Overview

As seen in the Analysis of Neighborhood Changes, Chandler is a unique city with many different neighborhoods. Gentrification may be changing those different neighborhoods, but are Community Land Trusts slowing the gentrification?

To evaluate Community Land Trusts, we will analyze neighborhoods that are not gentrifying in the City of Chandler and then specifically evaluate neighborhoods with Community Land Trusts.

Set-up

We need to set up the data and analysis completed in the Analysis of Neighborhood Changes.

library( dplyr )
library( knitr )
library( pander )
library( stargazer )
library( here )
library( scales )
library( ggplot2 )
library( ggthemes )
library( gridExtra )
library( geojsonio )   
library( sp )          
library( sf )          
library( tmap )        
library( cartogram ) 
library( maptools )

Predicting Change Based on 2000 Neighborhood Characteristics

By analyzing specific 2000 census data, we can predict neighborhoods that won’t gentrify if given the opportunity. First we must establish what consists of a neighborhood that can be a candidate for gentrification using median income. By finding tracts that have a lower 2010 income than the 2000 income adjusted for inflation, we can speculate that the tract might have potential for gentrification.

Next, we need to establish which factors signify whether a neighborhood has been gentrified. The following variables were used for this process:

Less Uneducated: We look for neighborhoods that show a decrease in uneducated people as a signal for gentrification.
Less Poverty: We look at neighborhoods that show a decrease in poverty in the ten year span as an indicator for gentrification.
Less Poverty for Families with Children: We look at neighborhoods that show a decrease in family poverty in the ten year span as an indicator for gentrification.
Less Renters: We look at tracts that have a decrease in renters, singifying that people are more likely moving into housing units instead, showing possible gentrification.
Less Minority: We look at tracts that have a decrease in minority population, signifying that white individuals are moving in and displacing minority individuals.

Create data for Potential Neighborhoods that Gentrified

To analyze gentrification, we need to determine which neighborhoods have the ability to gentrify and then determine which of those eligible neighborhoods did gentrify. These logic variables create a threshold for determining gentrification and then are used to filter for gentrified neighborhoods.

# education status
# poverty
# family poverty
# % renters
# % color

## tracts that did not gentrify

# proportion of college educated decreased
less.educated <- d$p.col.edu.change < 0

# proportion of poverty stayed the same or increased
more.poverty <- d$pov.change >= 0

# proportion of family poverty stayed the same or increased
more.fam.poverty <- d$pov.fam.change >= 0

# proportion of renters stayed the same or increased
more.renters <- d$p.rent.change >= 0

# proportion of minority stayed the same or increased
more.minority <- d$p.color.change >= 0

# Create logic variables that signify what were looking for

# income less than median
lower.income <- d$pay.00 > d$pay.10

g.flag <- lower.income & less.educated & more.poverty & more.fam.poverty & more.renters & more.minority

# Create subsets of the candidates and the gentrified tracts
d_candidates <- d[lower.income,]
d_nogentrified <- d[g.flag,]
d_nogentrified <- na.omit(d_nogentrified)

# Add up how many candidates and how many gentirfied tracts
num.candidates <-  sum( lower.income, na.rm=T )
num.nogentrified <- sum( g.flag, na.rm=T )

Demographics of the Candidate and Gentrified Neighborhoods

Candidate Neighborhoods

Looking at a summary of the candidates for gentrification below, we can see the overall demographics of the neighborhoods. We see a high percentage of individuals who are white that decreases from 2000 to 2010, an increase in education, an increase in renters, and a decrease in pay.

d_candidates <- data.frame(d_candidates)
stargazer( d_candidates, 
           type="html", 
           digits=0, 
           summary.stat = c("min", "p25","median","mean","p75","max") )


