Background

According to the U.S. Green Building Council (USGBC), Leadership in Energy and Environmental Design (LEED) is a green building certification program that evaluates the extent to which buildings are designed to be energy efficient. There are six LEED certification programs for different sectors, which include:

  • LEED for Building Design & Construction
  • LEED for Interior Design & Construction
  • LEED for Building Operations & Maintenance
  • LEED for Neighborhood Development
  • LEED for Homes
  • LEED for Cities & Communities

For this project, we will look at LEED for buildings specifically. Based on the USGBC’s evaluation, buildings can earn one of four LEED certification levels: CERTIFIED (40–49 points), SILVER (50–59 points), GOLD (60–79 points), or PLATINUM (80 points or more). Also, out of USGBC’s seven evaluation criteria for LEED certification, the following carry the most weight:

  • Sustainable Sites (26 points or 23.6%): Whether a project encourages interaction with the environment through physical activity, passive recreation, restoration of habitats, and reduction of rainwater runoff.
  • Energy and Atmosphere (35 points or 31.8%): Whether a project demonstrates a commitment to higher energy performance based on cost and greenhouse gas emissions.

Accordingly, governments at various levels have focused on incentivizing the private and public sectors to pursue LEED certification. In the case of the City of Atlanta, the following policies and programs encourage developers to make their buildings LEED certified:

  • Atlanta Better Buildings Challenge: As part of the Biden Administration challenge, many building owners in the city sought LEED certification.
  • Commercial Buildings Energy Efficiency Ordinance: Required buildings over 25,000 sqft to achieve certain energy and water usage benchmarks that align with LEED certification requirements.
  • Ongoing Municipal Policy: City-owned properties larger than 5,000 sqft are required to obtain LEED Silver certification for new construction and major renovation; City-owned properties greater than 25,000 square feet required to obtain LEED Certification for Existing Buildings.

USGBC argues that “LEED certification provides a framework for healthy, highly efficient, and cost-saving green buildings, which offer environmental, social and governance benefits.” Through these benefits, LEED certification is intended to enhance community resilience and promote equitable communities, highlighting its focus on the broader concept of sustainability, which includes not only environmental but also social and economic dimensions. As Purvis et al. (2019) argue, sustainability is a synthesis of three interconnected pillars: social, environmental, and economic sustainability. To achieve these three pillars, we need to better understand how building certification programs like LEED influence all aspects of sustainability, especially in the context of urban development and community resilience. While LEED emphasizes benefits such as energy efficiency, water conservation, and reduced carbon footprints, important questions remain about its impact on social and economic sustainability. How do LEED-certified buildings affect the social well-being of their communities, such as enhancing green placemaking or walkability? Are these buildings accessible to all residents? Do they contribute to gentrification of the neighborhoods? Several studies have explored these questions and summarized their findings as follows:

  • LEED has a central nexus in sustainable planning strategies (Szibbo, 2016; James, 2016). Governments provide tax/non-tax incentives to encourage green building practices, which are closely related to LEED (Brown et al., 2002; Shannon et al., 2008.)
  • Matisoff et. al (2015) explains that, ‘firms earn higher LEED scores to achieve a higher certification and provide a greener signal to stakeholders, indicating the presence of competition in green building; buildings certified just above the highest thresholds cluster spatially’.
  • Chen and Gou (2023) claim that the spatiotemporal distribution of green-certified buildings depends upon ‘demographic, socioeconomic, environmental, and policymaking factors’ and that ‘expedited permitting, reduced fees, and property tax credit or exemption are significant policy instruments that promote the implementation of LEED certified projects.’
  • Elzeyadi (2015) consider the ‘impact on the physical environment in terms of street networks, zoning, and walkability indicators have on…commuting behavior in LEED certified schools.’

Research Question & Data

Given the intentions of the USGBC to promote sustainable communities through LEED certification and the incentives provided by the City of Atlanta to increase LEED certification in the city, this project aims to explore the following research question: What is the relationship between LEED buildings and sustainability in Atlanta, GA? Specifically, the project seeks to answer the below questions related to the social and economic aspects of LEED certification:

  • Aim 1: Are LEED buildings distributed equitably across the city? Is there evidence of green signaling?
  • Aim 2: Are areas of the city with LEED buildings greener, in terms of tree canopy and walkability/bikeability?
  • Aim 3: What impact does LEED certification have on the affordability of the buildings?

