Mapping the Housing Crisis
Mapping the Housing Crisis is an interactive research tool used to identify the area of New York City that are most at risk for redevelopment. It was developed as a part of the Rethink the Block thesis team’s research in the Design and Urban Ecologies program at Parsons School of Design in December 2013.
A criteria was established to define the information we wanted to capture that could help us generate a spatial visualization of areas that may be at-risk for speculation and development. This included researching data sets that would provide information on:
- Total Median Income
- Minority Population
- Rent Stabilized Property
- Tax Exemptions
- Percent of Income Allocated Towards Rent
- Proximity to the Subway
Total Median Income
The relationship of the total median income within an area to housing is that income determines where someone can live or qualify to live. In New York, as well as across the United States, housing projects that access some form of public subsidy must develop measurements and formulas that set minimum and maximum income levels as one of the qualifications for housing. In our thesis research we investigated many “affordable housing” initiatives, including the New Housing Marketplace initiative launched by New York City’s outgoing Mayor Michael Bloomberg.
Our investigation revealed that as much as 90% of the housing developed in New York City does not meet the needs of the local community because the income requirements to qualify for a particular development far exceeds the actual medium income of the given community. As a result, we wanted to create a mapping tool that would show what the actual medium income was within particular census tracts. This tool would not only help address growing concerns around whether housing initiatives are really addressing the needs of low income familes but also allow for a more progressive and informed dialogue to occur within the various stages of planning by the different stakeholders.
For example, if a planned affordable housing project stipulates that the minimum and maximum incomes necessary to qualify for the housing is:
But the actual medium income of the area where the project is occurring is only $14,590. This means that those who fall within the $14,590 income bracket who are indigenous to the area would not be able to qualify for the units. Further, New Yorker’s citywide who earn below these minimum and maximum income ranges who are also in dire need of affordable housing-despite the project being labeled as affordable and using subsidies allocated for such-also do not qualify.
Again, the goal of this mapping tool is to raise the bigger question around who qualifies for the affordable housing being built and what happens to those who do not.
Methodology for the Total Medium Income Per Census Tract
In compiling the information for the average median income, we used the guided search option in American Fact Finder. We narrowed down our results to New York City and skipped the option to define race. The data that was then downloaded contained the average over a period of five years.
Before importing the data into QGIS, we cleaned up the document in excel to reflect only the data needed so that the imported .csv file would only have the Total Median Income column. Once imported the data was imported into QGIS, we selected the attribute table option, then toggled into edit mode and used the field calculator to transform that column from a string set into an integer set. Once that transformation was made we used style editor under layer properties to represent the data. Here the graduated symbology was selected, and then changed the mode to natural breaks (Jenks). After trying all of the other modes, we found that natural breaks (Jenks) had a more accurate measure. We used the Upper East Side around 760 Park Avenue, which should fall around the middle section of central park as the estimate to judge whether the visual reflected the correct data- in that the wealthiest of NYC reside in that area.
The reason we decided to map out the Tax Exemptions in the city is because it is representative of critical trends in housing development as it’s indicative of the subsidies being provided by the state to spur construction. By identifying spots where tax exemptions are heavily prevalent and clustered, one is able to gather the status of an area where ‘dispossession’ is occurring or reaching a maximums if it’s heavily clustered. Then comparing these locations with adjacent ones will help us place which areas need help. Adjacent areas with low numbers of exemption indicate that new developments are not being constructed in wide numbers YET; therefore, it becomes important as these areas could now be considered for the potential in speculation. The following methodology indicates how we tried to identify as many properties on the market, without being single family residences (which have a litany of tax exemption programs for energy efficiency), nonprofits, institutions, or government. The only issue with the numbers is that it was difficult to sift out all data concerning religious institutions as they do not have their own Ownership Type in the tax code.
Methodology for the Tax Exemption per Census Tract
List of Built Queries & Reasoning
“ZoneDist1” != ‘PARK’
Exclusion of Parks and Recreation property is necessary as these are common spaces; and, as much as they might evaluate at high assessment values according to ‘retail market prices’ – they cannot be evaluated at a market standpoint.
“ExemptTot” > 1
An exemption total greater than one indicates a property that is receiving a tax abatement (either through 421A, J-51, Inclusionary Housing, or other means), and that exemption total means that is an amount that is unable to be taxable for property tax purposes.