Statistic	Min	Pctl(25)	Median	Mean	Pctl(75)	Max

mhv.00	118,547	153,531	176,274	183,273	213,577	278,327
mhv.10	82,800	160,400	203,200	218,623	252,950	421,800
mhv.change	-35,747	11,089	26,952	35,350	57,004	188,057
mhv.growth	-30	7	15	18	28	80
p.white.00	32	69	72	69	77	97
p.black.00	1	3	3	4	4	7
p.hisp.00	2	11	15	21	23	61
p.asian.00	1	2	4	5	6	12
p.native.00	0	1	1	1	2	3
p.white.10	20	55	65	61	69	96
p.black.10	0	4	5	5	7	10
p.hisp.10	2	13	17	23	26	71
p.asian.10	1	5	7	9	11	53
p.native.10	0	1	1	2	2	3
p.color.00	3	23	28	31	31	68
p.color.10	4	31	35	39	45	80
p.col.edu.00	10	23	34	32	40	51
p.col.edu.10	8	28	36	36	46	61
pov.rate.00	0	2	4	6	7	22
pov.rate.10	0	5	8	12	11	85
pov.fam.00	0	1	2	4	5	15
pov.fam.10	0	0	4	6	7	28
p.rent.00	1	13	20	24	33	58
p.rent.10	6	22	33	35	45	83
pay.00	46,229	60,588	77,216	75,635	91,559	108,727
pay.10	37,784	49,621.5	62,964	64,969	78,661	104,559
pop.total.00	180,222	180,222	180,222	180,222	180,222	180,222
pop.total.10	241,278	241,278	241,278	241,278	241,278	241,278
pop.growth	0	0	0	0	0	0
pay.change	-35,120	-14,181	-9,927	-10,666	-5,136	-1,200
p.color.change	-3	6	10	10	12	39
pov.change	-7	1	3	5	7	82
pov.fam.change	-5	-0	2	3	4	14
p.col.edu.change	-15	-1	2	4	7	36
p.rent.change	-6	5	9	12	15	54

Non-Gentrified Neighborhoods

Looking at a summary of the neighborhoods that did not gentrify below, we can see the comparison to the candidate demographics. We see a higher percentage of minority change, higher rate of poverty, higher rate of family poverty, less education, and more renters - all indicators of gentrification not happening.

d_nogentrified <- data.frame(d_nogentrified)
stargazer( d_nogentrified, 
           type="html", 
           digits=0, 
           summary.stat = c("min", "p25","median","mean","p75","max") )


Statistic	Min	Pctl(25)	Median	Mean	Pctl(75)	Max

mhv.00	118,547	148,505	176,274	170,299	188,257	215,832
mhv.10	82,800	159,100	195,100	189,964	218,750	259,500
mhv.change	-35,747	10,595	18,826	19,664	37,854	66,346
mhv.growth	-30	7	12	10	19	34
p.white.00	32	69	72	70	77	83
p.black.00	2	3	4	4	5	6
p.hisp.00	11	12	15	19	19	61
p.asian.00	1	3	4	5	6	10
p.native.00	1	1	1	1	2	2
p.white.10	20	55	65	58	67	71
p.black.10	5	5	7	6	8	8
p.hisp.10	13	16	21	25	25	71
p.asian.10	1	6	8	8	10	13
p.native.10	1	2	2	2	2	3
p.color.00	17	23	28	30	31	68
p.color.10	29	33	35	42	45	80
p.col.edu.00	10	26	35	33	40	51
p.col.edu.10	8	24	30	30	37	51
pov.rate.00	0	2	3	5	5	21
pov.rate.10	4	7	10	11	12	29
pov.fam.00	0	0	2	3	3	15
pov.fam.10	2	5	6	9	8	27
p.rent.00	1	13	30	26	36	50
p.rent.10	14	27	37	39	52	68
pay.00	46,229	64,686	72,481	72,360	81,963	91,760
pay.10	39,085	49,621.5	62,964	60,428	69,846	79,688
pop.total.00	180,222	180,222	180,222	180,222	180,222	180,222
pop.total.10	241,278	241,278	241,278	241,278	241,278	241,278
pop.growth	0	0	0	0	0	0
pay.change	-27,662	-14,129	-10,573	-11,932	-7,030	-3,591
p.color.change	7	10	12	12	13	23
pov.change	2	5	6	6	8	9
pov.fam.change	2	3	4	6	7	12
p.col.edu.change	-6	-4	-2	-3	-1	-0
p.rent.change	3	7	13	13	16	37

