. As a corollary to findings relating to the distribution of clustered housing, the areas with the highest proportion of Single house approvals were mainly in the outer metropolitan areas, including Serpentine-Jarradale (99%) and Kwinana to the south; and Mundaring (95%) to the east. Over this period, the City of Rockingham had the most approvals for Single houses, with over 5,500 approvals at an average value of around $200,000. Many of these areas are characterised by recent land releases (mainly for single house development) and are attractive to first home-buyers and young families due to their relatively low land values.

Approvals for new Single houses had the lowest average approval value in Kwinana ($154,000) and Gosnells ($179,000), compared with $243,000 across the Perth SD. At the other end of the range, the highest average approval value was for Peppermint Grove ($1.4m), followed by other western suburbs SLAs including Mosman Park ($921,000), Nedlands ($831,000) and Cottesloe ($822,000).





Approvals for Flats, units and apartments

While this study is primarily concerned with densification arising from medium density (Clustered dwelling) development, the use of Flats, units and apartments to create higher urban densities in the Perth SD should not be overlooked. In the four years to 2009, approvals for Flats, units or apartments accounted for 12% of total building approvals across the Perth metropolitan area. Approvals for these high density dwellings continued to be concentrated in inner-city SLAs, including Perth - Inner (100%), Perth - Remainder (99%) and Fremantle - Inner (96%). An exception was the northern coastal SLAs of Joondalup, where there has been a trend towards higher density development in recent years.

Average values for approvals for Flats, units & apartments ranged between $125,000 in Wanneroo - North East and $892,000 in Claremont, compared with $350,000 for the whole Perth SD.






CONCLUSION

This study sought to find a way of measuring changes in housing density in the Perth metropolitan area in recent years. In particular, the study focussed on changes in the proportions of building approvals for low and medium density dwellings.

Utilising a similar methodology to that used in 2001, this study identified minimal change in the proportions of Clustered (medium density) dwelling approvals. There has, however, been a trend towards higher density within Clustered dwelling developments, as evidenced by three-dwelling clusters becoming more prevalent than two-dwelling clusters. In addition, the proportion of approvals for Flats, units and apartments increased since the earlier study. Although medium-density housing was the primary focus of this study, future investigations would need to take into account changes over time in both high and medium density developments.

Regional analysis of the distribution of dwelling approval types across the Perth Statistical Division identified tendencies for medium and higher density housing approvals to be concentrated in older, already-developed inner city suburbs, while lower density approvals comprised a large proportion of approvals in outer suburbs. Given current government objectives for urban in-fill and densification, future analysis of approvals may identify a greater 'spread' of medium to high density approvals across the metropolitan area.

The extent to which densification goals of government are being met cannot be measured solely in terms of the relative proportions of low, medium and high density dwelling approvals. Other factors such as the size of dwellings and the size of the parcels of land on which they are located must also be considered. Anecdotal evidence suggests that the size of dwellings continues to grow while the size of the plot, particularly in some inner suburbs, is diminishing: this is also being facilitated by changes to residential zoning in some council areas.

There is a need for reliable and more detailed statistics at the LGA/SLA level, including estimates of approvals for 'granny flats' (i.e. second dwellings on plots already containing single houses) as well as data on demolitions. Another important data need is a reliable measure of Gross Site Area for each dwelling approval. This would facilitate the process of estimating the number of dwellings per hectare in any given area.

The next base, or stock measure, of housing density that the ABS will undertake is the 2011 Census of Population and Housing. Post 2011, an estimation of dwelling density increase could be achieved for the intercensal years, using building approvals by development type. This approach would provide a clearer view of trends in housing density and would enable state and local governments to monitor progress towards their goals of increased urban densification.


REFERENCES

ABS (various years). Building Approvals, Australia, cat. no. 8731.0.

ABS (various years). Regional Population Growth, Australia, cat. no. 3218.0.

ABS 2001. ABS Functional Classification of Buildings, 1999, cat. no. 1268.0.55.001.

ABS 2001. 'A View of Housing Density in Perth, 2001' in Western Australian Statistical Indicators, Dec 2001, cat. no. 1367.5.

ABS 2006. Basic Community Profile - Perth Statistical Division, 2006 Census Community Profile Series, cat. no. 2001.0.

Council of Australian Governments 2009. 'National objective and criteria for future strategic planning of capital cities'. Appendix B in COAG Communique, 7 December 2009.