Measuring Gentrification

I have been puzzling over how to measure gentrification quantitatively. Drawing on the work of geographer Elvin Wyly and others who have used disciminant analysis to compare qualitative investigations of gentrified tracts with Census data, I look at 5 Census variables that have been shown to best "discriminate" between gentrified and non-gentrified tracts. These are: change in median household income, change in median rent, change in median house value, change in proportion of residents who have a BA, and change in proportion of residents working in "high-status" (managerial, professional, or technical) occupations.

I investigate how these have changed for all central-city Census tracts from 1990 to 2000. I use a relative measure of change--has the Census tract experienced an increase in each indicator that is greater than the average increase for the city in which the tract is located? A principal components analysis of changes in these 5 variables results in two "types" of gentrification indicators: (1) Changes in Capital: changes in median household income, rent, and house value; and (2) Changes in Status: changes in proportion of residents with a BA or working in high status jobs. Based on these two typologies and descriptive statistics on the changes from 1990 to 2000, I created thresholds for gentrification: (1) An increase that is 10 percentage points higher than the average change for the city in ALL three capital variables; and (2) An increase that is 5 percentage points higher than the average change for the city for BOTH status variables. From 1990 to 2000 in the United States, about 15% of all central-city tracts in the U.S. experienced gentrification based on changes in capital, about 8% experienced gentrification based on changes in status, and about 3% experienced gentrification based on both dimensions (i.e., they experienced changes above the thresholds for all 5 variables).

To see how these indicators stacked up, I mapped them for Boston. Non-shaded tracts experienced no gentrification, the lightest gray indicates gentrification based on capital, the medium gray indicates gentrification based on status, and the dark gray indicates gentrification based on both measures. I also included boundary lines for the Boston neighborhoods. Given my knowledge of Boston from 1990-2000, this map looks pretty good to me--we see the gentrification of Jamaica Pond, East Southie, the creation of the mixed income community of Harbor Point in Dorchester, and action in Allston-Brighton and Charlestown. I haven't investigated every tract--and perhaps being consistent with locally understood gentrification is less important than sticking with a defined theoretical concept (social upgrading of residents). But I think I've made some progress in measuring gentrification with Census data.