«SUBURBAN SOCIO-SPATIAL POLARISATION AND HOUSE PRICE CHANGE IN MELBOURNE: 1986 – 1996 Margaret Reynolds, Research Fellow, School of Geography & ...»
SUBURBAN SOCIO-SPATIAL POLARISATION
AND HOUSE PRICE CHANGE IN MELBOURNE:
1986 – 1996
Margaret Reynolds, Research Fellow, School of Geography & Environmental Science, Monash
Correspondence to Margaret Reynolds: Margaret.Reynolds@arts.monash.edu.au Associate Professor Maryann Wulff, School of Geography and Environmental Science, Monash University Correspondence to Maryann Wulff: Maryann.Wulff@arts.monash.edu.au This study examines the process and pattern of spatial polarisation in Melbourne Australia between 1986 and 1996. We construct a five-category polarisation typology based on the relative change in the bottom and top ends of local suburban household income distributions. The suburbs are classified as either: in- creasing advantage, increasing middle income, stable, polarising or increasing disadvantage. The research then examines the relationship between the suburb classification and house prices over the same period.
The spatial units are 327 Melbourne suburbs. The two primary data sources include Australian Bureau of Statistics (ABS) household income figures and Victorian state government house sale price data for 1986 and 1996. The maps reveal a contiguous sector of increasing advantage in Melbourne’s inner and nearby eastern suburbs encircled by an adjacent middle suburban ring characterised by growing disadvantage.
Spatially this picture of polarisation corresponds closely with the map showing median house price change between 1986 and 1996. The polarisation categories are closely related to real quartile house prices with the highest house price increases in suburbs of increasing advantage and the lowest gains (or declines) in suburbs increasing in disadvantage.
INTRODUCTIONIn this era of globalisation and rapid economic, political and social change, new patterns in the nature of social distributions in cities are emerging. In many urban regions, the numbers of low and high income households have been increasing at the expense of the middle income groups.
This widening gap between the rich and the poor, alongside the decline of the middle class, has been termed ‘social polarisation’. The spatial translation of social polarisation into the urban landscape is described as socio-spatial polarisation. Because of the acknowledged role that geo- graphic location plays in contributing to, and even intensifying, social advantage and disadvantage (Badcock 1984; Maher et al. 1992; Lee 1994; Van Kempen 1997), socio-spatial polarisation directly affects the well being of the population. Housing market operations play a fundamental role in generating socio-economic spatial patterns in cities (Van Weesep et al. 1992; Badcock 1995;Burbidge et al. 1996; Hamnett 1996; Andersen et al. 2000).
In Australia, the housing market is particularly important in understanding the links between socio-economic status and polarisation because of the economic and cultural significance that home ownership holds. Around 66 per cent of households either own their home outright or are purchasing (ABS 2002) and this figure has held steady since the 1960s. For most Australian households, their home is their single most valuable asset and the equity gained in the family
Figure 1 Location maps, Melbourne Australia Vic Map Digital, Land Victoria; ABS (2002)
04-2 SUBURBAN SOCIO-SPATIAL POLARISATION ARTICLESMelbourne’s urban form, similar to other Australian capital cities, is a low density sprawled metropolitan region. The metropolitan area covers 7,694 square kilometres and runs approximately 116 kilometres north to south and 122 kilometres east to west. Historically the major sociospatial division in Melbourne has been between the working class northern and western suburbs of the city, and the more affluent eastern and southern regions. Socially, ‘the dividing line in Melbourne, in general terms, was the Yarra River’ (Burnley 1980 p. 228). During Melbourne’s establishment in the 1800s, the marshy environment and flat basalt plains to the west were less physically pleasing to the growing population than the better soils, undulating terrain, higher rainfall and coastal plain found in the east and south. As a consequence, the eastern and southern parts of the city drew in higher status residential development (Johnston 1966). Later, the location of public transportation (trains and trams) shaped city growth and, following the Second World War, the rapid rise in automobile ownership contributed to the continuing sprawl of the city.
For the most part, however, the visible differential in socio-economic status in Melbourne was limited to two broad sectors of the city, divided by the Yarra River.
During the post war urbanisation process, Melbourne’s inner suburbs were largely working class and the housing stock consisted of 19th century terrace houses and semi-detached housing.
In the 1970s these suburbs began to experience intense gentrification as the attractiveness of the location and the value of dwellings soared (Maher 1982). Several industrial and manufacturing sites dotted the northern and western suburbs (alongside newly constructed public housing estates of single detached dwellings). The south-eastern fringe also contained a post war public housing estate built to house the workers and their families in a new manufacturing plant. Most of the dwellings in these estates were ‘modest and low-cost’ three bedroom fibro houses (Burnley 1974;
Wulff et al. 1983). In the meantime, the inner and middle-eastern suburbs continued as the high status locations.
During the 1980s and 1990s Melbourne’s neighbourhoods were influenced by economic, social and political shifts. According to Burbidge (2000) housing price increases and capital gains in real dollar terms have ‘had a substantial class bias’, favouring higher income groups and expensive locations more substantially than low income groups and lower priced areas. Growing divisions in the housing market also appeared between existing and aspiring home owners, particularly those in the inner Melbourne regions compared with others living in outer fringe suburbs (Burke et al. 1990). Maher (1994) identified distinct spatial differences in the distribution of house price changes in Melbourne in the late 1980s and argued that such differentials create inequties in: population and labour mobility; access to housing; the environment in which affordable housing is found; and access to public goods and services – all of which influence future urban development. Furthermore, a recent study that ranked Australian communities by their level of ‘opportunity’ or ‘vulnerability’ found that Melbourne contained a higher level of polarisation than found in other capital cities (Baum et al. 1999). Finally, aggregated household income data used in the current research also reveals growing economic polarisation within Melbourne over the decade 1986 to 1996. Figure 2 shows that at the aggregate city-level, Melbourne’s household income distribution became more polarised over the 10-year period. This is shown in the fact that both high and low income households increased in number, while the moderate and moderatehigh household income categories changed negligibly. The polarising trend was weighted heavily at the low end of the household income distribution, with growth in both low and low-moderate 04-3
SUBURBAN SOCIO-SPATIAL POLARISATION ARTICLESincome households exceeding that of high income households. In fact, the absolute number of low income households increased by more than 84,000 over the ten year period: more than three times the number of high income households. Significantly, the number of moderate and moderatehigh income households changed very little. Therefore, in a broad sense, Figure 2 characterises the widening ‘gap’ between Melbourne households in terms of their economic and social opportunities.
