![Sample our Health and Social Care journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5938/TOC-Subject-Sample-2021-Health-Social-Care.jpg"})
Abstract
The asset-based welfare approach, which has foremost encouraged homeownership, has led to rising homeownership rates, house prices and household debt levels. While this shift has helped raise the net worth of some among the middle and working classes who own property, the implications for the spatial distribution of wealth in cities have not yet been explored. This paper examines the spatial implications of the rise of policies promoting asset-based welfare, by examining statistically how variables related to homeownership rates and housing prices relate to measures of urban wealth segregation among neighbourhoods. Canadian cites are used as the main case study for the empirical analysis. The findings suggest that while homeownership in general has an equalizing effect, rising rates of homeownership (and to some extent, rising house prices) are associated not with greater spatial equalization and dispersal of wealth, but instead with greater spatial segregation and concentration of wealth within cities.
Funding
This work was supported by the SSHRC (Social Sciences and Humanities Research Council of Canada). The author takes all responsibility for the content.
Notes
1 The gini coefficient rose from 0.655 to 0.705, here calculated using adult-equivalent household net worth, to control for changes in household size and composition (ibid.). The maximum value of the gini coefficient is 1 (indicating that one unit has all the income or wealth), while the minimum value is 0 (indicating an equal distribution).
2 A related concept is “augmented wealth”, which adds to net worth the annuity value of private and government pensions, and various welfare-state benefits that are usually progressively redistributive. While broader, the latter conceptualization of wealth is also much more difficult to estimate, and is subject to uncertainty regarding future rates of return and benefits levels.
3 Canadian data calculated by the author from the 1991 and 2006 census of Canada. The United States went from a homeownership rate of 64.1 per cent in 1991 to a peak of 69 per cent in 2004 (US Census Bureau, Citation2007).
4 Brzozowski et al. (Citation2010) measure wealth and income using adjusted adult-equivalized data for working-age-headed households, which controls for changes in household structure and size. However, it is a different measure than that used by other researchers.
5 The gini, this time measured among all families unadjusted (instead of just working-age headed households) rose from 0.691 to 0.727 (ibid.).
6 Census tracts are spatial units created by Statistics Canada to represent neighbourhoods. They follow established natural and human-made routes and boundaries. The majority of tracts range between 4000 and 8000 in population, and good-quality census data are made universally available at this scale in the census. Census metropolitan areas (CMAs) are metropolitan areas with populations of 100,000 or more. Census agglomerations (CAs) are urban areas with populations between 10,000 and 100,000.
7 It is typically assumed that when spatial units are relatively smaller—when each unit holds a smaller proportion of the total population under study—that resulting segregation indices will provide larger coefficients than when the spatial units are larger. This has been shown for measures of racial segregation (Wong, Citation1997). Inclusion of the log of population size in the regression models controls for this effect in the current study.
8 All variables could not be input simultaneously as a single block, as the total number of variables (35) is larger than the number of metropolitan areas for which there is data (33), largely due to the inclusion of the occupational variables (necessary for testing for the presence of class and occupational restructuring effects). Tests of alternate methods of estimating the models using fewer variables suggest that these would not produce significantly different results, but would limit the testing of potential variable interactions.
9 All Pearson correlations in column 9 of Table are statistically significant at p < 0.01.
10 In cases where a corporation (such as a real estate investment trust, or REIT) owns and rents out properties, shares in those companies are included as financial liquid wealth in the data set.
11 While areas specialized in the FIRE sector (finance, insurance and real estate) reveal higher levels of segregation in financial assets, this would appear to be balanced out by higher levels of spatial concentrations in debt in more wealthy areas in such places.
12 Separate variables for the three most common visible minority groups (Blacks, South Asians, East Asians) were also tested in the models, but not found to have any effect on differences in the neighbourhood segregation of wealth among CMAs.
13 Furthermore, in Table there is similarly a significant association between rising housing values since 1991—an outcome of Canada’s attempt to stimulate homeownership through its mortgage credit policies—and higher ginis for neighbourhood segregation of both liquid assets and total marketable wealth. Note that the latter is not an effect of population growth, which is simultaneously controlled for in the models (and for which the coefficients are negative, suggesting that population growth has an equalizing effect on levels of wealth segregation overall). However, the variables for dwelling values, and dwelling value change, fall out of the OLS models for the Atkinson indices.