Gentrification and change in canadian metropolitan areas



Download 106 Kb.
Page4/12
Date12.01.2021
Size106 Kb.
1   2   3   4   5   6   7   8   9   ...   12
The Method

A two-step process was used to identify gentrifying neighbourhoods within 10 CMAs. The CMAs were selected to achieve a diverse regional representation and include Halifax, Quebec City, Montreal, Ottawa, Kingston, Toronto, Winnipeg, Regina, Edmonton and Vancouver. The first stage used census tract data to trace changes between 1981 and 2001 in income, rents and then relate these changes to the tracts’ 1981 characteristics. The second step sought the views of local market analysts on the location and determinants of gentrification in their cities. The 1981 and 2001 census tract data was merged for the quantitative analysis creating a unique data set that would show changes on a small geographic scale. Statistics Canada defines tracts to an average of about 4,000 people and most of the boundaries of the inner city tracts remained the same over the study period. The combined data yields a total of 2,182 tracts in the 10 CMAs. For the split tracts that were recombined, weighted averages were constructed for some variables while others could be added together. All rents and incomes were standardized to 2001-dollar values. While not all the variables in the two censuses are the same, there is enough correspondence to show changes in income, rent and housing stock characteristics at the census tract level of disaggregation.


The Principal Components

The 1981-2001 changes in the census tracts were related to their 1981 characteristics by using principal component analysis.1 The variables include the 1981-2001 changes in rent, income, dwelling unit density as well as a set of categorical variables describing the 1981 stock characteristics as listed in Table 1. The principal components that would help identify the gentrifying neighbourhoods would be the ones most closely correlated with the rent and income increases as well as with the variables identifying the tracts that had retained the highest proportion of old buildings in 2001.



The last two columns in Table 1 present the correlations of the variables with the first two principal components. The first component absorbs 21.2 percent of the variance in the original variables and the second absorbs 9.0 percent. The first principal component clearly points to the center-periphery attributes of urban regions. One end of this axis is defined by a high proportion of owner-occupied dwellings (correlation 0.352), a high proportion of single-detached houses, the concentration of family households, higher than average rents, a greater distance from the city centre and, higher personal incomes. The other side of this axis has a larger proportion of renters (correlation -0.351), proportionally more other building types, higher dwelling densities and greater proportion of dwellings built before 1920.

While it was not surprising that the inner city/suburb dimension forms the dominant axis within the 30 dimensional space of the original variables, it was a surprise that the second component so clearly points to gentrification. The next most prominent dimension within the data set, perpendicular to the center/suburb axis, shows that the largest rent and income increases were in the neighbourhoods that also had the highest proportion of older buildings in 1981. These older tracts also had higher proportions of older detached houses while high-rise apartments and new buildings define the other end of the axis, the side with the lowest increases in rent and income. The variables identifying the 10 CMAs are not associated with this principal component suggesting that the factors identified by this component are ubiquitous. The gentrification component and the profiles presented later do not pick out the largest CMAs as the places with the most gentrification suggesting that the process takes place in the small as well as the larger cities. This sheds light on initial speculation within the literature that it is not just a phenomena associated with top-ranked, i.e., global cities (Lees 2000, 390; Bailey and Robertson 1997, 562). The correlation coefficients are illustrated in Figure 1 where all the signs have been changed to positive.

Figure 2 helps illustrate the principal component by relating its scores to the proportional change in the average personal income in the Montreal’s census tracts. The scores tend to increases with income and show a large dispersion about the values predicted with a quadratic ordinary least squares regression. Figure 3 relates the scores to the average personal income in each tract for Montreal and shows that high scores were also gained by higher-income neighbourhoods in 1981. Figure 4 shows that Montreal’s low-rent areas experienced more up grading than somewhat high rent areas.2 Figure 5 relates the principal component scores to the proportion of the 2001 stock in each tract that was built since 1981. In Montreal, the slope is positive; the neighbourhoods with more new development 10d to have higher scores. The Toronto profile is flat and the pooled data show a shallow but statistically significant at the 0.01 level U shape. Figure 8 shows that new development plays an important role in the upgrading of old neighbourhoods.

The geographic distribution of the second principal component scores for Montreal, Toronto and Vancouver is illustrated in Figures 6, 7 and 8. A visual inspection of the maps and their comparison to other published maps and reports (Bourne 1993, Criekingen and Decroly 2003, Filion 1991, Germain and Rose 2000 Ley 1988, 1993, 2003, Millward 1988) develop three observations. First, high gentrification scores were noted for a large number of tracts at the periphery of some CMAs due the older ex-urban settlements being surrounded by new suburban development.3 Second, elite-upper income neighbourhoods, such as Westmont in Montreal, Forest Hill/Rosedale in Toronto, and Shaugnessy in Vancouver, recorded high gentrification scores mainly the result of their having a high proportion of older houses and large increases in average tract income. Three, not all of the principal component scores identify the kind of neighbourhood that one typically thinks of as gentrifying.





Share with your friends:
1   2   3   4   5   6   7   8   9   ...   12




The database is protected by copyright ©essaydocs.org 2020
send message

    Main page