“OwnerName” != ‘NEW YORK CITY HOUSING’
By removing New York City Housing [Authority] from our scope of data,
“OwnerType” = NULL
By removing C, X, O from the type, we have thoroughly eliminated major sources of tax exemption in the tax rolls from government, institutions, or charities. Having OwnerType set to equal null would result in eliminating a great portion of private ownership (i.e. single family homes); but, there could Because the M and X are sources of confusion, we’d need to find other points in the data to further eliminate the potential/possible government or non-profit owners within these categories.
- C – City Ownership
- M – Mixed City & Private Ownership
- O – Other
- Public Authority, State or Federal Ownership
- P – Private Ownership
- Either the tax lot has started an “in rem” action or it was once city owned.
- X – Mixed (Excludes property with a C, M, O, or P ownership code)
- Fully tax exempt property that could be owned by the city, state, or federal government; a public authority; or a private institution.
- NULL – Unknown (Usually Private Ownership)
“LandUse” != ‘01’
Removing the Land Use 1 in the query will leave off Tax Class 1 properties from the spread. This is important as taxation based on these lots are done on a purely market value and assessment system, unlike the income generators found in condo/rental buildings and commercial office space.
“ZoneDist1” != ‘PARK’ AND “ExemptTot” > 1 AND “OwnerName” != ‘NEW YORK CITY HOUSING’ AND “OwnerType” = NULL AND “LandUse” != ‘01’
We were interested in collecting data that had to do with the demographics of the city and see how it related with the larger issue of housing affordability in New York City. We found that housing affordability in certain neighborhoods is an indicator of the types of demographics that live in these areas. Neighborhoods, like Bushwick, that went through a major economic disinvestment that made the valuable extremely cheap brought an influx of immigrant and marginalized demographics to the neighborhood. The white population is mostly concentrated in areas with higher median income while the areas of the city that have a dense concentration of black, Hispanic, or other demographics tend to have a lower median income. When looking at the map in relation to the data set that indicates the percent of income that goes towards rent we found that in areas like Chinatown, Washington Heights and East Harlem, where the population is mainly minorities, they pay more than 60% of the income towards rent. Through our research we have found that these areas with a high concentration of a Blacks or Hispanics, like areas in Harlem and Brooklyn are quickly being speculated, developed and ultimately gentrified forcing the minority demographic out of the neighborhood.
In addition, this component of our research can be used to identify demographic changes overtime to build a storyline if certain indicators actualize within a neighbor where development is occurring.
Methodology for the Minority Population Per Census Tract
For the map on percentage of minority population per census tract we acquired the basic shape file from the “Bytes of the Big Apple” website, which is run by the department of City Planning, City of New York. The site is set to assist both government agencies and the public by providing policy analysis and technical assistance relating to housing, transportation, community facilities, demography, waterfront and public space. Further we got the data for the minority population from the “American FactFinder” website, which provides the data about the United States, Puerto Rico and the Island Areas. The data from the “American FactFinder” come from several censuses and surveys.
For our minority population map we searched specific data on minority population and created a table through guided search. As the topics we selected race and ethnicity, (we had to select each race separately) and then proceeded to selecting geographic type census tract, (the geographic location for our map was the State of New York), then we selected each of the five boroughs (Manhattan, Bronx, Brooklyn, Queens, and Staten Island). The next step was to select ACS demographic and housing estimates, and modifying the table by unselecting the categories we didn’t need, for example sex and age. After we selected the desired information we downloaded the table. Further we opened the annotated version of CSV table, and opened it in Excel and further edited it (we erased several columns, for example Margin Errors and Percents, and left just the total population in order to make the file lighter and to work just with the data that we needed in order to build our map). We saved the table in Excel as CSV file. To spatialize the minority demographic data we opened in QGIS the NYC census tract shape file that we acquired from the Bytes of the Big Apple. Then we added the layer with the minority demographic information and joined them in properties window. Next step was to visualize the data using natural breaks on a graduate scale, showing the census tract with the highest percentage darkest and the lowest percentage lightest. We created 11 levels on the graduate scale, with the lowest being 0 and highest 100. Based on selected levels we created ranking system with the highest level receiving value 10 and lowest value 1. After that we created a new column in the data set, and used the advanced search to select only the census tract that fell within each of the breaks in the graduate scale. After we used QGIS field calculator to add ranking to a new column named “RANK”.