To answer these questions, the project obtained various data from the below sources:

  • LEED Project Database by USGBC: List of all publicly available LEED-certified Projects in Atlanta, GA
  • ACS 2020 5-year Estimates by US Census: Socio-demographic data by census tract
  • Vegetation Data by GT Urban Climate Lab: Vegetation (trees and other greenery) from aerial imaging
  • Walk Score and Bike Score API by Walk Score Professional: walkability and bikeability score by census tract
  • Fulton County Tax Assessor’s Data 2015-2024: information on property values, and other characteristics by parcel

Aim 1. Profile, Distribution, and Green Signaling

First, the profile of LEED-certified buildings in Atlanta and their distribution across census tracts are analyzed.

1.1. Data Preparation

To conduct this analysis, we cleaned and geocoded the LEED dataset for all publicly available LEED certified buildings in the City of Atlanta. We then pulled Census data for the city using tidycensus and tigris packages. Finally, we spatially joined the LEED data with the census data, counted the number of LEED buildings in each census tract, and joined the LEED building count with the census demographic data.

1.2. Profile of LEED certified buildings in Atlanta

The LEED Project Database by USGBC reveals that a significant number of LEED-certified buildings in Atlanta lack project type data. Among those with available information, the most common project type is “Office.” Additionally, based on visual inspection, most LEED-certified buildings are concentrated in Downtown, Midtown, and Buckhead. This suggests that many LEED-certified projects without specified project type information are also likely to be office buildings.

1.3. Distribution and Green signaling of LEED certified buildings in Atlanta

Next, out of all the census tracts in Atlanta, 81 census tracts have LEED projects, while 133 census tracts do not have any LEED projects.Those that are distributed outside of Downtown, Midtown, and Buckead appear to be mostly schools and other public buildings.

As shown in the figure below, LEED-certified buildings are more concentrated in Downtown, where a high proportion of office buildings is present. This clustering suggests the possibility of green signaling whereby office building owners compete for business by improving sustainability. This is also indicative of the green building initiatives put forward by the City of Atlanta, incentivizing or mandating large buildings to become LEED certified.

Interestingly, the maps and figures below reveal no clear relationship between the number of LEED-certified buildings in a census tract and either median household income or the percentage of non-white residents. We presume this is because most LEED-certified buildings are office spaces, meaning the census data primarily reflects residential areas not captured by these developments.


Using a linear regression, no relationship is observed between the median household income of a census tract and the number of LEED-certified buildings within the census tract.

Using a linear regression, there is also no relationship between the percentage of non-white residents in a census tract and the number of LEED-certified buildings within it.

Aim 2. Greenness

Next, the relationship between the number of LEED projects in a census tract and the extent of greenness within those tracts is examined.

2.1. Vegetation

To conduct this analysis, we first spatially joined the LEED and Census data to the vegetation data, which includes the percentage of a variety of land cover types in each 100m x 100m grid cell across the city. For each grid cell, we used the percentage of vegetation (trees, grasses, and other greenery). We averaged the vegetation across each census tract, giving us the mean vegetation percentage for the census tract. The map below illustrates the average vegetation coverage in each census tract, along with the locations of LEED-certified buildings. Interestingly, the average vegetation in census tracts with LEED buildings is slightly lower than in those tracts without LEED buildings. Additionally, the percentage of vegetation appears to be more variable in tracts without LEED buildings.

The figure below shows there significant negative association between the percentage of vegetation in a census tract and the number of LEED buildings in that tract. As the percentage of vegetation increases in a census tract, the number of LEED buildings decreases. This relationship was expected, as most LEED-certified buildings are clustered in the highest-density areas of Atlanta, where vegetation is typically less prevalent.

2.2. Walk Score

Next, walk scores of each census tract and the extent to which they correlate with the number of LEED projects in those tracts are examined. To conduct this analysis, we first pulled the centroids of each census tract. We used this data frame to pull Walkscores for each census tract centroid, as the API provide Walkscores for each longitude/latitude and does not pull mutliple at the same time. The map below shows the walk score of each census tract and the locations of LEED buildings.

The figure below shows that the walk scores of census tracts with a higher number of LEED buildings are higher than those of census tracts with fewer LEED buildings.

