CPP 528 Project Group 4

Part I - Neighborhood Change

Descriptive Analysis and Prediction of Community Change

Introduction

There are three sections which focus on the exploration of median home value (MHV), operationalizing gentrification, and spatial patterns. While this chapter will garner insights on median home value, it will not exercise the utility of additional variables. Understanding median home value at this level can allow us to understand gentrification as we build our approach for hedonic pricing models and measuring the impact of federal programs as interventions in gentrifiable communities.

Summary Insights:

Through our analysis we find that the MHV data is skewed and by looking at it using a relative change in MHV after removing potential empty tracts, the quality of our insight improves. We see an average of 33% increase in MHV from 2000 to 2010.

After defining a criteria for gentrification we found that ~6% of tracts were gentrified. This is still open to interpretation because we can consider that criteria for gentrfiability by new market tax credit programs as capturing a much larger group of tracts, so an interesting question might be which are being gentrified based on a more rigorous criteria after participating in a program?

After taking a quick look at some spatial patterns, we can see that the concentration of people could be related to the percent change in MHV, this is also open to intepretation. Nonetheless, an important consideration for future analysis.

Overall, this provides good insight into understanding changes in median home value and it’s relationships to gentrification.

Note: This chapter imports multiple functions defined in the utilitiesDescriptiveAnalysis.R file within the respective folder. Please refer to this file for more detail. This chapter will contain plots and tabular views with descriptions on the insight gathered.

Exploration of Median Home Value

First, we can explore the initial conditions in 1990 to 2000: There was a decrease in MHV between 1990 and 2000 of $2,239.

# adjust 2000 home values for inflation 
mhv.90 <- d$mhmval90 * 1.357  
mhv.00 <- d$mhmval00

mhv.change90 <- mhv.00 - mhv.90

df <- data.frame( MedianHomeValue1990=mhv.90, 
                  MedianHomeValue2000=mhv.00, 
                  Change.90.to.00=mhv.change90 )

stargazer( df,  
           type = "html", 
           covariate.labels = c("MHV 1990", "MHV 2000", "MHV Change 1990 to 2000"),
           digits = 0, 
           summary.stat = c("median","mean","min","p25","p75","max"))
Statistic Median Mean Min Pctl(25) Pctl(75) Max
MHV 1990 117,380 152,526 0 79,792 192,423 678,501
MHV 2000 119,900 144,738 0 81,600 173,894 1,000,001
MHV Change 1990 to 2000 \-2,237 \-7,788 \-678,501 \-27,975 16,239 1,000,001

Next, we can explore the initial conditions in 2000 to 2010: There was an increase in MHV between 2000 and 2010 of $36,268. We should also consider the median change of 36k, which is a lot less, but the mean might be higher if we expected the data to be skewed. With minimum home values of 0 and 9,999, we can expect empty land or other cases when home value would be so low.

# adjust 2000 home values for inflation 
mhv.00 <- d$mhmval00 * 1.28855  
mhv.10 <- d$mhmval12

mhv.change <- mhv.10 - mhv.00

df <- data.frame( MedianHomeValue2000=mhv.00, 
                  MedianHomeValue2010=mhv.10, 
                  Change.00.to.10=mhv.change )

stargazer( df,  
           type = "html", 
           covariate.labels = c("MHV 2000", "MHV 2010", "MHV Change 2000 to 2010"),
           digits = 0, 
           summary.stat = c("median","mean","min","p25","p75","max"))
Statistic Median Mean Min Pctl(25) Pctl(75) Max
MHV 2000 154,497 186,502 0 105,146 224,071 1,288,551
MHV 2010 193,200 246,570 9,999 123,200 312,000 1,000,001
MHV Change 2000 to 2010 36,268 60,047 \-1,228,651 7,187 94,881 1,000,001

Histogram of MHV

Creating histograms of the change in median home values from 1990-2000.

#1990-2000
histChangeMVH1(mhv.change90)

Creating histograms of the change in median home values from 2000-2010.