There were 39 neighborhoods that were candidates for gentrification. Of these, 11 did not gentrify (approximately 28%).

num.candidates

## [1] 39

num.nogentrified / num.candidates

## [1] 0.2820513

Comparison of Non-Gentrified Neighborhoods and Community Land Trust Neighborhoods

We will now compare neighborhoods where Community Land Trust homes reside to the neighborhoods that did not gentrify to see if there is a correlation between the two groups by uploading the Community Land Trust data, merging, and sorting.

Community Land Trust tracts

Looking at a summary of the tracts that contain Community Land Trust homes below, we can see the overall demographics of the neighborhoods. We see on average a 3% increase in educated invdividuals, a 5% increase in poverty, 4% increase in family poverty, 7% increase in renter, and a 9% increase in minorities. On average, the income from 2000 to 2010 decreases by $7,700.

d_clt <- filter(d_clt, Number.CLT.Houses > 0)

d_clt <- data.frame(d_clt)
stargazer( d_clt, 
           type="html", 
           digits=0, 
           summary.stat = c("min", "p25","median","mean","p75","max") )


Statistic	Min	Pctl(25)	Median	Mean	Pctl(75)	Max

mhv.00	118,547	139,163	150,503	153,069	167,664	198,694
mhv.10	82,800	153,200	172,100	172,189	193,300	249,100
mhv.change	-35,747	13,535	16,914	19,121	27,115	78,320
mhv.growth	-30	9	12	12	18	46
p.white.00	32	53	71	63	76	80
p.black.00	2	3	4	4	4	7
p.hisp.00	11	14	21	27	40	61
p.asian.00	1	2	3	4	6	7
p.native.00	1	1	1	2	2	3
p.white.10	20	42	58	54	67	70
p.black.10	3	5	5	6	7	9
p.hisp.10	14	18	25	33	44	71
p.asian.10	1	4	6	6	7	11
p.native.10	1	1	2	2	2	3
p.color.00	20	24	29	37	47	68
p.color.10	30	33	42	46	58	80
p.col.edu.00	10	19	24	25	34	44
p.col.edu.10	8	20	28	28	33	46
pov.rate.00	1	4	6	9	14	22
pov.rate.10	3	8	10	14	19	34
pov.fam.00	0	2	4	6	8	15
pov.fam.10	2	4	6	10	12	28
p.rent.00	4	16	32	29	39	58
p.rent.10	14	28	34	36	49	57
pay.00	46,229	57,585	62,173	65,768	74,650	95,568
pay.10	39,085	47,106	56,929	58,068	64,019.5	95,673
pop.total.00	180,222	180,222	180,222	180,222	180,222	180,222
pop.total.10	241,278	241,278	241,278	241,278	241,278	241,278
pop.growth	0	0	0	0	0	0
pay.change	-17,901	-10,687	-7,690	-7,700	-4,131	2,902
p.color.change	-3	7	10	9	12	16
pov.change	-0	2	5	5	8	13
pov.fam.change	-1	1	3	4	7	14
p.col.edu.change	-5	-2	2	3	7	13
p.rent.change	-6	4	8	7	10	16
Census.Tract	422,209	522,903.5	523,102	613,214	810,800	811,700
Number.CLT.Houses	1	1	2	3	3.5	15

After merging the candidate tracts with the tracts that contain Community Land Trusts, 16 of the 19 tracts were both candidate tracts and contained Community Land Trust homes. We can take a closer look at those tracts displayed below.

clt_candidates <- merge(clt, d_candidates, by="tractid", na.omit=TRUE)

clt_candidates

Of the 16 census tracts that contain Community Land Trust homes and were candidates, 6 of the tracts were also identified as tracts that did not gentrify. Comparatively, 5 of the tracts that did not contain Community Land Trust homes did not gentrify.

nogent_clt <- merge(clt, d_nogentrified, by="tractid", na.omit=TRUE)

