Nonprofit Gentrification

I pulled a definition form a paper I like by Lance Freeman and Frank Braconi (2005) where they have a clear definiton of gentrificaiton. They define it as neighborhoods/tracts that

1. Are located in the central city of a metropolitan area;
2. Have a median income lower than the 40th percentile for the median income in the metropolitan area at the beginning of the study period;
3. Have a proportion of housing built within the past 30 years lower2 than the 40th percentile for the metropolitan area at the beginning of the study period;
4. Have an increase in the percentage of university graduates greater than the median increase in university graduates for that metropolitan area between periods;
5. Have an increase in real housing prices.

One of the things I like about that definition is it lets you identify both the tracts that did gentrify, and those that were eligible to gentrify but didn’t. It creates something of a treamtnet/control group to study what changes occur as a result of gentrification.

I constructed that defintion to see whether neighborhoods that did gentrify had more or fewer nonprofits than similarly poor neighborhoods in 2000 that didn’t gentrify.

Study period: 2000-2018

sample: all tracts in 2000 that were eligible to gentrify

• cntnp: count of nonprofits in each tract in 2019
• gentrified: 1 for those tracts that did gentrify between 2018 and 2000
• whtch: change in percentage white in neighborhood 2000-2018
• gini16: gini coefficient in 2018
• pop16: population in 2018

Table continues below

cntnp	gentrified	whtch	gini16
Min. : 0.000	Min. :0.0000	Min. :-0.50998	Min. :0.2491
1st Qu.: 0.000	1st Qu.:0.0000	1st Qu.:-0.02211	1st Qu.:0.4115
Median : 0.000	Median :0.0000	Median : 0.05070	Median :0.4498
Mean : 0.684	Mean :0.4636	Mean : 0.11646	Mean :0.4567
3rd Qu.: 1.000	3rd Qu.:1.0000	3rd Qu.: 0.21422	3rd Qu.:0.4939
Max. :40.000	Max. :1.0000	Max. : 0.98454	Max. :0.7913

pop16
Min. : 107
1st Qu.: 2377
Median : 3357
Mean : 3626
3rd Qu.: 4586
Max. :17871

	Estimate	Std. Error	t value	Pr(>|t|)
(Intercept)	-0.2617	0.03941	-6.641	3.309e-11
gentrified	0.1165	0.01045	11.16	1.062e-28
whtch	-0.1829	0.02286	-8	1.406e-15
gini16	0.9164	0.07945	11.53	1.488e-30
pop16	5.203e-05	3.035e-06	17.14	8.547e-65

Fitting linear model: log(cntnp + 1) ~ gentrified + whtch + gini16 + pop16

Observations	Residual Std. Error	\(R^2\)	Adjusted \(R^2\)
8577	0.4766	0.06029	0.05985

The results make sense, neighborhoods that gentrified had more nonprofits in 2019. Although the white result is somewhat odd to me. The variable stays negative whether I use the 2000 or 2016 % of white residents, so I’m not sure what to make of that.