After the ranking we compiled for all 10 levels, we finalized visualization the data on a graduated scale using the new RANK column and were able to show the 10 as the darkest and 1 as the lightest.
Level 1: 0
Level 2: 0 – 10.3
Level 3: 10.3 – 21.7
Level 4: 21.7 – 31.7
Level 5: 31.7 – 42.5
Level 6: 42.5 – 52.4
Level 7: 52.4 – 63.0
Level 8: 63.0 – 74.9
Level 9: 74.9 – 86.2
Level 10: 86.9 – 94.8
Level 11: 94.8-100%
Rent Stabilized Properties
There are a certain amount of apartments in New York City that are rent stabilized under certain programs like the 421a of J51. We thought it was be very valuable to be able to find a data set that had information on where these properties where in the five boroughs of the city. It was extremely interesting to look at the data set in relation to the other data sets we collected in relation to housing affordability. Not only did we find that there are very few areas of the city that have rent stabilized properties, we found that the most dense concentration of these were in Manhattan. We found that in East Harlem and in the Lower East Side, where buildings are older in age have more rent stabilized properties than industrial areas in Brooklyn or areas in Midtown Manhattan. While people who live in rent stabilized units might be comfortable for a while, these exemptions do not last forever; once the exemption expires, these properties have a high probability of being redeveloped.
Methodology for the Rent Stabilized Property Per Census Tract
We did an extensive online search of websites that could provide data on rent-stabilized units in New York City. We were not able to find a data set that identified units but were able to find one that had the properties listed; we found it on http://www.housingnyc.com/html/resources/zip.html#tables. One the site they have a pdf file for all five boroughs that list rent stabilized buildings in the five boroughs. We downloaded the five files and reviewed them to make sure they had addresses we could use to then map on QGIS. To be able to make use of our data we had to download a free converter online to transform our pdf into an excel format. One we completed this step we had to go into the excel files and organize the columns to make sure that everything was properly aligned and in the correct column. We organized the information into columns that included their zip code, street address, and type of structure. The data set also included there status and where or not the properties had some sort of type exemption. After the data files were organized, we saved each file according to their borough in a csv format.
In order to visualize, we first attempted to geocode the addresses, but were not getting great results. Instead, we decided to use the BBL method. Since the data sets included the Borough name, Block and Lot, we were able to process the data in Open Refine, to create a BBL column that we could then join with the MapPLUTO data for each of the 5 boroughs. In order to calculate percentages within census tracts, we converted these into polygon centroids, and overlayed a clipped census tract shapefile, which could be used to conduct a Points in Polygon function. Once we had a count of the properties, we also repeated these steps for all of the MapPLUTO lots so that we could get a percentage.
To create a ranking, census tracts that had the highest percentages of rent regulated properties were given a 10 and the lowest percentage 0 and 1 along a graduated scale. This was represented visually with darker colors going to the 10s and lighter colors the 1s, etc.
Proximity to the Subway
Another aspect we wanted to consider when constructing our final maps was the proximity of properties to the subway. We looked at properties that were within a five-minute walk from the subway. In cities one of the major drivers of development is the subway system. It is expensive infrastructure that is commonly built by public funds and is meant to transport large quantities of people to work and around cities. It is interesting to see how the system drive development in certain areas of the city. Private developers are attracted to these areas because it allows for convenient travel for their employees. A lot of private developers look at the subway system as an indicator for where they can potentially develop; areas that are within a 5 minute walking distance from the subway tend to have a higher property value because of the easy access to transportation. These areas will also tend to have a population of people with a higher income.
Methodology for the Proximity To The Subway Per Census Tract
It is well known that apartments that are located the closest to the Subway are typically more expensive than apartments that are further away. Therefore, we suspected that properties that the closest to the Subway system were the most likely to get redeveloped. In order to find this data, we used the MapPLUTO tax lot shapefile from Bytes of the Big Apple for each of the boroughs, minus Staten Island, which does not have a direct connection to the Subway system (we regarded the undesirable commute by car or the Staten Island Ferry to automatically rank the entire island a 0). Additionally, we used the subway entrance point shapefile from the MTA and via the Spatiality Blog to create a buffer of tax lots that were within a distance of 1320 feet (straight-line radius) or a 5-minute walk and also within a distance of 2640 feet (straight-line radius) or a 10-minute walk, creating new shapefile layers for each of the selections. We also need to subtract the 5-minute walk lots from the 10-minute walk lots because we did not them included.