2.3. Bike Score

Moving forward, bike scores of each census tract and the extent to which they correlate with the number of LEED projects in those tracts are examined. We used the same methods described above to pull Bikescores for each census tract centroid. The map below shows the bike score of each census tract and the locations of LEED buildings.

Similar to the result from the WalkScore and LEED project analysis, the figure below shows that the bike scores of census tracts with a higher number of LEED buildings are higher than the bike scores for census tracts with fewer LEED buildings. This suggests that, like walkability, bikeability is also positively correlated with the concentration of LEED-certified buildings.

2.4. Overall Greenness

Finally, a regression analysis of the included greenness factors (vegetation, walk score, and bike score) and the number of LEED buildings in each census tract does not show a significant relationship. This is likely due to the fact that the association between vegetation and LEED building count is negative, while the associations between walk & bike score and the number of LEED buildings are positive.
  Dependent = Number of LEED Buildings
Predictors Estimates std. Error CI p
(Intercept) -0.52 0.82 -2.13 – 1.09 0.525
bike_score 0.06 0.02 0.03 – 0.10 <0.001
Observations 212
R2 / R2 adjusted 0.058 / 0.053


We originally looked at greenness at the census tract level to be comparable with the socio-demographic data. However, we also wanted to investigate differences in greenness at the 100m x 100m grid cell level to give us a closer estimation to the area proximate to the LEED certified buildings. Similarly to the other analyses, we merged the LEED database with the vegetation data, summing the number of LEED buildings in each grid cell. We were unable to pull Walk and BikeScores at this level becuase of limits on the number of WalkScore calls allowed on the free trial. as an average at the grid cell.

The figure below shows the percent vegetation in each grid cell and locations of the LEED certified buildings. The average greenness in grid cells with LEED buildings is 14%, while the average greenness in grid cells without LEED buildings is 50%, and even more stark difference than when comparing between census tracts.

  Dependent = Number of LEED Buildings
Predictors Estimates std. Error CI p
(Intercept) 0.02 0.00 0.02 – 0.02 <0.001
Veg -0.02 0.00 -0.03 – -0.02 <0.001
Observations 36808
R2 0.009

Aim 3. Affordability

Moving forward, to evaluate whether LEED certification affects the affordability of buildings, the extent to which LEED certification of a building in Atlanta affects its price is analyzed.

3.1. Data Preparation

For this assessment, parcel-based analysis was considered as more effective than the census-tract based analysis done for previous analysis. Also, there was a challenge in data preparation, since the Fulton County Tax Assessor’s Data stores tax parcel data was stored by polygons, while the LEED project database features the location of LEED-certified projects was stored by points. Since there is no common identifier to join the two datasets directly, all polygons of tax parcel data are converted to centroid points for a spatial join. Also, due to inefficiencies in data preparation in R, the polygon-to-centroid conversion task was done using ArcGIS Pro. Additionally, it is important to note that this analysis was conducted only for parcels within the Fulton County side of the City of Atlanta. This limitation arose because the DeKalb County government does not provide tax parcel data on an annual basis.

For the spatial join, a 20-meter buffer around the LEED-certified buildings in Atlanta was created to identify the nearest tax parcel centroid to each building as the coordinate point for the LEED-certified buildings. This approach was used because the centroid points created may not match the coordinate points of the LEED-certified buildings from the USGBC’s LEED project dataset. As a result, USGBC’s LEED project data was spatially joined to Fulton County’s 10-year (2015-2024) Tax Parcel data, and a binary variable, ‘LEEDcert,’ was created to indicate whether each building in Atlanta is LEED-certified in each tax year.

Next, per square foot assessed value of buildings on yearly basis from 2015 to 2024 was calculated from the tax parcel data to evaluate how property values of buildings have changed over time. For the final dataset, and for accurate comparison, buildings with tax purpose codes relevant to LEED-certified office/commercial buildings (“B3”, “B4”, “C1”, “C3”, “C4”, “C5”) were filtered for further analysis.

3.2. Box Plot Analysis

Next, the box plots of the per square foot assessed value of buildings by LEED certification status group are examined to compare the per square foot assessed value of commercial/office buildings and their relationship with LEED certification. As shown in the box plots below, LEED-certified buildings have a higher median per square foot assessed value and are relatively homogeneous in their per square foot assessed value. In contrast, non-LEED-certified buildings have a lower median per square foot assessed value and are more heterogeneous in their per square foot assessed value. This suggests that LEED-certified commercial/office buildings tend to have similar characteristics (office buildings located in similar area), while non-LEED-certified commercial/office buildings exhibit more variation in use and characteristics.

