Mapping Gentrification in Mexico City Using Open Data
As in many cities around the world, gentrification is a key issue in Mexico City. In recent years, the city has fallen prey to a construction boom and the cost of living has greatly increased. This building boom has gone hand in hand with the rebranding of the city as a trendy tourist destination and the rise of Airbnb. This project aims to map risks of gentrification in Mexico City neighborhoods and identify what areas are most at risk within the city. We will center on the following question: Can a series of spatial variables help us determine gentrification risk? What factors make a neighborhood desirable for gentrifiers? What factors put residents at risk of being priced out?
Background
Mexico City has fallen prey to rampant gentrification over the past decade. Million-dollar condos on the market, unaffordable for the population that earns less than $350 USD a month, have become commonplace in the city’s real-estate market. In addition, the government has made an effort to rebrand the city as a trendy global destination, hoping to attract tourists and expats at the expense of poorer residents who are being pushed out of the central areas. The rise in property value has created conflict between residents and building owners, and has fostered forced evictions.
Key to gentrification in the city are two factors. The first of these is the building boom in the city. Real-estate speculation and corruption have led to mass construction of luxury developments or megaprojects within central areas of the city. Thus, new construction can be seen as a proxy for gentrification, as data on recent construction can help us see where the construction boom is occurring. Likewise, Airbnb has been blamed for hastening gentrification within the city, with developers increasingly converting buildings exclusively into Airbnbs, driving out long-term tenants. In addition, another common trait associated with gentrifying neighborhoods is the construction of sites for bike sharing services, which often prioritizes gentrifying areas. A study conducted by the city of Seattle determined that in conjunction to variables related to displacement and demographic change, it is important to keep in mind the desirability of a neighborhood as something that could potentially foster gentrification. These factors to keep in mind include transit access and amenities such as parks, cultural spaces, and others.
Thus, this project will use two types of data: elements that make a neighborhood more desirable to speculators and gentrifiers, and new buildings/Airbnb listings as proxies for gentrification processes.
Data Sources
Data will mostly come from Mexico City’s open data portal, datos.cdmx.gob.mx. The neighborhood and borough maps are from the portal, as are the bikeshare, parks, bike lanes, metro, and metrobus files. Data on new buildings and property values comes from the government’s sig.cdmx.gob.mx site. In addition, we used an Airbnb dataset from InsideAirbnb, which pulls from the site’s API, created by Australian data scientist and activist Murray Cox. Other data will come from Open Streets Map and will be obtained using the Quick OSM plugin. In addition, the satellite imagery comes from ESRI QMS.
Methodology
Multi-Criteria Evaluation is a GIS methodology that helps determine suitability based on the presence of certain attributes within a given project area. This project will use Multi-Criteria Evaluation (MCE) suitability analysis to determine which areas in a determined section of Mexico City are gentrifying or susceptible to gentrification. Boolean overlay will be used to rank areas’ susceptibility to gentrification within a determined region of Mexico City. Each criteria will be valued at one point and will include the following:
Determining gentrifiability (spatial variables)
Proximity to public transit
Proximity to bike infrastructure (bike-shares and bike lanes)
Proximity to parks
Proximity to cultural spaces (theaters and museums)
Proximity to nightlife (bars)
Proxies for gentrification (logical variables)
Presence of new buildings
Presence of Airbnb
STEP 1: Preliminary Spatial Analysis: Defining the Project Area
Using satellite imagery (ESRI Satellite), we determine that we can exclude farmland, forest, and other rural areas that fall within the city bounds from our analysis. We will also exclude boroughs that fall beyond the bounds of the metro and metrobus systems, as these tend to be either more suburban areas or informal settlements. According to the Social Development Index, the following boroughs score highest: Azcapotzalco, Coyoacán, Venustiano Carranza, Benito Juárez, Cuauhtémoc, and Miguel Hidalgo. Thus our project area will encompass these boroughs.
STEP 2: Preparing the Data
New Buildings
For this, we use the city’s “catastro” datasets. These datasets show information for all the buildings in a borough, including value, year built, and land use. The datasets are all separated by borough, so this requires merging them onto a single layer. We then select by attributes (year>2009) to obtain only buildings in the past decade and export.
Airbnbs
The Mexico City csv from InsideAirbnb is loaded onto QGIS. We then clip to the project area and select by attributes to obtain only listings from 2019 and 2020.
Transit Infrastructure: Metro, Metrobus, Bike lanes and Bikeshares
The shapefiles and csvs from the city open data portal are loaded onto QGIS and then clipped onto the project area.
Parks
We download the city’s green spaces shapefile. Firstly, we clip it to the project area. This dataset contains every green space so it needs to be narrowed down. Thus, the second step is to select by attributes to obtain only parks.
Museums and Theaters
These are obtained by running two queries through QuickOSM, as the data is not on the government data portal. As the data is from OpenStreetsMap, it might have things missing. The queries generate two temporary points layers. These are clipped and exported.