Since, we were using census tracts as our common geography it was necessary to calculate the number of properties in the given census tracts. In order to do that, we converted both of the 5-minte and 10-minute walk layers to polygon centroids so that we had point data that we can take a count of. Bringing in an already clipped census tract file, we used the Points in a Polygon tool to create a column with the total numbers for each layer. Also, it was necessary to create centroids and do a count for the entire MapPLUTO data set so that we can easily take an average. Lastly, we had to merge all of the separate borough layers together to that we can easily create a ranking for the entire city.
To create the ranking we assigned census tracts that had 80%-100% of their lots within a 5-minute walk a 10 and 0 – 20% of their tax lots a 6 (and everything in between). For rankings 5-1, we used the 10-minute walk data where 80%-100% of tax lots within a census tract AND the percentage within a 5-minute walk was 0 were given a 5. Here it was important that the percentage that were within a 5-minute walk 0, because it meant that these were further away from subway entrances. Census tracts that had neither 10 or 5-minute walk properties were given 0.
Like the other maps, this was visualized where 10s were given the darkest red, and 0 were white. The resulting map looked very familiar, as it was somewhat reminiscent of the Subway map that we see every day.
Percent of Income Allocated Towards Rent
To discover which parts of New York City suffer from the greatest strain in paying rent, we looked to the latest census data at American FactFinder, which has a data set available that shows the percentage of income going toward rent on a census tract level. We felt that this data could be useful in showing areas of the city where average income has not risen fast enough with the average rental rates. We suspect that these areas are also good representations of areas where the population is on the brink of displacement due to rising rental costs and signifying a rapid investment in redevelopment.
According to HUD, in order to maintain a high standard of living, they recommend that households pay no more than 30% of their income toward rent. Recognizing that the 30% threshold is not realistic, even in the highest income census tracts, we decided to focus on a 40% threshold to show this strain.
Methodology for the Percent Of Income Allocated Towards Rent
The census data is broken down into several percentile levels, the highest two being 40%-50% and 50% or more. The data is also broken down into total number of households, and the percentage is derived from the number of households who pay within the given threshold and the total number of households within the given census tracts. In order to get a combined 40% or more total, first we had to calculate the sum of the population who pay 40%-50% and then divide that number by the total population to achieve a new percentage.
To spatialize the data, we imported the revised CSV data set into QGIS and joined it with the census tract shapefile from NYS LION. Additionally, we used the borough boundaries shapefile from Bytes of the Big Apple to clip the census tract shapefile to the shoreline of the city. Then we visualized the data using natural breaks on a graduated scale, showing the census tracts with the highest percentage darkest. We decided to create 11 levels on the graduated scale, with the lowest being completed 0. This decision facilitated the creation of ranking system where the highest percentages would receive a 10 and the lowest a 1. We created a new column in the data set, and used the advanced search to select only the census tracts that fell within each of the breaks in the graduated scale. Once they were selected, we were then able to use the field calculator in QGIS to add the ranking to a new column called “RANK.”
After the rankings were complied for all 10 levels, we then re-visualized the data on a graduated scale using the new “RANK” column and were able to show the 10s as the darkest and the 0 and 1s as the lightest.
This data set showed an unsurprising revelation in that the areas of the city where people pay more than 40% of their income toward rent are in the areas the are deemed as up and coming or gentrifying neighborhoods, such as Sunset Park, Bed-Stuy, Bushwick, and areas of Queens where there are high numbers of minority populations.
As we begin to compare the ranking for each data set, we were able to produce a map of the city that identifies the nodes we found have the highest possibility for displacement because of speculation and ultimately gentrification of the area.
We began to produce maps using data such as neighborhood demographics, average median income, occupancy types, rent controlled properties, etc. The creation of these various maps was part of our methodology for the final project. Each map ranks the city census track and according to how high or low their rank is signifies their potential for re-development.
The maps we created will contribute extensively to our thesis research. They are breaking down the complex data and facts that highlight the various reasons for housing inequality and housing unaffordability in New York City. The data that we found is particularly important in disclosing the inequalities and supporting our claims with analytical data.
Click Here to go to the Project Site