3.3. Panel Data Analysis

Finally, panel data analysis was used to assess the extent to which LEED certification affects the per square foot assessed value of buildings. Panel data was chosen as the most appropriate dataset for the analysis because it captures the different dynamics of LEED certification in Atlanta from 2015 to 2024. During this period, some buildings are not LEED certified at all, some are LEED certified for the entire period, and others receive LEED certification mid-period. This variation reflects the changing dynamics of LEED certification for buildings in Atlanta over time, and it is important because these differences in the timing and type of LEED innovation could have varying impacts on the price setting of buildings

A fixed-effects analysis on panel data was conducted to assess the impact of LEED certification on the per square foot assessed value. This method focuses on how changes within each building over time affect the dependent variable while ignoring any unchanging characteristics of the buildings that might distort the analysis. In this case, the changes would be LEED certification, and unchanging characteristics would be location and building years. Therefore, the fixed-effects analysis would assess how the per square foot price of a building changes over time in relation to whether or not the building is LEED certified, while controlling unobserved factors such as location or history of buildings. The results shows that there is no statistically significant relationship between LEED certification and per square foot assessed price of buildings, indicating that having a building LEED-certified does not affect the per square foot assessed value of the parcel.

Dependent Variable: Per Sqft Assessed Value
  Dependent variable
Predictors Estimates std. Error CI p
LEED Certification 113.27 56.01 3.48 – 223.06 0.043
Observations 1469669
R2 / R2 adjusted 0.000 / -0.116

Conclusion, Limitations, and Recommendations

Conclusion

To conclude the analysis, to the research question ‘What is the relationship between LEED buildings and sustainability in Atlanta?’, major findings are as follows:

  • For the question “Are LEED buildings distributed equitably across the city?”, we concluded that the relationship is unclear.While LEED buildings are clustered within certain census tracts, their relationships with socio-demographic characteristics were found to be statistically insignificant.This was a difficult question to answer given that most of the buildings in the LEED database were office buildings, not aligning well with residential data.
  • For the question “Is there evidence of green signaling?”, we concluded that there is strong evidence of green signaling, primarily due to the clustering of buildings in Downtown, Midtown, and Buckhead.
  • For the question “Are areas of the city with LEED buildings greener?”, the analysis yielded mixed results.While the tree canopy is less dense in areas with LEED buildings, there are higher walk and bike scores in areas with LEED buildings compared to areas without them. Accordingly, the relationship between overall greenness and LEED certification was found to be statistically insignificant.
  • For the question “Does LEED certification affect the price affordability of buildings?”, we concluded that there is no statistically significant relationship between LEED certification and the per square foot assessed value of buildingsfrom the fixed-effects analysis of panel data.

Limitation

Throughout the project, several limitations were encountered:

  • A majority of the buildings lack information on project type or scale (e.g., floor versus building). Some included projects were only for a floor of a building which does not really allow for useful analysis. The lack of data on project type prevents meaningful analysis based on project type. If more buildings were residential than we were aware of, the socio-demographic analysis would be more meaningful.
  • The analysis is based on point-in-time data, which may not fully capture changes over time. Because some of the data collected was challenging to collect over time, and LEED buildings can be certified before the building is even constructed, we conducted a point-in-time analysis making it challenging to understand whether LEED certification led to changes in the area near the buildings.
  • The LEED certification dates may not be entirely accurate.

Recommendation

The analysis suggests the presence of green signaling, as LEED-certified buildings are clustered in two areas of the city, likely indicating office spaces. The clustering is also likely due to initiatives from the city for LEED certification. The observed associations with walkability and bikeability are likely due to the higher-density locations of these buildings, rather than stemming from specific sustainability initiatives. Therefore, the City of Atlanta’s focus on LEED certification may not fully address all pillars of sustainability, since LEED certification was found to have no statistically significant relationship with socio-demographic and price factors, and a negative relationship with tree canopy.

Therefore, it is recommended that the initiatives encouraging LEED certification be accompanied by other sustainability measures in the neighborhood, such as increased greenness and car-free transport access. Also, further research would benefit from more comprehensive data on LEED projects, particularly regarding project type.

References