Comparatively to the candidate tracts, non-gentrified tracts with Community Land Trusts had slightly more poverty but less renters. The home value is slightly lower in these non-gentrified tracts, but saw similar rates of growth. The percent of minorities is relatively the same, but the percent of uneducated is higher.

d_nogent_clt <- data.frame(nogent_clt)
stargazer( d_nogent_clt, 
           type="html", 
           digits=0, 
           summary.stat = c("min", "p25","median","mean","p75","max") )


Statistic	Min	Pctl(25)	Median	Mean	Pctl(75)	Max

Census.Tract	422,209	447,390	523,003	585,439	738,950.8	811,300
Number.CLT.Houses	1	1	1.5	2	2.8	6
mhv.00	118,547	137,295	159,587	155,528	176,564	183,361
mhv.10	82,800	152,475	175,400	170,417	207,325	226,100
mhv.change	-35,747	13,593	17,400	14,889	30,761	42,739
mhv.growth	-30	9	11	7	18	23
p.white.00	32	68	72	67	77	80
p.black.00	3	4	4	4	4	6
p.hisp.00	12	13	15	23	20	61
p.asian.00	2	3	4	4	6	7
p.native.00	1	1	1	1	2	2
p.white.10	20	55	61	56	67	70
p.black.10	5	5	6	6	8	8
p.hisp.10	14	18	22	29	26	71
p.asian.10	1	4	8	7	9	11
p.native.10	1	2	2	2	2	3
p.color.00	20	23	28	33	32	68
p.color.10	30	33	39	44	45	80
p.col.edu.00	10	21	26	28	36	44
p.col.edu.10	8	20	24	25	32	39
pov.rate.00	1	2	4	6	5	21
pov.rate.10	6	9	11	13	12	29
pov.fam.00	0	1	2	4	2	15
pov.fam.10	4	6	7	10	8	27
p.rent.00	6	15	33	28	37	50
p.rent.10	14	26	43	38	49	54
pay.00	46,229	59,619	67,028	67,505	76,124	88,436
pay.10	39,085	52,026.2	59,946.5	59,719	69,222	77,634
pop.total.00	180,222	180,222	180,222	180,222	180,222	180,222
pop.total.10	241,278	241,278	241,278	241,278	241,278	241,278
pop.growth	0	0	0	0	0	0
pay.change	-10,802	-9,852	-7,417	-7,786	-6,972	-3,591
p.color.change	7	10	12	11	13	13
pov.change	5	6	7	7	8	9
pov.fam.change	3	4	5	6	7	12
p.col.edu.change	-5	-4	-2	-3	-2	-1
p.rent.change	4	6	9	9	12	16

Comparatively to the candidate tracts, non-gentrified tracts with Community Land Trusts had slightly more poverty but less renters. The home value is slightly lower in these non-gentrified tracts, but saw similar rates of growth. The percent of minorities is relatively the same, but the percent of uneducated is higher.

Non-gentrified, non-Community Land Trust Tracts

We can evaluate the non-gentrified tracts without Community Land Trusts to the non-gentrified tracts with Community Land Trusts to see the impact of the Community Land Trusts in part. The home value change is higher in these non-gentrified tracts as well as the home growth. The poverty rate, family poverty rate, and education are all lower than tracts with Community Land Trusts. However, the proportion of renters and minorities are higher.

d_nogent_noclt <- anti_join(d_nogentrified, nogent_clt, by= "tractid")

d_nogent_noclt <- data.frame(d_nogent_noclt)
stargazer( d_nogent_noclt, 
           type="html", 
           digits=0, 
           summary.stat = c("min", "p25","median","mean","p75","max") )