#2000-2010
histChangeMVH(mhv.change)

The distribution of change in MVH from 2000 to 2010 shows a positive skew. Now that we know there isn’t a normal distribution of change, we may want to look at change through a different lense.

Comparing 2000-2010 distributions of MHV.

#dist comparison between 2000-2010
distMVH(mhv.00, mhv.10, "2000", "2010")

The distributions for MHV in 2000 and 2010 are look similar with some slight differences. In 2010, there is a spike around $240,000. Also, there is a wider range of concentration in 2010 hovering $100,000.

The scatterplot is very interesting because it shows more of a relationship between the years. From 2000 to 2010, the relationship shows an increase in MHV, this supports what we’ve already seen - higher home prices in 2010. As data points pass ~$500,000 the line starts to decline. We might consider some data points above the red line and under $10,000 in MHV during 2000 to be problematic since they may have change in MHV that are not part comparable as a counterfactual.

Comparing change in home value by as a relative change in percent.

As we have progressed, two adjustments that can be noted that will allow for better insight:

  1. By using an absolute change in MHV we find skewed distributions and some extreme changes in value. By computing change as a relative measure, it may now make more sense for us to visualization the group which contains various magnitudes in absolute value. Our distributions are expected to change.

  2. If there are additional data points that represent empty lots, or others alike, then we will expect to see very high changes in value relative to the MHV in 2000. By removing these we can find more comparable data for thinking about the counterfactual.

#Remove values in 10k
mhv.00[ mhv.00 < 10000 ] <- NA

#create a percent change in MHV
pct.change <- mhv.change / mhv.00

#Plot 2000-2010
PlotPctChange(pct.change)

Showing changes by percent yields similar results, but still different. The new insight is that there is a 33% change in MHV, or a 25% when considering a skew. The range between median and mean is much smaller in this case, remember we saw almost a 2x in the mean MHV from the median. Removing problematic data may have made a difference in the output with the help of using a relative change in MHV.

Using a relative change is important because we may expect there to be a different absolute change in MHV for a home that is valued at $100,000 vs $300,000 in the year 2000.

Group Growth Rates By Metro Area

As we continue to use a percent change, we can start looking at cities with the highest growth rate from 2000 to 2010.

#2000 to 2010 top 25 cities by growth 
d$mhv.change <- mhv.change 
d$pct.change <- pct.change
d$mhv.10 <- mhv.10
d$mhv.00 <- mhv.00

d %>%
  group_by( cbsaname ) %>%
  summarize( ave.change = median( mhv.change, na.rm=T ),
             ave.change.d = dollar( round(ave.change,0) ),
             growth = 100 * median( pct.change, na.rm=T ) ) %>%
  ungroup() %>%
  arrange( - growth ) %>%
  select( - ave.change ) %>% 
  head( 25 ) %>%
  pander()
cbsaname ave.change.d growth
Ocean City, NJ $154,667 93.07
Virginia Beach-Norfolk-Newport News, VA $97,955 71.81
Casper, WY $70,770 70.03
New York-Wayne-White Plains, NY-NJ $189,118 69.86
Kingston, NY $89,272 65.08
Barnstable Town, MA $146,088 65.07
Washington-Arlington-Alexandria DC-VA $139,136 64.82
Charlottesville, VA $104,487 62.74
Atlan City, NJ $94,239 62.32
Baltimore-Towson, MD $102,918 61.9
Bethesda-Frederick-Gaithersburg, MD $146,639 61.46
San Luis Obispo-Paso Robles, CA $172,722 61.01
Midland, TX $54,576 60.3
Los Angeles-Long Beach-Santa Ana, CA $145,968 60.3
Redding, CA $87,677 59.12
Honolulu, HI $194,665 58.85
Chico, CA $81,746 57.83
Nassau-Suffolk, NY $149,785 57.62
Edison, NJ $125,056 57.19
Santa Ana-Anaheim-Irvine, CA $177,367 56.05
Poughkeepsie-Newburgh-Middletown, NY $101,010 55.18
Flagstaff, AZ $71,072 54.67
Salisbury, MD $60,400 54.06
Winchester, VA-WV $73,165 53.76
Odessa, TX $26,240 53.37

From 2000 to 2010, the highest growth rate by metro area is 93%. Ocean City, NJ nearly doubled in value. If we refer back to the change by tract, we can see changes > 200%. When measuring at the metro area, the changes are more balanced.