Figure 2 Per cent change in household income categories, Melbourne 1986 to 1996 *Dollar values for each household income category are provided in Table 1.
ABS Census of Population and Housing, 1986 and 1996 This paper examines the nature of socio-spatial polarisation in relationship to housing market changes for Melbourne suburbs between 1986 and 1996. In this analysis, polarisation refers to a dynamic process, whereby the patterns of change over time are of a polarising or dividing nature (Walks 2001). This study classifies ten-year trends in household income and house prices for each of 327 Melbourne suburbs. The paper first develops a five category polarisation typology based on the relative growth or decline in household income groups. This typology is then mapped and the association between the stages of polarisation and quartile house price changes within suburbs is discussed.
RESEARCH DESIGNThe unit of analysis is the suburb1. The 327 Melbourne suburbs range in population size from around 650 households in growth areas to nearly 17,000 households in the more established areas and are the locations commonly referred to in real estate advertisements. Because many spatial polarisation studies have used very broad spatial units or city-level data, small area patterns tend to remain undetected. Dale et al. (1989), for example, used aggregate British data, and though they examined some regional differences they did not analyse intra-regional variations.
In a paper on social polarisation and welfare state regimes, Hamnett (1996) also examined differences at a broad, British regional scale. Likewise, in Australia, Baum’s examination of social polarisation in Sydney used aggregate, city-level data (Baum 1997), and similarly, Murphy et al.
(1994) employed city-level data to examine and compare social polarisation across a number of
04-4 SUBURBAN SOCIO-SPATIAL POLARISATION ARTICLESAustralian cities. By using a finer spatial unit than most studies, and then mapping these results, issues such as spatial proximity, spatial contiguity and spatial concentration within metropolitan Melbourne can be addressed.
Two major data sources form the basis of the analysis. First, Australian Bureau of Statistics (ABS) household income data were obtained for both 1986 and 1996 (in A$ 1996). Real household income was disaggregated into five income categories (Table 1)2, with the categories defined to align closely with the household income quintile values for Australia in 1986 and 1996.
Table 1 Gross weekly household income categories, 1986 and 1996 Derived from 1986 and 1996 Census questions on gross weekly income of all persons in the household aged 15 years or older. These categories were used in Yates et al. (2000).
The second data source consists of individual house price sales records from the Victorian Government’s Office of the Valuer General. The data set provides the price and address of each house sale in Melbourne in the calendar years of 1986 and 1996. Only residential dwellings coded as a separate ‘house’ were included in the analysis, thereby excluding units, flats or other dwelling types. In both 1986 and 1996, separate detached dwellings comprised around 75 per cent of the total dwelling stock in Melbourne. To analyse change in real prices over time, the Melbourne All Groups Consumer Price Index (CPI) was used to convert each 1986 house sale price into 1996 dollars.
Each individual house price record was cleaned and geocoded to its address point in the Melbourne metropolitan area3. Of the approximately 49,500 house sales in 1986, 86 per cent were geocoded to cadastral parcel level. In 1996, 89 per cent of just over 48,400 house sales were geocoded to the parcel level. Overall, 95 per cent of sales in both years have been geocoded to the suburb level. Once geocoded to this scale, the individual sales could be spatially aggregated to any set of user-defined areal units, in this case, the 327 Melbourne suburbs. The flow chart in Figure 3 summarises the data sources and data processing steps undertaken in the analysis.
Real median house prices for 1996 and house price growth between 1986 and 1996 are mapped by suburb. House prices for 1986 and 1996 are summarised according to real quartile values and quartile per cent changes for each polarisation type.
The polarisation typology is based on the changes in the household income distribution between 1986 and 1996 for each Melbourne suburb. This approach improves upon other methods that show household income distribution as an average, median or ratio value for individual spatial units. The latter do not reveal polarising trends within the spatial units. Polarisation is conceptualised in this research as a process of change at both ends of the household income structure: that is, the simultaneous increase in the number of households in both the low and high income categories. By mapping these results, the polarising trends within each suburb, and
The difference between this expected change, and the actual change, provides the adjusted index score. This score varies around 100. A score below 100 represents less than expected growth (or an absolute decline) in a household income category over time, while a score above 100 indicates that the category grew more than expected over the decade. In other words, a category with a score over 100 increased its share of households in the suburb. If the household income distribution in any particular suburb remained unchanged over the decade, then the number of households in each income category would be expected to change at the same rate as the overall household growth in that suburb. For example, if the number of households in a 04-7
SUBURBAN SOCIO-SPATIAL POLARISATION ARTICLESsuburb increased by 5 per cent between 1986 and 1996, then the number of households in each income category could also be expected to grow by 5 per cent. This would result in an unchanged income structure within the suburb.
This measure takes into account the disparity in household growth rates in the Melbourne metropolitan area. Figure 4 shows the household growth rates of Melbourne suburbs between 1986 and 1996. The map reveals the much higher household growth in Melbourne’s outer and fringe areas compared with the inner and middle areas.