Statistic	Min	Pctl(25)	Median	Mean	Pctl(75)	Max

mhv.00	154,111	165,708	193,154	188,025	211,322	215,832
mhv.10	162,500	188,000	200,300	213,420	256,800	259,500
mhv.change	-11,022	8,389	22,292	25,395	40,968	66,346
mhv.growth	-5	5	13	13	19	34
p.white.00	70	71	72	74	76	83
p.black.00	2	2	3	4	5	5
p.hisp.00	11	12	14	15	17	21
p.asian.00	1	4	5	5	5	10
p.native.00	1	1	1	2	2	2
p.white.10	48	59	65	62	66	71
p.black.10	5	5	7	6	7	8
p.hisp.10	13	16	19	21	22	35
p.asian.10	5	7	8	9	10	13
p.native.10	1	2	2	2	2	3
p.color.00	17	24	28	26	29	30
p.color.10	29	34	35	38	41	52
p.col.edu.00	31	35	39	39	40	51
p.col.edu.10	29	30	36	37	39	51
pov.rate.00	0	1	2	4	6	11
pov.rate.10	4	4	8	9	10	20
pov.fam.00	0	0	0	2	3.6	5
pov.fam.10	2	3	6	7	7	17
p.rent.00	1	17	18	23	30	50
p.rent.10	17	32	32	41	55	68
pay.00	62,817	72,481	78,946	78,186	84,927	91,760
pay.10	44,819	48,851	64,653	61,279	68,384	79,688
pop.total.00	180,222	180,222	180,222	180,222	180,222	180,222
pop.total.10	241,278	241,278	241,278	241,278	241,278	241,278
pop.growth	0	0	0	0	0	0
pay.change	-27,662	-23,376	-14,293	-16,907	-13,966	-5,239
p.color.change	7	10	12	13	12	23
pov.change	2	2	4	5	8	9
pov.fam.change	2	2	3	5	7	12
p.col.edu.change	-6	-3	-1	-2	-1	-0
p.rent.change	3	15	16	18	18	37

Mapping of Non-Gentrified Tracts, Community Land Trust Tracts, and the Overlap

To view the mapping, we must first upload a mapping package and merge with the selected Chandler data. To see the comparison, we can map the tracts that contain Community Land Trust homes, the non-gentrified tracts, and where the two overlap - the non-gentrified tracts that contain Community Land Trust homes.

# we have Maricopa County already packaged and on GitHub for easy load: 
github.url <- "https://raw.githubusercontent.com/DS4PS/cpp-529-master/master/data/phx_dorling.geojson"
phx <- geojson_read( x=github.url,  what="sp" )

clt_map <- readRDS( here("/data/rodeo/ChandlerCensusMap.rds" ))

data_map_clt <- clt_map
# create small dataframe for the merge
df1 <- data.frame( data_map_clt )

# create GEOID that matches GIS format

# create a geoID for merging by tract 
df1$GEOID <- substr( df1$tractid, 6, 18 )  # extract codes
df1$GEOID <- gsub( "-", "", df1$GEOID )    # remove hyphens

clt_map <- merge( phx, df1, by.x="GEOID", by.y="GEOID" )

clt_map <- spTransform( clt_map, CRS("+init=epsg:3395") )

bb <- st_bbox( c( xmin = -12470000, xmax = -12431368, 
                 ymax = 3925924, ymin = 3880074 ), 
               crs = st_crs("+init=epsg:3395")) 

clt_map1 <-
  tm_shape( clt_map, bbox=bb ) + 
  tm_polygons( col=c("CLT.Homes", "Not.Gentrified", "Nogent.CLT"),
               n=2, ncol=2, style="quantile", palette="Spectral" ) +
  tm_layout( "Dorling Cartogram", title.position=c("right","top"), legend.position = c("left", "bottom") )

clt_map1

Gentrification and Community Land Trusts

Overview

Set-up

Predicting Change Based on 2000 Neighborhood Characteristics

Create data for Potential Neighborhoods that Gentrified

Demographics of the Candidate and Gentrified Neighborhoods

Candidate Neighborhoods

Non-Gentrified Neighborhoods

Comparison of Non-Gentrified Neighborhoods and Community Land Trust Neighborhoods

Community Land Trust tracts

Non-gentrified, non-Community Land Trust Tracts

Mapping of Non-Gentrified Tracts, Community Land Trust Tracts, and the Overlap