All data has been reprojected to the Mexico ITRF92/UTM Zone 14N CRS in order to ensure accuracy of analysis.
STEP 3: Preliminary Thematic Mapping
We began by looking at several variables individually, some as points maps and others as choropleth maps using the “count points per polygon” feature to see which neighborhoods our variables are concentrated in before initiating the overlay analysis. The Natural Breaks (Jenks) classification is used for each of these choropleth maps. We found that areas corresponded with one another, for the most part. The areas where construction was happening were mostly the same areas where Airbnb listings were more prevalent. Amenities like cultural spaces and nightlife also correspond with these areas. Likewise, bike shares seem concentrated exclusively within areas with high numbers of new buildings and Airbnb listings, giving credence to the idea of a likely relationship between gentrification and the presence of bike shares.
STEP 4: Overlay analysis
We will conduct a Raster based overlay analysis based on Boolean Logic to determine susceptibility to gentrification using a series of seven conditions to be met. Each condition will have a value of 1 and will sum up to a total score of 7. The conditions are as follows:
Conditions:
500m from public transit (Metro/Metrobus) (1 point)
500m from a bikeshare OR 250m from a bike lane (1 point)
500m from a park (1 point)
500m from museums and theaters (1 point)
500m from bars (1 point)
Large (>20) presence of new buildings in neighborhood (1 point)
Large (>20) presence of Airbnbs in neighborhood (1 point)
For conditions 1–5, we create buffers with the relevant distance. For conditions 6–7, we select by attributes to create layers that only include the neighborhoods that meet these criteria. Each of these is dissolved into a single polygon and rasterized. The raster layers are added up using the raster calculator as follows:
airbnbs@1+bikeinfrastructure@1+culturalspaces@1+newbuildings@1+nightlife@1+parks@1+publictransit@1
The map produced shows the most susceptible areas concentrating in a central corridor along the Benito Juárez and Cuauhtémoc boroughs, as well as some bordering areas in Coyoacán, Miguel Hidalgo, and Venustiano Carranza. After vectorizing and clipping, we found that 135 neighborhoods contain areas with a susceptibility rating of 5 and over. The area in square meters is 34857317.92242432, or 15.372% of the total project area. A flaw in the methodology is that we have not accounted for socioeconomic factors. Thus, from these specific datasets we do not know the extent to which low-income populations are being driven out.
Of the neighborhoods that are susceptible (score>5), we calculate the area that is susceptible and divide it by the total neighborhood area. We then create a thematic map to show which neighborhoods have the highest percentage of area that is susceptible. There is a visible contiguous corridor of neighborhoods with a high percentage of susceptible areas along the central part of our project area, with some pockets extending outward. The areas in between this corridor and pockets might start to see more gentrification in the coming years. Around half of all neighborhoods in the project area have either no data or less than 20% as shown by the histogram below:
Findings
Gentrifying neighborhoods are clearly concentrated in the central part of the city. The presence of new buildings and Airbnbs seems to match up to areas with better access to public transit, bike infrastructure, and amenities. Ultimately, a likely relationship needs to be verified statistically in addition to spatially. A multilinear regression model could be built using these variables to establish correlation and predict future gentrification.
Additionally, more variables relating to displacement and socioeconomic risk could be added to the model. To what extent has gentrification in these areas displaced poorer residents? How much have rents increased or decreased? How have demographics changed? Comparisons evaluating poipulational differences between the 2010 and 2020 censuses once the latter is out will be enlightening in this regard.
Bibliography
Becerril, Josemaría (.2017). “Sobre la gentrificación de la colonia Juárez a 111 años de su surgimiento.” Nexos:
https://labrujula.nexos.com.mx/?p=1339
De Coss, Alejandro (2017). “La nueva Ciudad de México: desigualdad, capital y mercadotecnia.” Nexos:
https://labrujula.nexos.com.mx/?p=1449
Grabinsky, Alan (2016). “CDMX Inc.” Horizontal:
https://horizontal.mx/cdmx-inc/
Gil Olmos, José (2017) “Los cimientos podridos del boom inmobiliario” Proceso:
https://www.proceso.com.mx/reportajes/2017/9/26/los-cimientos-podridos-del-boom-inmobiliario-192149.html
Pearson, Tamara (2019) “Airbnb Rentals Are Displacing Mexico City Residents as Rents Surge” Truthout:
https://truthout.org/articles/airbnb-rentals-are-displacing-mexico-city-residents-as-rents-surge/
Leahy, Colleen (2018) “Bike-Share Programs: Two-Wheeled Symbols For Gentrification” WPR:
https://www.wpr.org/bike-share-programs-two-wheeled-symbols-gentrification
Government of Seattle (2016) “Analyzing Impacts on Displacement and Opportunity
Related to Seattle’s Growth Strategy”:
https://www.seattle.gov/Documents/Departments/OPCD/OngoingInitiatives/SeattlesComprehensivePlan/FinalGrowthandEquityAnalysis.pdf