There is a bit of a spread across the US, but the east coast seems to be popular in this top 25 listing some of the top areas in New Jersey, Virginia, New York, Mass., and Wyoming.

At this level, we may also be able to understand which cities are experiencing gentrification as a whole. Although, we would still expect certain tracts within a metro to not be experiencing gentrification. This is where we can isolate tracts that may experience gentrification based on their metro level measures.

Operationalizing Gentifrication

To operationalize gentrifcation we can pick certain criteria that would elicit gentrification. Such an analysis would allow us to understand gentrification at high level. E.g. the number of tracts that experienced gentrification from 2000 to 2010.

Let’s focus on the following criteria:

  1. Lower than average home value in metro
  2. Above average diversity for the metro
  3. Change in MHV greater than city
  4. Change in MHV greater than median for country
  5. Loss of diversity
# home value in lower than average home in a metro in 2000
poor.2000 <- d3$metro.mhv.pct.00 < 50  

# above average diversity for metro
diverse.2000 <- d3$metro.race.rank.00 > 50 

# home values increased more than overall city gains 
# change in percentile rank within the metro
mhv.pct.increase <- d3$metro.mhv.pct.change > 0

# faster than average growth  
# 25% growth in value is median for the country
home.val.rise <- d3$pct.change > 25 

# proportion of whites increases by more than 3 percent 
# measured by increase in white
loss.diversity <- d3$race.change > 3 

g.flag <- poor.2000 & diverse.2000 & mhv.pct.increase & home.val.rise & loss.diversity

num.candidates <-  sum( poor.2000 & diverse.2000, na.rm=T )
num.gentrified <- sum( g.flag, na.rm=T )

num.gentrified 

## [1] 1137

num.gentrified / num.candidates

## [1] 0.06240738

With this criteria, we find that 1137 or 6.2% of tracts are likely to have undergone gentrification.

Spatial Patterns

Virginia Beach-Norfolk-Newport News, VA came in #2 on our list, with country and city level criteria we can see how the tracts themselves might be changing in relation to the overall rank.

#plot the shape file to see outline
plot(msp.sp)


#plot dorling cartogram with size as population and color as pct change in MHV
tm_shape( msp_dorling ) + 
  tm_polygons( size="POP", col="pct.change", n=7, style="quantile", palette="Spectral" )

In this example, the tracts are concentrated on the bottom left, but otherwise somewhat spread out across the region. Overall, there is not a bad mix of high and low percent changes in MHV. Since the size of the circle denotes population size, it is clear that some areas in the bottom left contain larger populations than those further away. Those with largest populations tend to have a turqiose color, indicating high changes in MHV, but not the largest. The biggest changes (blue) tend to be smaller circles. In future analysis, we may want to consider the concentration of residents with the change in MHV.

References

*Census Geography: Bridging Data for Census Tracts Across Time*. n.d. Spatial Structures in the Social Sciences, Brown University. <https://s4.ad.brown.edu/Projects/Diversity/Researcher/Bridging.htm>.
*Low-Income Housing Tax Credit (Lihtc)*. n.d. Washington, DC: U.S. Department of Housing; Urban Development. <https://www.huduser.gov/portal/datasets/lihtc.html>.
*New Markets Tax Credit Program*. n.d. U.S. Department of the Treasury, Community Development Financial Institutions Fund. <https://www.cdfifund.gov/programs-training/programs/new-markets-tax-credit>.